PULMONARY ASPERGILLOSIS IN ASSOCIATION WITH TUBERCULOSIS AND HIV IN
UGANDA
A thesis submitted to The University of Manchester for the degree of Doctor of
Philosophy (PhD)
in the Faculty of Medical and Human Sciences.
2015
IAIN DUNSMUIR PAGE
SCHOOL OF MEDICINE / Institute of Inflammation and Repair
1
Table of Contents
ABBREVIATIONS ............................................................................................................................. 2 1 ABSTRACT ........................................................................................................................................ 6 2
AUTHOR DECLARATION .............................................................................................................. 7 3 COPYRIGHT STATEMENT ............................................................................................................ 8 4
DEDICATION .................................................................................................................................... 9 5
THE AUTHOR ................................................................................................................................. 11 6 INTRODUCTION ............................................................................................................................ 14 7 Part 1 – Thesis structure ...................................................................................................................... 14 8 Part 2 – Author contribution to enclosed papers ........................................................................ 16 9 Part 3 – Publication plan ...................................................................................................................... 18 10 Part 4 – Published review article -‐ Antibody testing in aspergillosis – quo vadis? .......... 20 11 Part 5 – CPA as Global Public Health issue ..................................................................................... 57 12
METHODOLOGY ............................................................................................................................ 62 13 Paper 1 – Performance of six Aspergillus-‐specific IgG assays for the diagnosis of chronic 14 pulmonary aspergillosis (CPA) and allergic bronchopulmonary aspergillosis (ABPA) 62 15 Paper 2 –Aspergillus-‐specific IgG levels in patients previously treated for pulmonary 16 tuberculosis in Gulu, Uganda ............................................................................................................. 75 17 Paper 3 -‐ Prevalence of chronic pulmonary aspergillosis (CPA) secondary to 18 tuberculosis: a cross-‐sectional survey in an area of high tuberculosis prevalence ........ 83 19 Paper 4 -‐ “Frequency of pulmonary aspergillosis in ‘smear-‐negative tuberculosis cases” 20 and Paper 5 “Frequency of Aspergillus co-‐infection in patients admitted to a Ugandan 21 hospital with pulmonary tuberculosis” .......................................................................................... 96 22
PAPER 1 -‐ Comparison of six Aspergillus-‐specific IgG assays for the diagnosis of 23 chronic pulmonary aspergillosis (CPA) and allergic bronchopulmonary 24 aspergillosis (ABPA) ................................................................................................................ 103 25
PAPER 2 -‐ Aspergillus-‐specific IgG levels in patients previously treated for 26 pulmonary tuberculosis in Gulu, Uganda ......................................................................... 139 27
PAPER 3 -‐ Prevalence of chronic pulmonary aspergillosis (CPA) secondary to 28 tuberculosis: a cross-‐sectional survey in an area of high tuberculosis prevalence.29 ......................................................................................................................................................... 160 30
PAPER 4 -‐ An estimate of the prevalence of pulmonary aspergillosis in HIV-‐31 positive Ugandan in patients diagnosed as smear-‐negative pulmonary 32 tuberculosis. ............................................................................................................................... 189 33
PAPER 5 -‐ Aspergillus co-‐infection may be common in Africans with active 34 pulmonary tuberculosis ......................................................................................................... 201 35
SUMMARY .................................................................................................................................... 212 36 REFERENCES ............................................................................................................................... 217 37
APPENDICES ................................................................................................................................ 245 38
2
ABBREVIATIONS 39
40
AAFB Acid and alcohol fast bacilli 41
42
ABPA Allergic bronchopulmonary aspergillosis 43
44
AIDS Acquired immunodeficiency syndrome 45
46
AU Arbitrary units 47
48
AUC Area under the curve 49
50
BAL Broncho-‐alveolar lavage 51
52
CCPA Chronic cavitary pulmonary aspergillosis 53
54
CD4 Cluster of differentiation 4 55
56
CF Cystic fibrosis 57
58
CFPA Chronic fibrosing pulmonary aspergillosis 59
60
CGD Chronic granulomatous disease 61
62
CI Confidence interval 63
64
CIE Counterimmunoelectrophoresis 65
66
COPD Chronic obstructive pulmonary disease 67
68
CNPA Chronic necrotizing pulmonary aspergillosis 69
70
CPA Chronic pulmonary aspergillosis 71
3
72
CT Computed tomography 73
74
CV Co-‐efficient of variation 75
76
CXR Chest X-‐ray 77
78
DD Double diffusion 79
80
DR Congo Democratic Republic of Congo 81
82
ELISA Enzyme-‐linked immunosorbent assay 83
84
EORTC European Organization for Research and Treatment of Cancer 85
86
ESCMID European Society of Clinical Microbiology and Infectious Diseases 87
88
FEIA Fluoroenzymeimmunoassay 89
90
GAFFI Global Action Fund for Fungal Infections 91
92
GM Galactomannan 93
94
GRRH Gulu Regional Referral Hospital 95
96
GVHD Graft versus host disease 97
98
HA Haemagglutination 99
100
HIV Human immunodeficiency virus 101
102
IA Invasive aspergillosis 103
104
4
IAV Intra-‐assay variability 105
106
IHA Immunohaemagglutination 107
108
ICU Intensive care unit 109
110
IDSA Infectious Diseases Society of America 111
112
IMMY Immuno-‐Mycologics 113
114
ISHAM International Society for Human and Animal Mycology 115
116
IRB Institutional Review Board 117
118
JCRC Joint Clinical and Research Centre 119
120
KEMRI Kenya Medical Research Institute 121
122
LA Latex agglutination 123
124
LFD Lateral flow device 125
126
ManRAB Manchester Respiratory and Allergy Biobank 127
128
MIND-‐IHOP Mulago Inpatient Noninvasive Diagnosis – International HIV 129
Opportunistic Pneumonia 130
131
MRI Manchester Royal Infirmary 132
133
MTA Material Transfer Agreement 134
135
MTB Mycobacterium tuberculosis 136
137
5
MRC Medical Research Council 138
139
NAC National Aspergillosis Centre 140
141
PCR Polymerase chain reaction 142
143
RAST Radioimmunoassay 144
145
ROC Receiver operating characteristic 146
147
SAFS Severe asthma with fungal sensitization 148
149
SCID Severe combined immunodeficiency 150
151
TB Tuberculosis 152
153
Th2 cells T-‐helper 2 cells 154
155
TREGS T-‐regulatory cells 156
157
UHSM University Hospital of South Manchester 158
159
UK United Kingdom 160
161
UNCST Uganda National Council for Science and Technology 162
163
USA United States of America 164
165
WHO World Health Organization 166
167
6
ABSTRACT 168 Thesis submitted to The University of Manchester in 2015 by Iain Dunsmuir Page for 169 the degree of Doctor of Philosophy entitled “Pulmonary aspergillosis in association with 170 tuberculosis and HIV in Uganda”. 171 172 Chronic pulmonary aspergillosis (CPA) is a serious disease that occurs secondary to 173 tuberculosis and is estimated to affect 1.2 million persons globally. Pulmonary 174 aspergillosis is found in 2-‐3% of all AIDS autopsies, but 90% of cases go undiagnosed 175 ante-‐mortem. Here the sensitivity and specificity of optimal diagnostic thresholds for 176 CPA have been defined in relation to six Aspergillus-‐specific IgG assays. The prevalence 177 of CPA in an area of high tuberculosis prevalence has been measured. 178 179 Receiver operating characteristic (ROC) curves were used to compare results of testing 180 with six Aspergillus-‐specific IgG assays in 241 patients with CPA and 100 healthy 181 controls. ThermoFisher Scientific ImmunoCAP and Siemens Immulite had ROC area 182 under curve (AUC) results of 0.995 and 0.991 respectively. Both were statistically 183 significantly superior to all other assays. Both had a sensitivity of 96% and specificity of 184 98% using diagnostic cut offs of 20 mg/L and 10 mg/L respectively. 185 186 Eighty patients with allergic bronchopulmonary aspergillosis (ABPA) were also 187 assessed. ROC AUC results were 0.959 for ImmunoCAP and 0.932 for Immulite. The new 188 thresholds produced specificities of 98% for both assays and sensitivities of 78% and 189 81% respectively. Levels in ABPA patients were also compared to asthmatic controls. 190 191 398 patients with treated tuberculosis in Gulu, Uganda were assessed in a cross-‐192 sectional survey. CCPA diagnostic criteria were; 1 – Cough or haemoptysis for one 193 month, 2 – Progressive cavitation on serial chest X-‐ray or either paracavitary fibrosis or 194 aspergilloma on CT scan and 3 – Raised Siemens Immulite Aspergillus-‐specific IgG. All 195 three were required for diagnosis. CCPA was present in 5.7% of patients and simple 196 aspergilloma in 0.7% of patients. There was a non-‐significant trend to less frequent 197 CCPA in HIV positive patients (p=0.18). 198 199 Aspergillus-‐specific IgG levels were measured in stored sera from two adult in patient 200 groups at Mulago Hospital, Kampala, Uganda. 26% of 39 patients with HIV infection and 201 subacute respiratory illness and no diagnosis after extensive investigation had raised 202 levels. 47% of 57 patients with proven active pulmonary tuberculosis had raised levels. 203 204 The Immulite and ImmunoCAP assays both have good sensitivity and specificity for the 205 diagnosis of CPA. New diagnostic thresholds improve the performance of all assays. 206 CCPA has been shown to complicate pulmonary tuberculosis in Gulu, Uganda. Subacute 207 invasive pulmonary aspergillosis is likely to affect many patients with AIDS and 208 subacute respiratory illness. CPA may begin during active pulmonary tuberculosis 209 infection. CPA associated with tuberculosis constitutes a significant unrecognized public 210 health problem, which is probably being incorrectly identified as ‘smear-‐negative 211 tuberculosis’ clinically and in public health data. Prospective studies are now needed to 212 confirm the prevalence of CPA secondary to tuberculosis and define the optimal 213 strategy for routine CPA screening, followed by studies to define optimal treatment 214 regimes for use in research poor-‐settings, where most cases of CPA are likely to occur. 215
7
AUTHOR DECLARATION 216
217
The author has not submitted any portion of the work referred to in the thesis in 218
support of an application for another degree or qualification of this or any other 219
university or other institute of learning. 220
221
Results of Aspergillus-‐specific IgG from 100 healthy controls are compared to other 222
groups throughout this thesis. These results from healthy controls were also used by 223
Mr. Richard Kwizera as part of his 2014 MSc (Medical Mycology) thesis at The 224
University of Manchester. In this work he compares results in healthy controls to 225
patients with chronic obstructive pulmonary disease (COPD). While the author 226
provided some assistance to Mr. Kwizera on this project, his role was peripheral and as 227
a result it does not form part of this PhD thesis. 228
229
230
8
COPYRIGHT STATEMENT 231
232
The author of this thesis (including any appendices and/or schedules to this thesis) 233
owns certain copyright or related rights in it (the “Copyright) and he has given The 234
University of Manchester certain rights to use such Copyright, including for 235
administrative purposes. 236
237
Copies of this thesis, either full or in extracts and whether in hard or electronic copy, 238
may be made only in accordance with the Copyright, Designs and Patents Act 1998 (as 239
amended) and regulations issued under it or, where appropriate, in accordance with 240
licensing agreements which the University has from time to time. This page must form 241
part of any such copies made. 242
243
The ownership of certain Copyright, patents, designs, trade marks and other intellectual 244
property (the “Intellectual Property”) and any reproductions of copyright works in this 245
thesis, for example graphs and tables (“Reproductions”), which may be described in this 246
thesis, may not be owned by the author and may be owned by third parties. Such 247
Intellectual Property and Reproductions cannot and must not be made available for use 248
without the prior written permission of the owner(s) of the relevant Intellectual 249
Property and/or Reproductions. 250
251
Further information on the conditions under which disclosure, publication and 252
commercialization of this thesis, the Copyright and any Intellectual Property and/or 253
Reproductions described in it may take place is available in the University IP Policy (see 254
http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=487), in any relevant Thesis 255
restriction declarations deposited in the University Library, The University Library’s 256
regulations (see 257
http://www.manchester.ac.uk/library/aboutus/regulations) and in the University’s 258
Policy on presentation of Theses. 259
260
261
262
9
DEDICATION 263
264
The author would like to thank his supervisors Profs David Denning, Malcolm 265
Richardson and Angela Simpson for all their guidance and support throughout this 266
project. Many thanks are due to Dr Julie Morris for her assistance with statistical 267
planning and analysis throughout this work. I am also grateful to the staff of the 268
Mycology Reference Centre at University Hospital of South Manchester and the 269
pathology laboratory of Christie Hospital, Manchester for providing training and 270
support throughout the laboratory aspects of this work. 271
272
Further thanks are due to the UHSM Academy charity and to the commercial companies 273
Astellas Pharma, Siemens Immulite, Serion, Genesis, Dynamiker and OLM Medical, all of 274
whom provided support to this work in the form of grants, donations of test kits or 275
provision of accommodation and insurance. Without this support this study would not 276
have been possible. 277
278
This study has required many collaborative efforts. Particular thanks are due to John 279
Opwonya and the staff at the Gulu District Health Office, who played a critical role in 280
patient recruitment for the main CPA prevalence study in Gulu. The reporting of chest 281
X-‐rays was crucial to this study and required a substantial commitment by Dr Cyprian 282
Opira (Senior Radiologist and Clinical Director, St. Mary’s Hospital, Lacor, Uganda), Dr 283
Sharath Hosmane (Specialty Radiology Registrar, University Hospital of South 284
Manchester) and Dr Richard Sawyer (Senior Consultant Radiologist, University Hospital 285
of South Manchester). 286
287
I am also indebted to Mr. Nathan Onyachi (Clinical Director, Gulu Hospital) for his 288
assistance in planning the Gulu survey and to Drs. William Worodria, Alfred Andama, 289
Irene Akaka and the MIND-‐IHOP study group at Mulago Hospital, Kampala. This 290
substantial study was undertaken in collaboration with the University of California, San 291
Francisco aiming to identify the range of conditions present in patients admitted with 292
chronic cough. They were kind enough to provide me with stored sera from selected 293
patients in this study, the analysis of which forms part of this thesis. 294
10
295
Above all I must dedicate this thesis to my wife Sarah, whom I married six weeks into 296
this PhD study period. She has been a source of endless support and tolerance, 297
especially when I moved to Uganda to commence clinical work just a few months into 298
my recovery from a very severe illness. 299
300
11
THE AUTHOR 301
302
303 304
Dr Iain Dunsmuir Page MBChB, BSc, MRCP, DTM&H 305
306
The author graduated from The University of Edinburgh Medical School in 2002. He was 307
awarded an intercalated BSc in Virology with 2.1 Honours. He gained membership of 308
the Royal College of Physicians of Edinburgh in 2006 and was awarded the Diploma of 309
Tropical Medicine and Hygiene with Distinction by the University of Liverpool in 2007. 310
311
He worked as a junior doctor in Edinburgh, Glasgow and Leeds between before taking 312
the post of Clinical Lecturer at the University of Malawi from 2007 to 2008. He then 313
returned to the UK to become a Specialty Registrar in Infectious Diseases and General 314
Medicine, working at Blackpool Victoria Hospital and North Manchester General 315
Hospital. From 2012 he has been working as a Clinical Research Fellow at the University 316
of Manchester, based at the National Aspergillosis Centre at University Hospital of South 317
Manchester, with fieldwork in Gulu, Uganda. 318
319
The authors BSc included a research thesis entitled ‘Expression of Human Herpes Virus 320
8 protein vOx-‐2’, which was awarded 2.1 Honours. He has published four research 321
articles, listed below, prior to his current post. He also authored the gastro-‐intestinal 322
medicine section of the 2008 Malawian National Prescribing guidelines. 323
324
In addition to the contents of this thesis, the Dr Page also authored two conference 325
abstracts relating to pulmonary aspergillosis. He was an invited speaker at the 6th 326
Advances Against Aspergillosis conference (Madrid, February 2014) and the Global 327
Action Fund for Fungal Infections Forum (GAFFI, Seattle, February 2015). He was an 328
12
invited panel member at the International Society for Human and Animal Mycology 329
(ISHAM) expert group on azole resistance (Copenhagen, October 2013), which will be 330
published in due course. He co-‐authored the UK National Aspergillosis Centre’s 331
submission to the UK All-‐Party Parliamentary Group on Global Tuberculosis in 2014. 332
333
Significant difficulties were encountered in the course of this work. Fieldwork was 334
delayed by several months due to outbreaks of Ebola fever in Uganda in 2012 and 2014. 335
The author contracted leptospirosis while on honeymoon in the second month of the 336
study period and was admitted to intensive care with multi-‐organ failure. While he was 337
able to return to work and was granted a 3-‐month extension to his study period, he 338
suffered persistent hepatitis and severe fatigue for a year after discharge from hospital. 339
He has now fully recovered. 340
341
Prior Publications 342
343
1 -‐ Page I, McKew S, Kudzala A, Fullwood C, van Oosterhout J and Bates I. Screening HIV 344
infected adults in Malawi for anaemia: need for a new hemoglobin threshold to 345
determine eligibility for antiretroviral therapy. International Journal of STD & AIDS. 346
2013; 24:449-‐53 347
348
2 -‐ Page I, Phillips M, Flegg P, Palmer R. The impact of new National HIV Testing 349
Guidelines at a District General Hospital in an area of high HIV sero-‐prevalence. Journal 350
of the Royal College of Physicians of Edinburgh. 2011; 41:9-‐12. 351
352
3 -‐ Page I, Hardy GD, Fairfield J, Orr D, Nichani R. Implementing the Surviving Sepsis 353
Guidelines in a District General Hospital. Journal of the Royal College of Physicians of 354
Edinburgh 2011;41:309-‐15. 355
356
4 -‐ Ameyaw E, Nguah SB, Ansong D, Page I, Guillerm M, Bates I. The outcome of a test-‐357
treat package versus routine outpatient care for Ghanaian children with fever: a 358
pragmatic randomized control trial. Malaria Journal 2014; 13:461. 359
360
361
13
Conference presentations – unpublished work not included in thesis 362
363
1 – Muldoon EG, Page I, Bishop P, Denning DW. Aspergillus pulmonary nodules; 364
presentation, radiology and histology features. Poster 100 – 6th Advances Against 365
Aspergillosis, Madrid, Spain 27th February – 2nd March 2014. 366
367
2 – Richardson MD, Page I, Richardson RR. Detection of Aspergillus antibodies by a new 368
indirect haemagglutination assay. Poster 97 – 6th Advances Against Aspergillosis, 369
Madrid, Spain 27th February – 2nd March 2014. 370
371
372
14
INTRODUCTION 373
374
Part 1 – Thesis structure 375
376
This thesis is submitted in the alternative format. This format was selected as the 377
completed research takes the form of five separate studies. Each of these has been 378
completed with positive results that will be submitted for publication in due course. The 379
alternative format is therefore suitable. The introduction is based on a review article 380
published in the journal Medical Mycology, although this article has been edited from its 381
published format in response to corrections produced at the PhD viva examination. 382
There is then a further review of the literature relating to the global prevalence of 383
chronic pulmonary aspergillosis (CPA) in association with tuberculosis and human 384
immunodeficiency virus (HIV) infection. 385
386
The primary goal of this thesis is to measure the prevalence of CPA secondary to 387
tuberculosis. Additional studies provide evidence of the existence of pulmonary 388
aspergillosis in adult patients admitted to an African hospital with acquired 389
immunodeficiency syndrome (AIDS) and chronic cough. Before these tasks could be 390
undertaken, it was first necessary to demonstrate the sensitivity and specificity of 391
existing tests for the diagnosis of CPA. The key test in the diagnosis of CPA is 392
measurement of Aspergillus-‐specific IgG. While many assays are available to measure 393
this, the sensitivity and specificity of these assays for the diagnosis of CPA was barely 394
described prior to this work. 395
396
The first paper in this thesis describes a comparison of six assays for the measurement 397
of Aspergillus-‐specific IgG in cohorts of patients with known CPA and ABPA at the UK 398
National Aspergillosis Centre plus healthy controls. This paper demonstrates that two of 399
the available assays perform well in the diagnosis of CPA and ABPA. The Siemens 400
Immulite assay was then selected for use in the prevalence studies as the manufacturer 401
kindly offered to donate kits for this purpose. 402
403
15
The third and fourth papers in the thesis describe the results of a cross-‐sectional study 404
to measure the prevalence of CCPA in patients with treated pulmonary tuberculosis in 405
Gulu, Uganda. This required two surveys two years apart in order to detect progressive 406
cavitation on chest X-‐ray. Clinical assessment was performed and Aspergillus-‐specific 407
IgG measured. CT scan was performed in patients with evidence of possible CPA from 408
the first survey. The first study commenced in September 2012 and is described in 409
paper two. The second survey was completed in February 2015 and is described in 410
paper three. 411
412
In addition to these cross-‐sectional studies in Gulu, two collaborative studies were 413
arranged with the MIND-‐IHOP study group at Mulago Hospital, Kampala. This forms the 414
basis of papers four and five. This study group had been collecting sera from adult 415
patients admitted to their hospital with chronic cough. They had performed extensive 416
investigations to diagnose the underlying conditions in these patients, but did not have 417
access to Aspergillus serology. We therefore measured Aspergillus-‐specific IgG in 418
selected stored sera from the MIND-‐IHOP study. The first group was from patients with 419
a clinical presentation that was consistent with subacute invasive pulmonary 420
aspergillosis. This forms the basis of paper four. The second group was from patients 421
with proven pulmonary tuberculosis. Here a positive test for Aspergillus-‐specific IgG 422
provides evidence of Aspergillus co-‐infection. This forms the basis for paper five. 423
424
Overall this thesis provides definitive data on the sensitivity and specificity of various 425
Aspergillus-‐specific IgG assays. It demonstrates that two assays have statistically 426
significantly superior performance for the diagnosis of CPA. It then uses one of these 427
assays to measure the prevalence of CCPA in an area of high tuberculosis prevalence for 428
the first time. Evidence is also provided of undiagnosed pulmonary aspergillosis in 429
patients with AIDS and chronic cough and of Aspergillus co-‐infection in patients with 430
proven pulmonary tuberculosis. 431
432
It demonstrates CPA is a sufficiently common problem to be considered a public health 433
issue in Uganda. For the first time it defines the radiological and serological 434
characteristics of CPA in HIV-‐infected people, after they have recovered from 435
16
tuberculosis. The work described in this thesis provides the first substantial evidence 436
that CPA is a significant global public health issue. 437
Part 2 – Author contribution to enclosed papers 438
439
The author of this thesis will be the first author of all papers included in this alternative 440
format thesis. In each case he took the lead role in all aspects of the studies including 441
planning, data gathering, analysis and presentation as well as ensuring compliance with 442
all regulatory requirements. All clinical aspects of this work took place in Uganda, 443
where the author worked independently for 18 months of the study period. 444
445
The concepts for papers one and two were devised by the author’s lead supervisor (Prof 446
David Denning) and discussed prior to commencement of the PhD. The concepts for 447
papers three to five originated with the author and were developed after commencing 448
the PhD. 449
450
The author was awarded funding for the Clinical Research Fellow post following an 451
open application process. This post did not include research funding, but £26,000 452
funding was made available by UHSM Academy to complete the first CPA prevalence 453
study in Uganda. The author then applied for and was awarded a further £24,000 by 454
pharmaceutical firm Astellas Pharma in an open applications process. 455
456
The first paper was completed in collaboration with several commercial test 457
manufacturers. The author contacted all known manufacturers of Aspergillus-‐specific 458
IgG assays regarding the serology comparison studies. Test manufacturers Siemens 459
Immulite (Germany), Serion (Germany), Genesis (UK), OLM Medical (UK) and 460
Dynamiker (China) all agreed to donate kits for use in the comparison study. Dynamiker 461
and Serion both also agreed to provide £2500 towards laboratory costs. The total value 462
of funding and donations arranged by the author is around £100,000. 463
464
Once kits were donated the author performed around 95% of assays involving these 465
kits, with the remainder performed by a laboratory assistant (Matthew Kneale) who 466
was trained in their use. 467
17
468
The Ugandan prevalence studies were led by the author, who personally assessed 469
around 95% of patients, with the remainder assessed by study assistant Matthew 470
Kneale following training by the author. The author was assisted by two Ugandan 471
employees who performed translation and venepuncture. The Ugandan prevalence 472
studies required the collaboration of local health workers. The author visited Gulu for 473
one week prior to the PhD with his supervisor Prof. David Denning and was introduced 474
to most of the local collaborators. He then went on to develop these relationships over 475
the next three years. 476
477
Gulu District Health team played a key role in identifying eligible patients by 478
communicating with village health workers. Laboratory staff at the Joint Clinical and 479
Research Centre (JCRC), Gulu laboratory performed serum separation and storage 480
during the first Gulu survey as well as CD4 counts on HIV infected patients. During the 481
second survey this role was performed by staff at the Gulu Regional Referral Hospital 482
laboratory, who also performed GeneXpert tuberculosis PCR testing on sputum samples 483
from patients with productive cough. Control samples from healthy blood donors were 484
acquired by the Gulu Blood Transfusion service. 485
486
Chest X-‐rays were performed by staff at the radiology department of St. Mary’s Hospital, 487
Lacor. CT scans were performed by staff at the Kampala Imaging Centre. While the 488
author reported all chest X-‐rays and scans, results in paper two are based on reports by 489
three blinded radiologists. This process is underway for chest X-‐rays and CT scans for 490
paper three, but is not complete at the time of thesis submission. Radiological results in 491
paper three are therefore based on the author’s own reports. 492
493
Papers four and five describe the results of opportunistic testing of stored samples from 494
another study. The original MIND-‐IHOP study was a collaborative venture between 495
Mulago Hospital, Kampala and the University of California, San Francisco. The author 496
contacted the Kampala team after reading published results of their study. He then set 497
up two new collaborative studies after it became clear that the initial MIND-‐IHOP study 498
did not include adequate testing for pulmonary aspergillosis. The MIND-‐IHOP study 499
took place in 2010-‐11. The author played no role in the design or conduct of this study. 500
18
After stored samples were identified the author arranged shipment to the UK and 501
performed Aspergillus-‐specific IgG testing in the Manchester laboratory. He then 502
analysed these results in relation to clinical data provided by the MIND-‐IHOP group and 503
took a lead role in the subsequent presentation of these results. 504
Part 3 – Publication plan 505
506
The review article based on the PhD literature review has now been published in the 507
journal Medical Mycology. Final data for each of the original research papers was only 508
received from collaborators between January and March 2015. As a result none of these 509
have been published yet. Paper one has been submitted for publication in a peer 510
reviewed journal. All other studies refer to the diagnostic cut offs and sensitivity and 511
specificity results from that paper. The other papers will be submitted for publication 512
once paper one is in print. Interim data has, however been presented at academic 513
conferences as listed below:-‐ 514
515
Review article publication 516
517
Page ID, Richardson M, Denning D. Antibody testing in aspergillosis – quo vadis? 518
Medical Mycology. 2015:53;417-‐39. Doi 10.1093/mmy/myv020. 519
520
Interim results presented at conferences 521
522
Page ID, Kwizera R, Richardson M, Denning D. Comparative efficacy of five Aspergillus-‐523
specific IgG ELISAs for the diagnosis of Chronic Pulmonary Aspergillosis (CPA) and 524
Allergic Bronchopulmonary Aspergillosis (ABPA). 25th European Conference on Clinical 525
Microbiology and Infectious Diseases, Copenhagen, 25th-‐28th April 2015. 526
527
Page ID, Onyachi N, Opwonya J, Opira C, Odongo-‐Aginya E, Mockridge A, Byrne G, 528
Richardson M, Denning DW. Chronic Pulmonary Aspergillosis (CPA) is likely to be a 529
common complication of pulmonary tuberculosis: initial results of a cross-‐sectional 530
survey. 25th European Conference on Clinical Microbiology and Infectious Diseases, 531
Copenhagen, 25th-‐28th April 2015. 532
19
Page ID, Opwonya J, Onyachi N, Opira C, Ondongo-‐Aginya E, Mockridge A, Byrne G, 533
Denning D. Chronic pulmonary aspergillosis complicating treated pulmonary 534
tuberculosis in Gulu, Uganda. British Society of Medical Mycology, Manchester 28-‐29th 535
May 2014. 536
537
Page ID, Opwonya J, Onyachi N, Opira C, Ondongo-‐Aginya E, Mockridge A, Byrne G, 538
Denning D. Chronic pulmonary aspergillosis complicating treated pulmonary 539
tuberculosis in Gulu, Uganda. 44th Union International Conference on Tuberculosis and 540
Lung Health, Paris, France 1-‐3rd November 2013. 541
542
Page ID, Opwonya J, Onyachi N, Opira C, Ondongo-‐Aginya E, Mockridge A, Byrne G, 543
Denning D. Chronic pulmonary aspergillosis complicating treated pulmonary 544
tuberculosis in Gulu, Uganda. 6th Trends in Medical Mycology conference, Copenhagen, 545
Denmark 11-‐14th October 2013. 546
547
Page ID, Denning D. Pulmonary aspergillosis secondary to AIDS or tuberculosis in 548
Uganda. Royal Society of Tropical Medicine Research in Progress Conference, London, UK 549
Dec 2012. 550
551
552
20
Part 4 – Published review article -‐ Antibody testing in aspergillosis – quo vadis? 553
554
Iain D Pagea,b,c#, Malcolm Richardsona,b,c, David W Denninga,b,c 555
556
Institute of Inflammation and Repair, The University of Manchester, UKa, Manchester 557
Academic Health Science Centre, UKb, National Aspergillosis Center and Mycology 558
Reference Centre, University Hospital of South Manchester, UKc. 559
560
#Address correspondence to Iain D Page, [email protected] 561
562
Medical Mycology. 2015:53;417-‐39. Doi 10.1093/mmy/myv020. 563
564
565
566
567
568
569
570
571
572
573
574
21
ABSTRACT 575
576
Humans are constantly exposed to airborne Aspergillus spores. Most develop 577
Aspergillus-‐specific antibodies by adulthood. Persons with chronic lung disease or 578
Aspergillus airway colonization often have raised levels of Aspergillus-‐specific IgG. It is 579
not known whether this signifies an increased risk of future aspergillosis. 580
Chronic and allergic forms of pulmonary aspergillosis are estimated to affect 581
over three million people worldwide. Antibody testing is central to diagnosis of these 582
conditions, with raised Aspergillus-‐specific IgG in chronic pulmonary aspergillosis and 583
raised Aspergillus-‐specific IgE in allergic aspergillosis. Antibody levels are also used to 584
monitor treatment response in these syndromes. Acute invasive disease is less common. 585
There is a more limited role for antibody testing in this setting as immunosuppression 586
often results in delayed or absent antibody response. 587
Many methods exist to detect Aspergillus-‐specific antibodies, but there are 588
limited published data regarding comparative performance and reproducibility. We 589
discuss the merits of the available tests in the various clinical settings and their 590
suitability for use in resource-‐poor settings, where the majority of cases of aspergillosis 591
are thought to occur. We summarise the gaps in existing knowledge and opportunities 592
for further study that could allow optimal use of antibody testing in this field. 593
594
22
INTRODUCTION 595
596
Aspergillus is a mould that causes disease in humans1. Infection can lead to a spectrum 597
of clinical syndromes, ranging from rapidly fatal acute invasive infection to chronic 598
debilitating pulmonary disease2. The latter can normally be characterized as either 599
allergic airways disorders closely associated with asthma3,4 or chronic lung infection 600
that can be complicated by progressive fibrosis and massive haemoptysis5–8. 601
Understanding of these conditions has improved significantly over the course of several 602
decades, with associated changes in the case definitions and terminology used to 603
describe disease4,5. 604
It is likely that chronic and allergic forms of pulmonary aspergillosis are 605
sufficiently common to be considered a public health issue on a global scale9–13. The 606
most common form of aspergillosis is undoubtedly chronic pulmonary aspergillosis 607
secondary to treated tuberculosis7,8,14–16. It is therefore likely that most patients with 608
pulmonary aspergillosis will be living in the resource-‐poor settings where tuberculosis 609
is most common. 610
Treatment with antifungal medication is associated with clinical and / or 611
radiological stabilization or improvement in all common forms of aspergillosis17–20. It 612
can be successfully delivered in resource poor settings18. Surgery can cure chronic 613
pulmonary aspergillosis in selected patients with localized disease15,21 and can also be 614
performed in resource poor settings16. 615
Diagnosis of aspergillosis is challenging. Unfortunately the clinical presentation 616
of chronic and allergic aspergillosis overlaps considerably with other, better-‐recognised 617
conditions and it is likely that the vast majority of cases go undiagnosed5,14,22. The 618
development of assays to detect antigenaemia has led to improved ability to diagnose 619
invasive infections promptly and the interpretation and performance of these antigen 620
detection assays have been reviewed extensively23–28. Chronic and allergic forms of 621
aspergillosis are much more common than invasive disease11–13,29, but have been 622
relatively neglected. Antibody testing is central to the diagnosis of these conditions. 623
It is the goal of this article to describe the antibody response that occurs in 624
Aspergillus infection and its role in the diagnosis and management of aspergillosis. The 625
strengths and limitations of the various techniques available to measure Aspergillus-‐626
23
specific antibodies will be described, together with a review of the evidence of their 627
comparative efficacies. 628
629
CLINICAL SYNDROMES DUE TO ASPERGILLUS INFECTION 630
631
It is likely that human exposure to Aspergillus spp. is near universal, as Aspergillus spp. 632
are consistently recovered from air samples in urban and rural areas throughout the 633
year30,31. Human disease due to Aspergillus spp. has also been recorded worldwide10. 634
The vast majority of patients with aspergillosis have one or more underlying disorders 635
and the presentation of aspergillosis varies in line with the underlying disorder2,14,22. 636
While there can be a significant degree of overlap between syndromes it is nonetheless 637
useful to summarise the commonly observed syndromes. The antibody response to 638
Aspergillus and thus the role of antibody measurement in diagnosis and management 639
varies greatly from one syndrome to another. 640
641
Superficial aspergillosis 642
643
Cutaneous aspergillosis is uncommon as the physical barrier provided by the epidermis 644
prevents Aspergillus inoculation. Aspergillus spp. do cause keratitis, otitis externa and 645
onychomycosis in immunocompetent persons, but antibody response is not normally 646
seen in these conditions and diagnosis relies on microscopy and culture32–36. 647
648
Aspergillus bronchitis 649
650
Aspergillus can grow in the human respiratory tract. This can occur in asymptomatic 651
patients and in these circumstances is termed colonisation37,38. However in some 652
patients with no significant immune deficit, Aspergillus growth in the respiratory tract 653
occurs and is associated with cough and recurrent chest infections, but without 654
radiological evidence of pulmonary aspergillosis. These patients are considered to have 655
Aspergillus bronchitis39. This is well described in persons with cystic fibrosis40, but is 656
not restricted to this group39. Evidence of Aspergillus growth is provided by either 657
24
recurrent culture growth from respiratory samples or raised levels of Aspergillus-‐658
specific antibodies. 659
660
Acute invasive aspergillosis 661
662
Acute invasive disease can occur in immunocompromised persons and is termed 663
invasive pulmonary aspergillosis, invasive rhinosinusitis, invasive tracheo-‐bronchial 664
aspergillosis or disseminated aspergillosis depending on the site of the invasive 665
infection41–43. These conditions are mostly associated with severe neutropaenia, but can 666
also be seen in association with a large range of conditions including corticosteroid use, 667
intensive care unit (ICU) admission, diabetes, liver failure, tuberculosis, chronic 668
obstructive pulmonary disease (COPD), chronic granulomatous disease (CGD), graft 669
versus host disease (GVHD), solid organ transplantation and acquired 670
immunodeficiency syndrome (AIDS)42–47. 671
Pneumonia is the most common initial presentation, but lesions involving the 672
kidneys, cardiac valves, brain and skin have been documented42,44,46. Clear diagnostic 673
guidelines have been published by the European Organization for Research and 674
Treatment of Cancer (EORTC)48. Measurement of Aspergillus-‐specific antibodies do not 675
form part of these criteria, with diagnosis resting on biopsy evidence for proven disease 676
or a combination of risk factors, radiological change and microbiological evidence in the 677
form of culture growth or antigen detection for probable disease. 678
679
Subacute invasive pulmonary aspergillosis 680
681
In addition to this well-‐recognised acute presentation of invasive disease, there can also 682
be a more indolent presentation with progressive destruction of the lung over several 683
weeks or months. This has been frequently referred to as chronic necrotising 684
pulmonary aspergillosis or semi-‐invasive aspergillosis in the past6,49, but the term 685
subacute invasive pulmonary aspergillosis has been adopted more recently50 and will 686
be used throughout this article. The condition is normally seen in patients with mild 687
immunosuppression due to diabetes, steroid use, alcoholism, COPD, tuberculosis or 688
25
AIDS6,49,51–53. A similar condition occurs in the sinuses, where is termed chronic invasive 689
fungal rhinosinusitis41. 690
Diagnosis of subacute invasive aspergillosis is based on a combination of 691
symptoms, radiological changes and laboratory tests, including antibody and antigen 692
tests or culture6,53. 693
There is a large degree of overlap between subacute invasive pulmonary 694
aspergillosis and chronic pulmonary aspergillosis7. The duration of symptoms is the 695
main difference, over one month of symptoms considered appropriate for subacute 696
invasive aspergillosis6,53. In the absence of treatment, death from progressive lung 697
destruction and massive haemoptysis is common. Those who survive subacute invasive 698
pulmonary aspergillosis can go on to develop chronic pulmonary aspergillosis6. 699
700
Chronic pulmonary aspergillosis 701
702
The term aspergilloma refers to a fungal ball in a lung cavity. The cavity may be pre-‐703
existing or be created by Aspergillus as an aspergilloma forms. This can be an incidental 704
radiological finding in an asymptomatic person and is termed simple aspergilloma in 705
these cases15. Fungal balls are also well described in the sinuses41. 706
707
Formation of new cavities and fibrosis of surrounding lung tissues often occurs in 708
response to chronic Aspergillus infection. This process has been referred to as complex 709
aspergilloma15,54–56, but is now preferably referred to as chronic pulmonary 710
aspergillosis (CPA)5,7,8. It can be subdivided into chronic cavitary pulmonary 711
aspergillosis (CCPA) and chronic fibrosing pulmonary aspergillosis (CFPA)5. CPA occurs 712
in patients with underlying lung conditions, including treated tuberculosis, atypical 713
mycobacterial infection, sarcoidosis, COPD, pneumothorax, prior lung surgery, 714
rheumatoid arthritis or lung cancer7,8,14. CPA can also complicate subacute invasive 715
pulmonary aspergillosis6 or allergic bronchopulmonary aspergillosis13. Progressive lung 716
destruction due to fibrosis and cavitation occurs, with massive life-‐threatening 717
haemoptysis complicating advanced disease5,7,8. CPA is estimated to affect 3 million 718
people worldwide11–13,57. The five-‐year mortality of CPA is up to 85%7. 719
26
Diagnosis is based on a combination of chronic symptoms, radiological changes 720
and laboratory tests5,7,8. Unfortunately the symptoms of cough and breathlessness can 721
overlap greatly with the underlying lung diseases. Radiological changes of cavitation, 722
fibrosis and pleural thickening can also overlap greatly with underlying conditions, with 723
distinctive aspergilloma detected only in a minority of patients5,8,58. Laboratory testing 724
is therefore crucial in differentiating patients with CPA from those with underlying lung 725
disease alone. Serum galactomannan has been documented in up to 50% of CPA 726
cases8,59,60, but may not have any diagnostic value due to a high false positive rate59. 727
Culture of sputum is positive in up to 44% of CPA cases59,61, but raised levels of 728
Aspergillus-‐specific immunoglobulin G (IgG) antibodies are almost always present and 729
are central to diagnosis5,7,8. 730
731
Allergic aspergillosis 732
733
Sensitisation to Aspergillus can occur in asthmatics and such patients are more likely to 734
have severe asthma with life-‐threatening complications9,62,63. This is referred to as 735
severe asthma with fungal sensitization (SAFS)64. Allergy to Aspergillus can also result in 736
rhinosinusitis41. Diagnosis of sensitization can be achieved by skin testing or by 737
detection of raised levels of Aspergillus-‐specific immunoglobulin E (IgE) antibodies65. 738
Allergic bronchopulmonary aspergillosis (ABPA) complicates 1-‐4% of adult 739
asthma cases, many of whom have otherwise healthy lungs and no 740
immunocompromise13,22. ABPA can also complicate cystic fibrosis66 and occasional 741
cases are also seen in persons with neither condition67. ABPA is characterized by 742
recurrent exacerbations resulting in cough and breathlessness with lung infiltrates on 743
chest X-‐ray and can be complicated by the development of bronchiectasis or CPA22. In 744
contrast to other forms of aspergillosis, steroids are the main treatment, with 745
antifungals used as steroid sparing agents in some cases4. 746
The International Society for Human and Animal Mycology (ISHAM) has recently 747
revised the diagnostic criteria for ABPA4. Diagnosis requires the presence of cystic 748
fibrosis or asthma plus a total serum IgE of > 1000 IU/ml and evidence of Aspergillus 749
sensitivity from either skin prick testing or raised Aspergillus-‐specific IgE antibodies. 750
Two of the three following minor criteria must also be present; radiographic changes 751
27
consistent with ABPA, raised eosinophil count or raised levels of Aspergillus-‐specific 752
precipitating or IgG antibodies. 753
The diagnostic criteria for different clinical syndromes in aspergillosis are 754
summarised in table 1. Figure 1 is a visual representation of the number of patients with 755
each clinical syndrome overlaid with the bars showing the total number of patients 756
where each test is diagnostic. 757
758
ANTIBODY RESPONSE TO ASPERGILLUS 759
760
Asymptomatic persons 761
762
While human airways are constantly exposed to Aspergillus spores present in the 763
air30,31, these spores are rendered immunologically inert by the presence of surface 764
hydrophobin68. In healthy persons the innate immune system ensures that most spores 765
are promptly destroyed69. Those that germinate into hyphae are normally recognised 766
and killed by neutrophils before they can invade host tissue70. 767
Nonetheless, antibodies to Aspergillus are formed in healthy persons, with mean 768
levels increasing from childhood into adulthood71,72. In accordance with this, tests for 769
Aspergillus-‐specific antibodies are normally positive using sensitive methods such as 770
enzyme-‐linked immunosorbent assay (ELISA), with abnormal results defined as a raised 771
level above a cut-‐off related to the range of antibody levels seen in healthy 772
persons38,73,74. 773
Asymptomatic persons with Aspergillus airway colonisation may develop raised 774
levels of Aspergillus-‐specific antibodies and the correct interpretation of this is not clear. 775
Some authors classify raised levels in this population as false positives38. However to 776
our knowledge there are no published studies describing the long-‐term outcome in 777
colonised patients. It is therefore not clear whether patients described in the literature 778
as colonised are at higher risk of developing pulmonary aspergillosis in the future or 779
not. If this were the case then raised levels of Aspergillus-‐specific antibodies in 780
asymptomatic persons might be an indication of pre-‐clinical disease. 781
Aspergillosis normally develops in patients with underlying diseases4,14. The 782
range of levels of Aspergillus-‐specific antibodies in persons with these diseases may not 783
be the same as healthy persons. Up to 20% of patients with treated tuberculosis have 784
28
positive tests for Aspergillus-‐specific antibodies75, this rises to 25% when lung cavities 785
are present76 and 36% in those with haemoptysis77. Raised Aspergillus-‐specific IgG 786
levels are also seen in 13% of Indian asthmatics62, 24% of British cystic fibrosis 787
patients78, 25% of Indian children with thalassemia and human immunodeficiency virus 788
(HIV) infection79 and 8% of all patients attending a Brazilian tertiary respiratory 789
clinic80. 790
These surveys did not include further tests to diagnose pulmonary aspergillosis 791
and some of these patients with raised antibody levels may therefore have undiagnosed 792
aspergillosis. Nonetheless these results suggest that the range of Aspergillus-‐specific 793
antibody levels in patients at risk of developing aspergillosis may be different from the 794
ranges described in healthy individuals. Indeed the mean level of Aspergillus-‐specific 795
IgG in cystic fibrosis patients without ABPA is higher than the manufacturers upper 796
limit of normal78. 797
Further work is needed to define the range of Aspergillus-‐specific antibody levels 798
in other patient groups who are at risk of, but have not developed aspergillosis, as this is 799
the population that is most likely to undergo testing for CPA or ABPA. It may be that 800
existing diagnostic cut-‐offs for Aspergillus-‐specific antibody levels, which were defined 801
using healthy persons as a control, are not appropriate for those at most risk of 802
developing aspergillosis. 803
804
Aspergillus bronchitis 805
806
Seventy one percent of patients with symptomatic Aspergillus bronchitis have raised 807
Aspergillus-‐specific IgG and 29% have positive precipitins39. While raised levels of 808
Aspergillus-‐specific IgE are not typical of Aspergillus bronchitis, there is considerable 809
overlap between the clinical presentation of Aspergillus bronchitis and that of ABPA. 810
Measurement of total and Aspergillus-‐specific IgE would therefore be appropriate in 811
patients with symptoms of Aspergillus bronchitis, with the aim of identifying cases of 812
serological ABPA4. 813
814
815
29
Acute invasive aspergillosis 816 817
Acute invasive aspergillosis normally occurs in patients with profound immune 818
dysfunction, meaning that antibody production may not occur in response to infection81. 819
However Aspergillus-‐specific IgG antibodies are detectable by ELISA in 29-‐100% of 820
patients during the course of acute invasive aspergillosis82–89. Sensitivity is higher in 821
non-‐neutropaenic patients (48%) than neutropaenic patients (6%)90. 822
When antibodies do develop in acute illness, they take a mean of 10.8 days to 823
appear86 and historically a majority of patients with invasive aspergillosis died without 824
producing antibodies83,90. This greatly reduces their utility for diagnosis of acute disease 825
as early treatment is crucial for survival19. Nonetheless when a patient with suspected 826
invasive aspergillosis does develop newly raised Aspergillus-‐specific IgG antibodies this 827
finding does provide evidence of acute infection84. 828
There may be other uses for antibody testing in invasive aspergillosis other than 829
diagnosis of acute disease. A retrospective survey described an increase in all-‐cause 830
mortality in Aspergillus colonised lung transplant patients, with a hazard ratio of 2.291. 831
Another similar study failed to show this association47, but this cohort was complicated 832
by the fact that colonised patients considered high risk for development to invasive 833
aspergillosis were not included. This suggests that patients colonised with Aspergillus 834
might then benefit from antifungal prophylaxis or early empirical antifungal treatment 835
when immunosuppressed. Screening patients for raised Aspergillus-‐specific IgG 836
antibodies prior to initiation of immunosuppressive therapy might be a convenient 837
method of identifying such patients89,92. 838
There can also be a role for serial measurement of Aspergillus-‐specific IgG 839
antibodies after commencing treatment for presumed invasive aspergillosis. In this 840
situation a fall in Aspergillus-‐specific IgG levels is a bad prognostic marker93,94. This 841
most likely relates to failure of the immune system to mount a response to the infection. 842
A rise in Aspergillus-‐specific IgG antibodies can retrospectively confirm the diagnosis in 843
those who recover following empirical treatment for suspected invasive aspergillosis23. 844
This knowledge might affect decisions about whether to forgo further 845
immunosuppressive therapy or to provide antifungal prophylaxis with it. 846
847
30
Subacute invasive aspergillosis 848 849
Raised levels of Aspergillus-‐specific IgG antibodies are more likely to occur and thus are 850
of greater use for diagnosis in this group than in acute disease6,53. In lung transplant 851
recipients, invasive aspergillosis often develops months after transplantation and can 852
evolve slowly. A rise in Aspergillus-‐specific IgG titres preceded radiological changes by 853
1-‐2 weeks and diagnosis of invasive aspergillosis by 2-‐20 weeks in this group95. Raised 854
levels of Aspergillus-‐specific IgG antibodies were detected in 93% of 43 Korean patients6 855
and 77% of 45 Japanese patients with subacute invasive pulmonary aspergillosis53. 856
Sensitivities of serum (1,3)-‐ß-‐D glucan and galactomannan testing in the Japanese 857
patients were 60% and 64% respectively. 858
The sensitivity of galactomannan antigen testing is much lower when 859
Aspergillus-‐specific antibodies are present than when they are absent96. This effect may 860
be due to direct binding of anti-‐Aspergillus antibodies to the galactomannan antigen97. It 861
is therefore possible that both antigen and antibody testing will both needed to achieve 862
acceptable sensitivity for the diagnosis of subacute invasive aspergillosis in mildly 863
immunosuppressed patients. 864
865
Chronic pulmonary aspergillosis 866
867
Raised levels of Aspergillus-‐specific IgG antibodies are almost always found in CPA5,8,98. 868
Production of specific Immunoglobulin M (IgM) is also noted in up to 50% of CPA 869
cases72,88,99–102. This might be considered unusual in a chronic disease, as raised levels of 870
specific IgM are typically associated with the acute phase of an infection. 871
Ongoing growth of Aspergillus produces numerous different antigens at different 872
stages in its growth cycle that interact with the immune system in different ways103. IgM 873
might therefore be repeatedly re-‐stimulated as an immune response develops to each 874
new, individual Aspergillus antigen over time. An assay that detects IgM antibodies to a 875
wide range of Aspergillus antigens could therefore remain positive for some time. The 876
specificity of Aspergillus-‐specific IgM testing is poor, limiting its utility72,89,100. 877
Persistently raised levels of specific Immunoglobulin A (IgA) are found in up to 878
76% of CPA cases72,88,99–102. This immunoglobulin type is normally associated with 879
31
mucosal immunity and it may be persistently raised as the mucosa is constantly 880
exposed to fungal growth. Aspergillus-‐specific IgE levels are also sometimes raised in 881
CPA cases and may indicate the presence of underlying ABPA when present5 . 882
Measurement of Aspergillus-‐specific IgG antibodies had a higher sensitivity than 883
either IgM, IgA or IgE in all these studies and it should therefore be considered the most 884
appropriate test for screening. However small numbers of cases of CPA have been 885
identified which have normal Aspergillus-‐specific IgG, but raised Aspergillus-‐specific IgA 886
or IgM88,99,104. This may be explained by the fact that Aspergillus-‐specific IgA and IgM 887
can bind different Aspergillus antigens than Aspergillus-‐specific IgG94,100. Overall 888
Aspergillus-‐specific IgM probably has little to offer due to poor specificity, but there may 889
be a role for Aspergillus-‐specific IgA and IgE testing, in patients with symptoms and/or 890
radiological changes of CPA, but normal Aspergillus-‐specific IgG levels. 891
Measurement of Aspergillus-‐specific IgG has additional uses beyond initial 892
diagnosis of CPA. Precipitins titres fall following surgical resection of aspergilloma105 893
and rise in correlation with clinical treatment failure106. Aspergillus-‐specific IgG levels 894
have been successfully used to monitor response of CPA to medical therapy8,58,107–109. 895
896
Allergic aspergillosis 897
898
In this context the patient may initially have healthy lungs and an intact immune 899
function. However an exaggerated inflammatory response develops in response to 900
fungal allergen exposure. This is characterised by over-‐expression of T helper (Th) 2 901
and Th17 CD4+ cells and down-‐regulation of T-‐regulatory cells (TREGs). This results in 902
the high levels of both total and Aspergillus-‐specific IgE in serum in patients with SAFS 903
and ABPA64,110. Raised total and Aspergillus-‐specific IgE in serum are also noted in 904
patients with allergic fungal rhinosinusitis caused by Aspergillus. In this patient group 905
raised levels of Aspergillus-‐specific IgE can also be found in the ‘allergic mucin’ extracted 906
from the sinuses themselves111,112. 907
Raised Aspergillus-‐specific IgG has been described as an exclusion criteria for the 908
diagnosis of SAFS on the grounds it implies more complex disease and airways 909
infection110. It should be noted though, that 10% of all asthmatics have raised 910
Aspergillus-‐specific IgG or precipitins22. There are therefore likely to be some cases of 911
32
asthma with Aspergillus sensitization where Aspergillus-‐specific IgG is raised in addition 912
to IgE, but all diagnostic criteria for more severe conditions such as ABPA are not met. 913
Precipitating Aspergillus-‐specific antibodies were frequently found in ABPA cases 914
in early studies113,114. They were then considered a mandatory diagnostic criteria for 915
ABPA by some authors115,116, whereas others regarded it only as a supporting criteria110. 916
Reports on the frequency of raised Aspergillus-‐specific IgG or precipitins in ABPA will of 917
course be heavily dependent on whether or not it is considered a mandatory diagnostic 918
criteria, but 14% of ABPA cases have recently been reported as having a negative 919
precipitins test67. 920
ABPA can be complicated by the development of CPA, which is characterized by 921
raised levels of Aspergillus-‐specific IgG. However elevated Aspergillus-‐specific IgG is 922
much more common in ABPA than is the development of CPA117,118. Levels of 923
Aspergillus-‐specific IgG are generally higher in CPA than ABPA. Unusually high levels in 924
patients with ABPA may therefore suggest that CPA has developed and should prompt 925
further investigation4,5. Raised Aspergillus-‐specific IgA has also been noted in ABPA119, 926
but it occurs only in a minority of patients and is of limited diagnostic value. 927
In patients with underlying cystic fibrosis (CF) quantitative measurement of 928
Aspergillus-‐specific IgG has been suggested as a means to differentiate CF with ABPA 929
from CF without APBA78. It was found that the mean Aspergillus-‐specific IgG 930
concentration in CF patients without ABPA was 51.1 mg/L, compared to 132.5 mg/L in 931
CF patients with ABPA. The authors of this study suggested an Aspergillus-‐specific IgG 932
cut off of 90 mg/L to differentiate the two patient groups with a sensitivity of 91% and 933
specificity of 88%. 934
Latent class analysis is a statistical technique used to find groups or subtypes of 935
cases in multivariate categorical data. A recent publication used this technique to 936
identify different disease groups in relation to Aspergillus infection in CF66. Four disease 937
groups were identified; 1 – patients with no evidence of Aspergillus disease, 2 – patients 938
with serological ABPA, 3 – patients sensitized to Aspergillus and 4 – patients with 939
Aspergillus bronchitis. 940
Aspergillus-‐specific IgG could be used to differentiate between Aspergillus 941
sensitization and serological ABPA with a sensitivity of 96% and a specificity of 90% 942
when a cut off of 75 mg/L was used. Aspergillus-‐specific IgE could differentiate between 943
Aspergillus bronchitis and serological ABPA with 100% sensitivity and specificity using 944
33
a cut off 3.7 kUA/L. Patients with no Aspergillus disease could be differentiated from 945
patients with Aspergillus sensitization using Aspergillus-‐specific IgE with a cut-‐off of 2 946
kUA/L. To our knowledge the performance of the diagnostic cut-‐offs suggested by these 947
two studies have not been confirmed in populations other than the ones used to define 948
the cut-‐offs. 949
Total IgE falls with effective treatment of ABPA110,115,120–122. Aspergillus-‐specific 950
IgE can also fall with treatment110, but this effect was noted later than the fall in total 951
IgE in this study and was not reported in the majority of other treatment studies. 952
Aspergillus-‐specific IgG has also been noted to fall in line with treatment116, but this 953
occurred at the same time as a fall in total IgE and provided no additional information. It 954
therefore appears that total IgE is currently the most appropriate test for monitoring 955
response to treatment in ABPA. 956
957
LABORATORY METHODS FOR DETECTION OF ASPERGILLUS-‐SPECIFIC ANTIBODIES 958
959
Multiple techniques are available to measure levels of Aspergillus-‐specific antibodies in 960
human serum in the laboratory in different ways. Since raised Aspergillus-‐specific IgG, 961
IgE, IgA and IgM all have different interpretations in different clinical scenarios it is 962
important to understand which assays measure which antibody types when 963
interpreting results. 964
965
Precipitation in gels. 966
967
Detection of Aspergillus-‐specific antibodies in serum was first achieved by precipitation 968
of antibody-‐antigen complexes in gels123,124. This method has also been referred to as 969
double diffusion, immunodiffusion or the precipitins test. Antigens and antibodies are 970
placed in separate wells within the gel and are allowed to diffuse towards one another. 971
The presence of multiple binding sites on antibodies such as IgM125 allows the 972
formation of immune complexes that ‘precipitate’ when they become too large to pass 973
through the gel. These ‘precipitin bands’ are visible to the naked eye with non-‐specific 974
staining. 975
All antibody classes precipitate, but IgG predominates. The method takes around 976
five days to perform, is labour intensive and relies on human interpretation of results. 977
34
No complex equipment is needed. Commercial preparations of A. fumigatus antigens for 978
use in precipitins tests are available from Microgen (UK), Bio-‐Rad Laboratories (France) 979
and Immuno-‐Mycologics (IMMY) Inc. (USA). 980
981
Counterimmunoelectrophoresis. 982
983
An improvement on the precipitation method was described with the development of 984
counterimmunoelectrophoresis (CIE)124. Movement through the gel is accelerated by 985
application of an electric current and precipitation occurs within a few hours126. 986
987
Figure 2 is a picture of a CIE gel with visible precipitin bands. 988
989
Haemagglutination. 990
991
Haemagglutination tests use erythrocytes pre-‐coated with antigens. These erythrocytes 992
clump together when antibodies cross-‐react with antigens on more than one cell. The 993
resulting ‘plaque’ prevents erythrocytes from settling at the bottom of the test well. The 994
difference in appearance between positive and negative wells is visible to the human 995
eye127,128. This method produces a result in around two hours and does not require 996
complex equipment, but does rely on human interpretation of results. It is commercially 997
produced by ELITech Group (France). Antibody levels are considered raised if a positive 998
reaction takes place at a titer greater than the manufacturers stated cut off level. 999
1000
Figure 3 is a picture of a haemagglutination plate showing positive and negative results. 1001
1002
Complement Fixation. 1003
1004
Complement fixation tests rely on the fact that human complement will both react with 1005
antibody-‐antigen complexes and also lyse sheep erythrocytes that are pre-‐bound to 1006
anti-‐sheep erythrocyte antibodies129. Complement is removed from human serum by 1007
heating. Aspergillus antigens, complement and sheep erythrocytes, pre-‐bound to anti-‐1008
sheep erythrocyte antibodies are added in steps. In the absence of Aspergillus-‐specific 1009
antibodies a reaction takes place that results in lysis of the erythrocytes and thus colour 1010
35
change visible to the naked eye84. The method is fairly labour intensive and relies on 1011
human interpretation of results. Kits are produced by and Serion/Virion (Germany) and 1012
IMMY (USA). 1013
All of the above techniques can produce semi-‐quantitative results by following 1014
serial dilutions of serum. 1015
1016
ELISA. 1017
1018
This well-‐described technique allows the detection of individual types of antibody (IgG, 1019
IgM, IgA etc.). Antibodies from patient sera bind to antigens and are then detected by 1020
anti-‐human antibodies. Enzyme reactions produce a colour change that is measured 1021
with a spectrophotometer. ELISA has been used in diagnosing aspergillosis for 1022
decades130,131. It can be fully automated, which reduces labour costs and can produce 1023
results within two hours. The reaction can also be performed manually. ELISA produces 1024
a positive result in most sera, with a cut-‐off provided by the manufacturer to 1025
differentiate raised levels from normal ones. 1026
Commercial Aspergillus-‐specific IgG plate ELISA tests are currently produced by 1027
Serion, (Germany), IBL (Germany/USA), Dynamiker/Bio-‐Enoche, (China), Bio-‐Rad 1028
(France), Bordier (Switzerland) and Genesis (UK). Siemens Immulite (Germany) supply 1029
an automated Aspergillus-‐specific IgG ELISA system (Immulite) and ThermoFisher 1030
Scientific/Phadia (multi-‐national) supply an automated Aspergillus-‐specific IgG 1031
fluoroenzymeimmunoassay (FEIA) system (ThermoFisher Scientific ImmunoCAP), 1032
which is an ELISA variant. The Serion and Bio-‐Rad Aspergillus-‐specific IgG assays can 1033
also be automated. Siemens Immulite and ThermoFisher Scientific both produce 1034
automated Aspergillus-‐specific IgE ELISA/FEIA tests. Serion and IBL produce 1035
commercial Aspergillus-‐specific IgA and IgM ELISA tests. The units of measurement 1036
often differ from one assay to another. 1037
1038
1039
36
Immunoblot. 1040 1041
Gel electrophoresis is used to separate Aspergillus antigens by molecular weight. 1042
Antigens are then transferred to a membrane to which human serum is added. An 1043
identical series of reactions to ELISA is then performed, producing a colour change 1044
visible to the naked eye at the location of the antigen on the membrane when positive. It 1045
does not require complex equipment but is fairly labour intensive83. A commercial A. 1046
fumigatus immunoblot was released in 2012 by LDBIO diagnostics (France). 1047
The attributes of a selection of commonly used methods for detection of 1048
Aspergillus antibodies are summarised in table 2. 1049
1050
SOURCES OF ANTIGENS FOR USE IN ANTIBODY DETECTION ASSAYS 1051
1052
Extraction of antigens from fungal cultures 1053
1054
The traditional methods of antigen preparation for use in tests involves growth of 1055
Aspergillus culture in the laboratory, followed by either mechanical disintegration of 1056
intact cells to provide somatic antigens or culture filtration to provide extra-‐cellular 1057
antigens. The latter have often been referred to as ‘metabolic’ antigens in the literature 1058
and product information sheets. This terminology is, however inaccurate as many of the 1059
antigens are not metabolites. These crude processes produce mixtures of many of the 1060
different antigens produced by Aspergillus. Up to 52 separate precipitins bands have 1061
been identified on double diffusion testing using this type of antigen preparation132 and 1062
electrophoresis of culture extracts has identified up to 200 bands, each representing a 1063
potential antigen that might react with human sera133. 1064
While the extraction of antigens from Aspergillus cultures has been taking place 1065
for decades there have been several difficulties encountered in attempts to provide 1066
consistent and reliable antigens for use in tests. It is clear that different laboratory 1067
strains of Aspergillus fumigatus produce different groups of antigens128,134–138. Even 1068
when a single strain is used somatic and culture filtrate methods produce different 1069
groups of antigens134, which can produce different results when tested against patient 1070
sera139. Various factors such as the culture medium used, pH, and culture temperature 1071
37
have all been noted to affect the nature of antigens produced by cultures140. Antigens 1072
also vary with the age of the culture98,135. 1073
Even when identical methods are used, batch to batch variation from a single 1074
strain processed in the same lab has been noted141. In addition to antigens, culture 1075
extracts also contain enzymes and toxins, which might interfere with test 1076
performance142. When the same antigen extracts are used in different test formats they 1077
can produce widely variable results143. The antigen mixtures produced from culture 1078
extracts have also been shown to cross-‐react with antibodies produced against other 1079
fungi and bacteria144,145. 1080
As these traditional antigen extraction techniques can be performed in any 1081
mycology laboratory, reference laboratories often produce their own internally 1082
manufactured antigens for use in assays59,86,96. However the extensive difficulties noted 1083
above mean that quality control in Aspergillus antigen production is exceedingly 1084
challenging and by their nature internally manufactured assays in reference 1085
laboratories are not amenable to validation in inter-‐laboratory studies. In contrast 1086
commercially manufactured assays can be performed and assessed across multiple 1087
laboratories and can also be compared to other assays under identical conditions in a 1088
single laboratory. 1089
1090
Measurement of antibodies in non-‐fumigatus aspergillosis 1091
1092
All the tests described above are designed to detect A. fumigatus. However, in India the 1093
most prevalent Aspergillus species causing fungal sinusitis is A. flavus10. This species 1094
also accounts for 38% of Aspergillus cultures from patients with chronic lung diseases in 1095
India146. In Brazil Aspergillus niger is a common cause of chronic pulmonary 1096
aspergillosis147. The frequency of growth of different Aspergillus species in association 1097
with human disease in selected countries is shown in table 3. 1098
Evidence on the performance of antibody detection assays in these cases is 1099
extremely limited. Culture filtrate antigens from A. fumigatus are positive in around 1100
50% of cases with aspergilloma caused by A. flavus or A. niger148. A. niger-‐specific 1101
precipitins were positive in 78% of 23 patients with CPA due to A. niger in Brazil147. 1102
Other species-‐specific precipitins tests are available and might prove effective, but have 1103
been tested on very few patients98. Siemens Immulite produce ELISA tests for IgG 1104
38
specific to Aspergillus niger, nidulans, terreus and flavus, but to our knowledge there are 1105
no published data on the performance of these assays. 1106
1107
Detection of antibodies specific to individual Aspergillus antigens 1108
1109
Early experience with precipitins testing demonstrated that precipitin bands of 1110
consistent molecular weight appeared in many patients with aspergillosis and 1111
corresponded to enzymes associated with the fungus132,149. Individual antigens were 1112
identified, which had variable sensitivity and specificity for the diagnosis of 1113
aspergillosis. Many specific antigens reacting with human IgG and IgE have since been 1114
identified150,151 and the genes relating to these antigens have been sequenced152. This 1115
has allowed the production of recombinant antigens by expressing these genes in 1116
genetically modified bacteria or fungi, which then produce pure extracts of single 1117
antigen. 1118
Mitogillin-‐specific IgG is positive in 100% of aspergilloma cases, 64% of invasive 1119
pulmonary aspergillosis cases and only 1.3% of healthy volunteers in a single study88. 1120
Antibodies to purified recombinant Afmp1p, an Aspergillus cell wall 1121
galactomannoprotein, are positive in 100% of patients with aspergilloma and 33% of 1122
patients with invasive aspergillosis153. To our knowledge the performance of these 1123
assays has not been confirmed in other laboratories and the assays have not been 1124
released commercially. 1125
Testing for IgG specific to recombinant catalase, ribonuclease and 1126
dipeptidylpeptidase V showed sensitivity of 77%, 81% and 79% respectively for 1127
aspergilloma. This increased to 95% by using all three antigens together154. Bio-‐Rad 1128
(France) released a commercial recombinant assay following this study. It has shown 1129
good agreement with Serion culture filtrate ELISA in a retrospective survey38. Bio-‐Rad 1130
has not revealed which antigens are used in their commercialized test. 1131
The Dynamiker Aspergillus-‐specific IgG ELISA assay utilizes purified 1132
galactomannan as its sole antigen. No data has yet been published on the sensitivity and 1133
specificity of this test for the diagnosis of aspergillosis, but an earlier study detected 1134
antibodies to galactomannan in only 26% of aspergilloma cases155. 1135
Many efforts have been made to identify individual antigens against which 1136
specific IgE is formed in allergic aspergillosis156,157. These antigens are commonly 1137
39
referred to as allergens in this context. To date 23 Aspergillus-‐specific allergens have 1138
been recognised by the International Union of Immunology Societies158. This is likely to 1139
be an under-‐representation of the true number of Aspergillus-‐specific allergens as 81 1140
IgE binding Aspergillus proteins have been identified using a highly sensitive phage 1141
display detection method159. 1142
Attempts have been made to develop individual allergen-‐specific IgE assays for 1143
use in allergic aspergillosis and to use them to differentiate between different diseases. 1144
IgE to allergen Asp f1 is found in 60-‐85% of ABPA cases160,161, but has also been 1145
detected in the sera of Aspergillus sensitized asthmatics without ABPA156. Genomic 1146
studies have demonstrated that sensitization to this allergen is produced only by a small 1147
number of fungi162, suggesting that there is likely to be limited cross-‐reactivity with this 1148
recombinant protein. 1149
One study found IgE specific to allergens Asp f2/3/6 were all raised in both 1150
asthma and ABPA, but not in other forms of pulmonary aspergillosis. However another 1151
study found that IgE specific to allergens Asp f1/2/3/4/6 were all present at 1152
significantly higher concentrations in ABPA than asthma151. 1153
In patients with underlying cystic fibrosis, one study showed mean IgE to Asp f1 1154
was ten times higher in those with ABPA than those without163, but another study failed 1155
to show this differentiation164. IgE to Asp f4 and Asp f6 were found to differentiate CF 1156
with ABPA from CF without ABPA in this second study. A similar result was later found 1157
when these same antigens were used in skin prick testing165. 1158
A more recent study showed that no single allergen was absolutely effective in 1159
differentiating CF patients with and without ABPA166. IgE to Asp f1 showed non-‐specific 1160
binding with ABPA cases and controls, IgE to Asp f2 was consistently present in the sera 1161
of CF patients with ABPA, but was frequently also present in CF patients without ABPA. 1162
IgE to Asp f3 was highly specific for ABPA in CF but had poor sensitivity. Aspergillus-‐1163
specific IgG subtypes and IgA were also analyzed and found not to differentiate CF with 1164
ABPA from CF without ABPA. 1165
Attempts have also been made to identify single antigen-‐specific antibodies for 1166
the diagnosis of acute invasive aspergillosis83,167, but to our knowledge no commercial 1167
assays have been released for this purpose and detection of serum antigenaemia is 1168
preferred in this patient group due to its superior sensitivity48. 1169
40
Overall, while the detection of antibodies specific to individual antigens might 1170
eventually result in more accurate and reproducible diagnosis of aspergillosis, existing 1171
study results are mostly inconsistent or unconfirmed. No individual antigen or group of 1172
antigens has been consistently shown to be more efficacious than traditional methods of 1173
antigen extraction for the diagnosis of any form of aspergillosis. 1174
1175
COMPARATIVE PERFORMANCE OF DIFFERENT LABORATORY METHODS 1176
1177
Invasive aspergillosis 1178
1179
Antibody measurement plays a peripheral role in the diagnosis of invasive aspergillosis 1180
and data on the comparative performance of different techniques is limited23. 1181
Aspergillus-‐specific IgG ELISA was more sensitive than precipitins or CIE in two studies 1182
with a total of 18 patients82,87. Comparison of haemagglutination and Aspergillus-‐1183
specific IgG ELISA showed superior sensitivity for haemagglutination in one study with 1184
14 patients84, but superior sensitivity for ELISA in another study with 26 patients86. To 1185
our knowledge there are no comparisons of currently commercially produced 1186
Aspergillus antibody assays in this patient group, although the Serion Aspergillus-‐1187
specific IgG ELISA formed part of a mix of methods for antibody detection that were less 1188
sensitive than galactomannan antigen test in one comparison90. 1189
1190
Chronic pulmonary aspergillosis 1191
1192
The original reports of precipitins tests for diagnosis of aspergilloma reported 1193
sensitivity of 98% against patients with definite histological or radiological diagnosis of 1194
aspergillosis, with no positive results in healthy controls98. However it should be noted 1195
that the radiological methods available at the time did not include computed 1196
tomography (CT) scanning and would thus only have detected cases with fairly 1197
advanced disease. 1198
Since then precipitins detection has been used as part of the diagnostic criteria 1199
for chronic forms of aspergillosis. The lack of a clear gold standard creates a difficulty in 1200
41
subsequent studies. Sensitivity is normally measured against clinical diagnosis recorded 1201
in the patients’ medical records. Precipitins will often have formed part of the 1202
diagnostic criteria. It is therefore difficult to prove that other tests are more sensitive 1203
than precipitins in study populations defined in this way. 1204
In more recent studies the interpretation of reported sensitivity rates against 1205
diagnoses of CPA taken from case notes might be further complicated by the fact that 1206
many patients will be on antifungal treatment, which leads to reduction in Aspergillus 1207
specific IgG levels58. It is not known whether this would affect all tests equally or bias 1208
results in favor of one technique. Prospective studies comparing performance of tests in 1209
patients not yet diagnosed with aspergillosis would resolve these issues, but are 1210
difficult to conduct due to the low frequency of new diagnoses. One such study recently 1211
demonstrated that Aspergillus-‐specific IgG is more sensitive than serum galactomannan 1212
antigen for the diagnosis, but did not compare different Aspergillus-‐specific IgG assays59. 1213
Many retrospective studies have shown equally excellent sensitivity when 1214
precipitins testing is compared to CIE or ELISA in cases of aspergilloma168–173. 1215
Precipitins were even reported as being more sensitive than other methods in one 1216
comparison174. However not all cases of aspergilloma or CPA are precipitins 1217
positive7,175. Negative precipitins results might occur as not all antibody-‐antigen 1218
complexes precipitate in gels176. 1219
The one prospective study comparing precipitins to CIE showed that CIE is more 1220
sensitive than traditional precipitins for detection of Aspergillus-‐specific antibodies177. 1221
However CIE has also been reported as producing more false positive results than 1222
precipitins170,178,179. A recent retrospective study suggested that the sensitivity of ELISA 1223
for the diagnosis of CPA was 30% higher than precipitins74. 1224
Unlike ‘home-‐brew’ assays using internally manufactured Aspergillus antigens, 1225
commercially available assays can be compared to one another in head-‐to-‐head 1226
comparisons. Commercial ELISAs with published performance data for the diagnosis of 1227
CPA include ThermoFisher Scientific Aspergillus-‐specific IgG FEIA, Serion culture filtrate 1228
Aspergillus-‐specific IgG ELISA and Bio-‐Rad recombinant Aspergillus-‐specific IgG 1229
ELISA38,73. Each showed good correlation with precipitins test results and superior 1230
reproducibility when automated. The Siemens Immulite and ThermoFisher Scientific 1231
assays have good head-‐to-‐head correlation, but the Siemens Immulite assay produces a 1232
higher absolute result with a mean ratio of 1.78180. This study also demonstrated 1233
42
acceptable inter-‐laboratory reproducibility for the ThermoFisher Scientific with a co-‐1234
efficient of variation of 7.3 -‐ 18.1%. 1235
ThermoFisher Scientific ImmunoCAP Aspergillus-‐specific IgG FEIA, Bio-‐Rad 1236
recombinant Aspergillus-‐specific IgG ELISA and CIE using Microgen antigens were 1237
compared in 116 patients with CPA, who were mostly on antifungal treatment74. 1238
Sensitivity was 86% for ThermoFisher Scientific ImmunoCAP , 85% for Bio-‐Rad and 1239
56% for CIE. However 4% of cases were positive on precipitins testing only. This may 1240
be due to the ability of precipitins to detect IgM and IgA in addition to IgG. In the case of 1241
the Bio-‐Rad recombinant antigens assay, false negative results may also occur in 1242
patients who do not have antibodies to the selected antigens within their spectrum of 1243
anti-‐Aspergillus antibodies. These results suggests that while these ELISAs are more 1244
sensitive than precipitins testing for first line screening, there may be a role for 1245
precipitins testing in patients suspected of CPA with unexpectedly negative ELISA 1246
results. 1247
The Bio-‐Rad test has also been directly compared to Serion ELISA in 51 cases 1248
with CPA38. Sensitivities of 94% and 92% respectively were noted. Specificity in healthy 1249
controls was 100% for Bio-‐Rad and 96% for Serion. 1250
The published comparisons of the sensitivity of different methods of Aspergillus-‐1251
specific antibody detection in patients with CPA are summarised in table 4. 1252
1253
Allergic pulmonary aspergillosis 1254
1255
A recent review compared the performance of different diagnostic tests for identifying 1256
new cases of ABPA in Indian asthmatics using latent class analysis62. Aspergillus skin 1257
prick testing was 95% sensitive and 80% specific, total IgE of >1000 IU/ml was 97% 1258
sensitive but only 40% specific, raised Aspergillus specific IgE was 100% sensitive and 1259
70% specific, whereas Aspergillus precipitins testing was only 43% sensitive, but 97% 1260
specific. 1261
These results suggest that Aspergillus-‐specific IgE testing is the most appropriate 1262
screening test for ABPA and can be used in place of skin prick testing where available. 1263
However the high specificity of precipitins testing means that the diagnosis of ABPA can 1264
be made with high confidence in asthmatic patients with both raised Aspergillus-‐specific 1265
43
IgE and positive Aspergillus precipitins. Unfortunately most patients with ABPA in this 1266
study did not meet all of these criteria. 1267
CIE has been reported as more sensitive than precipitins for the detection of 1268
precipitating antibodies in cases of allergic aspergillosis181. There are no published 1269
direct comparisons of the performance of the commercially available Aspergillus-‐1270
specific IgE assays, but it should be noted that marked variation has been noted 1271
between Aspergillus-‐specific IgE levels and skin prick test results, with concordance of 1272
only 14-‐56%65,182,183. There is also marked variation between the Siemens Immulite and 1273
ThermoFisher Scientific assays in tests for peanut-‐specific IgE184. The Siemens Immulite 1274
system produces Aspergillus-‐specific IgG results roughly 2 fold higher than the 1275
ThermoFisher Scientific ImmunoCAP system180. Results of Aspergillus-‐specific IgE 1276
assays from different commercial assays should therefore be compared with caution. 1277
The published comparisons of the sensitivity of different Aspergillus-‐specific 1278
antibody assays are summarised in table 5. 1279
1280
SUITABILITY OF AVAILABLE LABORATORY TECHNIQUES FOR RESOURCE-‐POOR 1281
SETTINGS 1282
1283
As noted earlier the majority of patients suffering from pulmonary aspergillosis are 1284
likely to be located in resource poor settings. We would suggest that many commonly 1285
used assays are not ideal for use in such settings. Automated ELISAs require equipment, 1286
which is expensive to purchase and requires both a reliable electricity supply and 1287
regular maintenance. Manual ELISAs might be suitable, but still require a properly 1288
maintained spectrophotometer that may not be available in many resource poor 1289
settings. Such manual ELISAs have been described as having much poorer 1290
reproducibility than automated systems74. 1291
Precipitation in gels requires less high-‐tech equipment than ELISA, but is time 1292
consuming, requires significant operator training and produces subjective results. 1293
Complement fixation and immunoblot have similar difficulties. We consider 1294
haemagglutination assays a potentially attractive option as no complex equipment is 1295
required, but to our knowledge there are no published data describing the sensitivity 1296
and specificity of the sole commercially available haemagglutination test (ELITech). 1297
44
The lateral flow device (LFD) is well known for its use in point-‐of-‐care pregnancy 1298
tests. This format is also widely used for the diagnosis of HIV and malaria in resource 1299
poor settings185,186. To our knowledge no LFD for the detection of Aspergillus-‐specific 1300
antibodies exists at this time. An LFD that detects an Aspergillus antigen has recently 1301
been developed and seems to perform well using serum for the diagnosis of acute 1302
invasive aspergillosis in mostly neutropaenic patients187–189. It is also effective using 1303
BAL fluid to diagnose invasive aspergillosis in non-‐neutropaenic patients with 1304
underlying lung disease190. However to our knowledge there is no published evidence 1305
regarding its sensitivity and specificity for the diagnosis of CPA. It is possible that in this 1306
context the sensitivity of this LFD will be low, as the alternative galactomannan antigen 1307
assay has poor sensitivity in this patient group8,59,60. 1308
Figure 4 shows examples of tests that are unsuitable and potentially suitable for 1309
use in resource-‐poor settings. 1310
1311
CONCLUSIONS 1312
1313
Aspergillosis has been estimated to affect millions of persons worldwide, with CPA as 1314
the most common clinical syndrome. Many of these patients are likely to reside in 1315
resource-‐poor countries, given the current and previous prevalence of tuberculosis in 1316
these locations. Improved diagnosis of CPA is a critical need in the battle to improve 1317
CPA outcomes and expanding access to Aspergillus-‐specific IgG testing in areas of high 1318
tuberculosis prevalence is key to achieving this goal. 1319
Expanding the diagnosis of aspergillosis presents many challenges. The clinical 1320
and radiological signs of aspergillosis often overlap significantly with associated 1321
underlying diseases and so cannot be relied upon to diagnose aspergillosis. Culture can 1322
be helpful, but the sensitivity of culture for the diagnosis of aspergillosis is sub-‐optimal 1323
and access to reliable fungal culture is frequently challenging or even non-‐existent in 1324
poorly resourced countries. 1325
Serological testing is therefore of crucial importance. For acute invasive 1326
aspergillosis this mostly means antigen testing, which has been reviewed extensively 1327
elsewhere. However there may be a secondary role for antibody testing in this setting 1328
for retrospective diagnosis of recovering patients. The screening of patients for 1329
45
evidence of Aspergillus colonisation prior to immunosuppressive therapy may also be 1330
useful. Outside of this setting the interpretation of raised levels of Aspergillus-‐specific 1331
antibodies in asymptomatic colonised patients is not well described and follow up 1332
studies of such patients that describe their risk of developing symptomatic forms of 1333
aspergillosis would be welcome. 1334
In chronic and allergic aspergillosis measurement of Aspergillus-‐specific 1335
antibodies is central to diagnosis, with raised Aspergillus-‐specific IgG found mostly in 1336
chronic disease and raised total and Aspergillus-‐specific IgE found mostly in allergic 1337
disease. It is important to note, though that there is a degree of overlap between these 1338
clinical syndromes and many patients will have clinical and serological features of both. 1339
Similarly subacute invasive aspergillosis occurs in mildly immunosuppressed 1340
patients with a presentation that overlaps acute invasive disease and CPA. Here patients 1341
may have positive antigen tests, raised Aspergillus-‐specific IgG or both simultaneously. 1342
As a result it is possible that this group of patients will need to be tested for both 1343
Aspergillus-‐specific IgG and Aspergillus antigens to achieve early diagnosis with good 1344
sensitivity. 1345
Measurement of antibodies can also be used to monitor response to treatment. A 1346
falling Aspergillus-‐specific IgG indicates poor prognosis in acute invasive aspergillosis, 1347
but a good response to therapy in CPA. For allergic aspergillosis, total IgE remains the 1348
best method for monitoring treatment response, although it is far from optimal. 1349
Many methods exist for the measurement of Aspergillus-‐specific antibodies, with 1350
differing performance characteristics. It is thus unfortunate that they are frequently 1351
mislabeled in the literature with the term ‘precipitins’ often used to refer to Aspergillus-‐1352
specific IgG ELISA rather than precipitation in a gel and ‘RAST’ often used to refer to 1353
Aspergillus-‐specific IgE ELISA rather than the older radioimmunoassay. 1354
Evidence of sensitivity and specificity of different methods is sparse, but 1355
Aspergillus-‐specific IgG ELISA is likely to be more sensitive than precipitation in gels. 1356
However there are some patients with CPA with normal Aspergillus-‐specific IgG ELISA 1357
results and positive precipitins tests or raised levels of Aspergillus-‐specific IgA. 1358
Performing these assays in patients suspected of CPA with negative Aspergillus-‐specific 1359
IgG ELISA would therefore probably result in better overall sensitivity. 1360
Aspergillus-‐specific IgM ELISA is probably not useful for diagnosis of CPA due to 1361
poor specificity, although it should be noted that the specificity data comes from studies 1362
46
of ‘home-‐brew’ assays. The commercially produced Aspergillus-‐specific IgM assays 1363
might have different performance characteristics, but to our knowledge there are no 1364
published data on this topic 1365
The product inserts of most commercial ELISAs report good specificity at the 1366
manufacturers diagnostic cut-‐offs, but the evidence for these statements is often not 1367
published in peer-‐reviewed journals. It should be noted that these cut-‐offs are normally 1368
calculated against the range of antibody levels found in a cohort of healthy volunteers. 1369
This is probably an appropriate comparator for most invasive aspergillosis patients. 1370
However healthy volunteers may not be the ideal comparator for CPA or ABPA, as these 1371
conditions almost always occur in persons with underlying chronic lung disease or 1372
chronic immune dysfunction. Unfortunately, to our knowledge there is no published 1373
data on the distribution of Aspergillus-‐specific IgG levels in patients with these chronic 1374
underlying conditions, with the exception of cystic fibrosis. Our research team is 1375
undertaking a study measuring Aspergillus-‐specific IgG levels in patients with treated 1376
tuberculosis, COPD and asthma using several assays. The diagnostic cut-‐offs for CPA and 1377
ABPA may need to be changed in response to this data. 1378
Global standardization of assays has proved difficult, with many laboratories 1379
using assays derived from antigens manufactured ‘in-‐house’. By their nature these 1380
assays are impossible to validate in other laboratories. Many commercially produced 1381
Aspergillus-‐specific IgG and IgE tests exist, but to our knowledge only one 1382
(ThermoFisher Scientific / ThermoFisher Scientific ImmunoCAP) has published inter-‐1383
laboratory variability data. The Bio-‐Rad recombinant Aspergillus-‐specific IgG has been 1384
tested against reasonable number of persons with CPA at more than one centre with 1385
good sensitivity reported. The IBL and ThermoFisher Aspergillus-‐specific IgG assays 1386
have been tested in reasonable numbers of patients with CPA at single sites. Most 1387
patients in all of these studies will have been on treatment and it is not known how this 1388
may have biased the results. Many other assays have no published performance data at 1389
all. 1390
The publication of data from studies demonstrating the reliability of available 1391
assays both in terms of sensitivity and specificity in untreated patients and in terms 1392
inter-‐assay and inter-‐laboratory reliability is a pre-‐requisite for their use in the large 1393
scale screening that will be necessary to achieve diagnosis of the predicted number of 1394
47
cases. Our unit is currently undertaking a single centre study with this goal, but studies 1395
across multiple laboratories will be needed to determine inter-‐laboratory variability. 1396
Many attempts have been made to develop ELISAs for the detection of antibodies 1397
specific to one or more individual Aspergillus antigens and commercially produced tests 1398
based on this principle do exist. In theory this should allow production of a reliable test 1399
and resolve the many problems that exist with traditional antigen extraction 1400
techniques. However, to our knowledge there is no published evidence that these assays 1401
are consistently either more reliable or efficacious than traditional techniques for the 1402
diagnosis of either allergic or chronic aspergillosis. Assays based on culture filtrate or 1403
somatic antigens remain in common usage. 1404
As the majority of patients with pulmonary aspergillosis are predicted to live in 1405
resource-‐poor settings it will be necessary to identify a reliable test that is suitable for 1406
widespread use in such settings if such patients are to be diagnosed and treated. The 1407
haemagglutination assay may be suitable for use in this setting, but requires further 1408
validation. The Aspergillus antigen LFD is in the ideal test format, but is likely to have 1409
poor sensitivity for the diagnosis of CPA. An LFD that detects Aspergillus-‐specific IgG 1410
may need to be developed to allow widespread access to testing in resource poor 1411
settings.1412
48
1413
Table 1 – Abbreviated diagnostic criteria for acute pulmonary IA, sub acute 1414
pulmonary IA , CCPA and Aspergillus bronchitis 1415 Proven
Invasive48 Probable Invasive48
Sub Acute Invasive (aka CNPA)6
CCPA5,7,8,21 Aspergillus bronchitis39
ABPA4
Clinical criteria
not required
neutropaenia OR stem cell transplant OR high dose corticosteroids for >3 weeks OR Immune-‐suppressant drugs OR CGD OR SCID
>1 MONTH SYMPTOMS; weight loss OR productive cough OR haemoptysis AND absence of host factors for acute invasive disease
3 MONTHS SYMPTOMS; weight loss OR productive cough OR haemoptysis AND absence of host factors for invasive disease
persistent productive cough OR recurrent chest infections AND does not meet diagnostic criteria for chronic or allergic aspergillosis
asthma OR cystic fibrosis
Radiological criteria on CXR or CT scan
not required
dense lesions +/-‐ halo sign OR air-‐crescent sign OR one or more cavities
new cavitation OR expanding cavity OR paracavitary infiltrates
new cavitation OR expanding cavity OR paracavitary infiltrates
absence of changes consistent with CPA or ABPA
transient opacifications or permanent evidence of bronchiectasis of pleuro-‐pulmonary fibrosis (see other criteria below)
Laboratory criteria
culture from a sample from a normally sterile site OR histology
culture from sputum or BAL OR GM in blood or BAL OR ß(1,3)-‐D-‐glucan in blood
culture from sputum or BAL OR GM in blood or BAL OR ß(1,3)-‐D-‐glucan in blood OR raised Aspergillus-‐specific IgG OR histology
raised Aspergillus-‐specific IgG OR culture from sputum or BAL OR GM in blood or BAL* OR ß(1,3)-‐D-‐glucan in blood*
raised Aspergillus-‐specific IgG AND EITHER recurrent culture growth from sputum or BAL OR persistently positive PCR from sputum or BAL
Obligatory Criteria total IgE > 1000 IU/ml AND raised Aspergillus-‐specific IgE (or positive skin prick test) Other criteria (2 of 3 needed) raised eosinophil count OR raised Aspergillus-‐specific IgG / precipitins OR radiological changes as above
CNPA = chronic necrotising pulmonary aspergillosis, CCPA = chronic pulmonary aspergillosis, ABPA = allergic 1416 bronchopulmonary aspergillosis, CGD = chronic granulomatous disease. SCID = severe combined immunodeficiency, 1417 CXR = chest X-‐ray, CT = computed tomography, BAL = bronchoalveolar lavage, GM = galactomannan antigen test, IgG 1418 = immunoglobulin g, IgE = immunoglobulin e, PCR = polymerase chain reaction. Unless stated otherwise patients 1419 must meet all 3 criteria for diagnosis of each condition. *GM and ß(1,3)-‐D-‐glucan are less sensitive than 1420 Aspergillus serology in CPA and so not included in all published case definitions, but are consistent with CPA when 1421 present together with appropriate clinical and radiological features. 1422
49
Table 2– Comparison of the features of selected commercial Aspergillus antibody 1423
assays 1424 Test CIE Thermo
Fisher Scientific IgG FEIA
Siemens IgG ELISA
Bio-‐Rad IgG ELISA
Serion IgG ELISA
Dynamiker IgG ELISA
ELITech HA
LDBIO Immuno blot
Antigen type
fungal extract
fungal extract
fungal extract
unspecified recombinant antigen
fungal extract
galacto-‐mannan
fungal extract
fungal extract
Volume (µL)
10 140 (dead volume = 100 )
255 (dead volume = 250)
10 10 1 50 10
Dilutions titres as required
1 if result > 200mg/L.
1 if result > 200mg/L.
1 pre-‐test and second in samples with high result
2 pre-‐test and third in samples with high result
1 pre-‐test and second in samples with high result
titres as required
none
Units dilution titres
mg/L mg/L AU/ml
U/ml AU/mL dilution titres
n/a
No samples tested per batch
30 + 2 controls*
continuous testing
continuous testing
92 + 4 controls
92 + 4 controls.
92 +6 controls
94 + 2 controls*
1
Equipment needed
gels antigens Coomassie blue stain, de-‐stain and washing solution CIE tank
Phadia 100 analyzer and antigen packs. test tubes. barcode labels
Siemens Immulite analyzer and antigen packs. test tubes barcode labels
kit pipettes test tubes incubator spectro-‐photometer OR automated analyzer
kit pipettes test tubes moist chamber incubator distilled water spectro-‐photometer OR automated analyzer
kit pipettes test tubes incubator distilled water spectro-‐photometer
kit pipette
pipette tweezers rocking tray
Suitable for a resource poor laboratory?
YES NO NO YES (if manual)
YES (if manual)
YES YES YES
Total batch time
2 days 3 hours 2 hours 4 hours 4 hours 4 hours 2 ½ hours
3 hours
Hands on time-‐ approx
4 hours 30 mins 30 mins 2 hours 2 hours 2 hours 30 mins 1 hour
CIE = counterimmunoelectrophoresis, IgG = immunoglobulin g, FEIA = fluoroenzyme immunoassay, 1425 ELISA = enzyme immunoassay, HA = haemagglutination, AU = arbitrary units. *Represents total number 1426 of sera wells per test. Can perform this many screening tests in one batch or use 1 well for each serial 1427 dilution if dilutional titres are required. 1428
1429
50
Table 3 – Frequency of different Aspergillus species grown in different respiratory 1430 conditions 1431 1432 Paper Country Disease No of
cases A. fumigatus (%)
A. niger (%)
A. flavus (%)
A. terreus (%)
Baddley 2009191
USA invasive aspergillosis
274 isolates
66 10 10 9
Herbrecht 2002 20
International invasive aspergillosis
110 77 8 6 5
Denning 20035
UK CPA 10 100 none none none
Baxter 201366
UK cystic fibrosis 39 100 none 3 none
Camuset 2007108
France CPA 21 95 none 5 none
Nam 20106 South Korea subacute invasive aspergillosis + CPA
34 91 9 3 none
Jhun 20138 South Korea CPA 18 78 22 17 none Ohba 20127
Japan CPA 75 68 15 4 none
Kurhade 2002192
India treated tuberculosis
14 79 14 7 none
Shahid 2001146
India ‘chronic lung diseases’
12 67 33 none none
Michael 2008193
India allergic Aspergillus rhinosinusitis
125 11 3 79 1
invasive Aspergillus rhinosinusitis
34 26 9 59 6
Prateek 2013194
India Aspergillus rhinosinusitis
16 19 none 75 6
CPA = chronic pulmonary aspergillosis. Note multiple species identified in some cases. 1433
1434
1435
51
Table 4 –Direct comparisons of sensitivity of antibody tests in proven CPA / 1436
aspergilloma 1437 Paper No of
patients DD (%)
CIE (%)
HA (%)
Culture filtrate IgG ELISA (%)
ThermoFisher Scientific ImmunoCAP FEIA (%)
Bio-‐Rad recombinant IgG ELISA (%)
Bio-‐Rad galactomannan antigen test (%)*
Dee 1975168
9 89 89 -‐ -‐ -‐ -‐ -‐
Warnock 1977171
5 100 100 -‐ -‐ -‐ -‐ -‐
Kurup 1978170
23 87 91 100 -‐ -‐ -‐ -‐
Kauffman 1983169
13 100 -‐ -‐ 100 -‐ -‐ -‐
Mishra 198387
17 100 100 -‐ 100 -‐ -‐ -‐
Gugnani 1990173
5 100 -‐ -‐ 100 -‐ -‐ -‐
Faux 1992172
11 100 -‐ -‐ 100 -‐ -‐ -‐
Kitasato 200960
28 89 -‐ -‐ -‐ -‐ -‐ 50
Guitard 201238
51 -‐ -‐ -‐ 92 (Serion)
94 -‐
Baxter 201374
116 56 -‐ -‐ -‐ 86 85 -‐
Jhun 20138
47 -‐ -‐ -‐ 99 (IBL)
-‐ -‐ 23
Shin 201459
168 98 -‐ -‐ -‐ -‐ -‐ 23
1438 CPA = chronic pulmonary aspergillosis, DD = double diffusion (precipitins), CIE = 1439 counterimmunoelectrophoresis, HA = haemagglutination, IgG = immunoglobulin g, 1440 ELISA = enzyme immunoassay, FEIA = fluoroenzyme immunoassay. *galactomannan 1441 positive when index ≥0.5 1442 1443
1444
1445
1446
1447
1448
52
1449
Table 5 – Direct comparisons of sensitivity of antibody and antigen tests in 1450
invasive aspergillosis 1451
1452
Paper Clinical group
No. of patients
DD (%)
CIE (%)
HA (%)
IgG ELISA (%)*
Serum GM (%)
Holmberg 198082
autopsy proven IA
10 -‐ 70 -‐ 80 -‐
Mishra 198387
IA 8 37 50 -‐ 75 -‐
Manso 199485
mixed proven and probable IA
18 55 -‐ -‐ -‐ 38 (LA)
Kappe 199684
biopsy proven IA
14 -‐ -‐ LD – 29 Roche -‐ 36 Fumouze – 36
29 -‐
Kappe 200486
biopsy proven IA
26 -‐ -‐ 8 22 -‐
Herbrecht 200296
definite IA 31 68 64 probable IA 67 58 16 possible IA 55 70 25 all IA 133 64 29
Cornillet 200690
neutropaenic IA
52 6.25 (mix of DD, CIE and Serion ELISA)
64
non-‐neutropaenic IA
36 48 (mix of DD, CIE and Serion ELISA)
65
all IA patients
88 30 (mix of DD, CIE and Serion ELISA)
65
DD = double diffusion (precipitins), CIE = counterimmunoelectrophoresis, HA = 1453 haemagglutination, IgG = immunoglobulin g, ELISA = enzyme immunoassay, GM = 1454 galactomannan antigen test (ELISA unless stated otherwise), LA = latex agglutination, IA 1455 = invasive aspergillosis. *IgG ELISA tests are internally manufactured by the research 1456 laboratory unless stated otherwise. 1457
53
FIGURES 1458
1459
1 – Visual representation of the number of patients with each condition and the number 1460
of patients where each test is diagnostic 1461
1462 Note – IA = invasive aspergillosis, CPA = chronic pulmonary aspergillosis, ABPA = allergic 1463
bronchopulmonary aspergillosis. The size of each circle is relative to the estimated European 1464
population affected by the disease, with prevalence used for the chronic conditions ABPA 1465
(887,000 cases) and CPA (240,000 cases) and incidence for the acute condition IA (63,000 1466
cases)29. 1467
1468
The length of the bars represents the total number of patients where each test is diagnostic by 1469
combining frequency of positive results annually for each condition. Aspergillus-specific IgE 1470
is raised in almost all cases of ABPA4,22 and up to 66% of CPA cases5. Aspergillus-specific 1471
IgG is raised in 65% of ABPA (Smith and Denning, unpublished data), up to 100% of CPA5,8 1472
and up to 65% of cases of IA96. Aspergillus antigen tests are positive in around 62% of cases 1473
of IA in adults24 and 23% of cases of CPA8. 1474
2 - Picture of a CIE gel, with visible precipitin bands 1475
1476 Blue-stained precipitin lines are formed where antigens and antibodies meet and precipitation 1477
of antibody-antigen complexes occurs. They represent a positive result. Sera in the left hand 1478
column produced no lines and are negative. 1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
55
3 – Haemagglutination assay 1491
ELITech(Aspergillus(IHA(–(fresh(sera(
(((((((1:80(((((((1:160(((((((((1:320((((((((1:640((((((1:1280(((((1:2560((((((Serum(((((Reagent((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((control(((((control(
(Microgen(+(
Kit(+(
Kit(G(
44305((1:8)(
44380((1:4)(
44426((1:4)(
44429((neg)(
44460((1:2)(
Samples((precipiKn(Ktre)(
1:320(
1:320(
>1:2560(
1:640(
1:320(
1:80(
Neg(
>1:2560(
1492 The ELITech haemagglutination assay can be performed with no equipment other than a 1493
pipette. Results are visible to the naked eye. In the image above each row is a test sample 1494
with dilutional titres increasing from left to right. Result is the last titre at which a ‘plaque’ is 1495
still visible as shown. 1496
1497
1498
56
1499 4 – Suitability of tests in resource-poor settings 1500
1501 The automated ELISA machine shown is of little use in settings with no regular electricity, 1502
whereas lateral flow devices, such as the Aspergillus antigen LFD above, are ideal. 1503
1504
1505
1506
57
Part 5 – CPA as Global Public Health issue 1507
1508
The presence of a fungal ball in a human was first documented by Plaignaud in the 18th 1509
century195, with fungal growth in human tubercular lung cavities described half a 1510
century later by Bennett196. The first survey of the frequency of pulmonary aspergillosis 1511
secondary to tuberculosis was undertaken by the British Medical Research Council 1512
(MRC) in 197076,197 and the modern syndrome of CPA was defined in 20035. CPA occurs 1513
in patients with underlying structural lung disease. Underlying conditions include 1514
COPD, sarcoidosis, and cystic fibrosis, but the most common underlying condition in the 1515
UK is treated pulmonary tuberculosis14. Large CPA case series have now been reported 1516
in Europe, India, China, Korea and Japan and the large majority of CPA cases described 1517
are secondary to tuberculosis7,8,14,15,18,108,198. 1518
1519
Aspergillosis has been known to exist in Africa for some time. Aspergillus sinusitis has 1520
been well recognized in Sudan since the 1960s199 and Aspergillus has documented as the 1521
most common cause of otomycosis in Nigeria200. The first documented case of 1522
pulmonary aspergillosis in Africa was that of an aspergilloma in a South African farmer 1523
in 1965201. Over 170 cases of CPA have since been reported throughout Africa, including 1524
South Africa, Nigeria, Ivory Coast, Senegal, Central African Republic, Djibouti, Ethiopia, 1525
Tanzania and Uganda16,201–212. Over 90% of these cases occurred in persons previously 1526
treated for pulmonary tuberculosis. Cases of co-‐infection with Aspergillus and active 1527
pulmonary tuberculosis have also been documented in Tunisia and Egypt213,214. 1528
1529
An estimated 9 million people developed tuberculosis in 2013215. It was associated with 1530
1.5 million deaths14, of which only 210,000 are estimated to be due to multidrug 1531
resistant strains of M. tuberculosis. Many of these deaths in patients with initially drug-‐1532
susceptible disease will be due to lack of diagnosis, poor access to treatment or 1533
inadequate compliance, given that they mostly occur in resource-‐poor countries with 1534
weak health infrastructure. However some may be due to undiagnosed CPA. 1535
1536
CPA presents with progressive pulmonary cavitation associated with weight loss, 1537
persistent cough and haemoptysis5,7,8. This presentation is near identical to that of 1538
58
pulmonary tuberculosis itself216. Chest X-‐ray often cannot distinguish the two 1539
conditions because cavities, pleural thickening and fibrosis are characteristic of both 1540
tuberculosis and pulmonary aspergillosis5,8,217,218. Aspergilloma is distinctive, but it is 1541
present in only 36% of cases of CPA8 and probably represents a late manifestation of 1542
infection219. Raised levels of Aspergillus-‐specific IgG are present in almost all cases of 1543
CPA5,7,8,59, but this test is essentially unavailable in Africa220. 1544
1545
The UK Medical Research Council (MRC) study, performed between 1968 and 1970, 1546
remains the only published measure of the prevalence of CPA secondary to pulmonary 1547
tuberculosis. It investigated 544 patients with residual lung cavities after tuberculosis 1548
treatment76,197. Precipitating antibodies to Aspergillus fumigatus were present in 34% of 1549
patients, of whom 63% went on to develop an aspergilloma within the 2-‐year follow-‐up 1550
period. Haemoptysis occurred in 42% of those with aspergilloma. This study has not 1551
been replicated, but positive Aspergillus-‐specific antibodies have been found in 20-‐27% 1552
of patients previously treated for pulmonary tuberculosis in Japan, India and 1553
Brazil80,146,192,221. 1554
1555
The MRC study had a number of limitations. The radiological criteria used to define 1556
‘aspergilloma’ are not well described. It was performed before the invention of the CT 1557
scan and therefore probably underestimates the frequency of aspergilloma and other 1558
characteristic features of CPA. It also used precipitation-‐in-‐gel technique105,123 to detect 1559
Aspergillus-‐specific antibodies, which probably has inferior sensitivity in comparison to 1560
modern ELISA techniques74. Furthermore it was restricted to those with visible 1561
cavitation on chest X-‐ray after tuberculosis. Aspergillosis might also develop in cavities 1562
smaller than those detected on chest X-‐ray. These cases would not be detected with this 1563
study design. 1564
1565
The global prevalence of CPA secondary to tuberculosis was nonetheless calculated on 1566
the basis of the 1968-‐70 MRC study results, together with current prevalence data for 1567
pulmonary tuberculosis and residual cavitation following pulmonary tuberculosis11. 1568
The rate of post-‐tuberculous cavitation was assumed to be 22% everywhere except 1569
Europe where it was estimated to be an arbitrary 12%. As CPA may occur in those 1570
without visible cavitation, an arbitrary 2% of post-‐tuberculous patients without visible 1571
59
cavitation were also includes in the estimates. The 5 year period prevalence of CPA was 1572
estimated to be 1,173,881 million people secondary to tuberculosis11. The 5-‐year point 1573
prevalence of CPA secondary to tuberculosis in DR Congo was estimated to be 43 per 1574
100,000 population11. However any inaccuracies in the original 1968-‐70 study would be 1575
reflected in this prediction. Multiple other underlying diseases are associated with CPA 1576
in addition to tuberculosis14. The total global prevalence of CPA is estimated at around 3 1577
million cases when these are included11–13. 1578
1579
There has never been a survey measuring the prevalence of CPA secondary to 1580
tuberculosis in a current area of high tuberculosis prevalence. CPA prevalence in these 1581
areas might differ from the UK in 1968-‐70. Rates of Aspergillus rhinitis and keratitis are 1582
higher in countries with warm climates and many subsistence farmers10. The same 1583
pattern might apply to CPA. Frequent exposure to wood smoke is common in areas of 1584
high tuberculosis prevalence and is associated with increased frequency of respiratory 1585
diseases222. CPA rates might also be higher in this group. Most importantly HIV co-‐1586
infection might alter the rate of CPA, either increasing it due to 1587
immunosuppression52,223,224 or decreasing it by reducing the rate of residual cavitation 1588
after tuberculosis treatment225–227. 1589
1590
Fungal infections are a well-‐documented aspect of AIDS. Oral candidosis affects 90% of 1591
AIDS patients and more than 10% develop oesophageal candidiasis228. Cryptococcus 1592
neoformans can cause both pulmonary infection229 and meningitis230. Pnuemocystis 1593
jirovecii, Blastomyces dermititidis, Coccidioidomyces imitis and Paracoccidioidomyces 1594
brasiliensis are all known to cause pneumonia in patients with AIDS231–234. 1595
1596
Aspergillus tracheo-‐bronchitis is perhaps the most well known form of invasive 1597
aspergillosis associated with AIDS43. However, there is also a documented association 1598
between pulmonary aspergillosis and AIDS, in the absence of pulmonary 1599
tuberculosis52,223,224,235,236. Many of these cases occurred in patients with drug-‐induced 1600
neutropaenia, but in 44% of cases neutrophil counts were normal235. Advanced AIDS is, 1601
however associated with impaired neutrophil function237 and almost all of these 1602
patients had CD4 count below 100 cells/μL235. There is therefore a plausible mechanism 1603
by which AIDS could directly place the patient at risk of invasive aspergillosis. 1604
60
Most cases of pulmonary aspergillosis in AIDS are best described as subacute invasive 1605
disease, rather than CPA52,236. Indeed AIDS has been considered an exclusion criteria for 1606
CPA on the grounds that marked immunosuppression is likely to result in more rapid 1607
disease progression5. Chronic presentations have, however also been described in 1608
AIDS223. In reality there is probably a spectrum of disease related to the severity of AIDS 1609
related immunosuppression. The precise nature of the clinical syndrome is less 1610
important than the outcome. Autopsy studies from Italy, India and Uganda have 1611
demonstrated that aspergillosis is associated with 3-‐11% of all AIDS related deaths, 1612
with only 10% of these cases diagnosed during life207,238–240. 1613
1614
In resource-‐poor settings pulmonary tuberculosis is often diagnosed on the basis of 1615
World Health Organization (WHO) approved ‘smear-‐negative’ criteria, with 1616
microbiological proof of tuberculosis infection not required241. These clinical diagnosis 1617
protocols are used extensively in areas of high tuberculosis and HIV prevalence242. In 1618
Uganda 54% of HIV positive out-‐patients commencing tuberculosis therapy are 1619
diagnosed in this manner, but only 35% of patients in the ‘smear-‐negative’ group who 1620
submit sputum for culture ultimately grow mycobacterium tuberculosis243. In Ugandan 1621
HIV positive in-‐patients there is essentially no correlation between the results of the 1622
WHO smear-‐negative diagnostic protocol and eventual confirmation of tuberculosis244. 1623
1624
Comparing smear-‐negative pulmonary tuberculosis cases to smear-‐positive pulmonary 1625
tuberculosis cases, the hazard ratio is 1.49 for 2-‐month mortality in DR Congo245 and 2.2 1626
for 7-‐year mortality in Malawi246. One potential explanation for the excess mortality in 1627
the smear-‐negative group is that some cases of ‘smear-‐negative tuberculosis’ are in fact 1628
undiagnosed and untreated cases of pulmonary aspergillosis. 1629
1630
Treatment greatly reduces mortality in pulmonary aspergillosis. In invasive 1631
aspergillosis 12-‐week survival rates of 71% and 57% have been achieved with 1632
voriconazole and amphotericin, in comparison to almost 100% mortality without 1633
treatment247–249. Amphotericin is already used effectively to treat cryptococcal 1634
meningitis in resource-‐poor settings230. CPA is also treatable. Oral itraconazole has 1635
been shown to prevent clinical and radiological progression18,198 and is available in low-‐1636
cost generic preparations. Voriconazole and posaconazole have also been associated 1637
61
with positive outcomes58,108,198. Surgery is curative in selected patients with localized 1638
disease15,21 and has been safely delivered in resource-‐poor settings16,54,212. Identifying 1639
and treating patients with CPA could therefore result in a reduction of the mortality rate 1640
currently ascribed to pulmonary tuberculosis. 1641
1642
The existing evidence therefore raises a real possibility that pulmonary aspergillosis 1643
commonly occurs in association with both tuberculosis and AIDS. The situation is 1644
complicated by the fact that CPA both complicates pulmonary tuberculosis and mimics 1645
its clinical and radiological presentation14,217,218. It therefore might well be frequently 1646
misdiagnosed as recurrent ‘smear-‐negative tuberculosis’ in those with previous treated 1647
pulmonary tuberculosis215. Pulmonary aspergillosis can also occur as a direct 1648
consequence of AIDS52,223,236. In this circumstance patients are also at high risk of 1649
pulmonary tuberculosis and the presentation of the two conditions would be near 1650
identical. 1651
1652
The central goal of the work contained in this thesis is to determine the frequency of 1653
pulmonary aspergillosis in association with tuberculosis and AIDS. A prerequisite of this 1654
is to define the performance characteristics of Aspergillus-‐specific IgG assays in a well-‐1655
defined CPA population. The UK National Aspergillosis Centre has the world’s largest 1656
collection of stored sera from known CPA cases, which have been used to define the 1657
sensitivity and specificity of various assays. The best available assay is then used in 1658
surveys to measure the prevalence CPA complicating treated pulmonary tuberculosis. 1659
This study includes both HIV infected and uninfected persons to determine the impact 1660
of HIV infection on CPA prevalence. Two additional Ugandan cohorts are also assessed 1661
for evidence of primary pulmonary aspergillosis; those with proven pulmonary 1662
tuberculosis and those with AIDS and ‘smear-‐negative pulmonary tuberculosis’, but no 1663
microbiological proof of diagnosis. 1664
1665
1666
62
METHODOLOGY 1667
1668
Paper 1 – Performance of six Aspergillus-‐specific IgG assays for the diagnosis of 1669
chronic pulmonary aspergillosis (CPA) and allergic bronchopulmonary 1670
aspergillosis (ABPA) 1671
1672
Patients 1673
1674
241 patients with CPA and 80 patients with ABPA were identified at the UK National 1675
Aspergillosis Centre (NAC) who had a stored sample of serum taken at a time when they 1676
were either off treatment, or had only started treatment in the last 3 months. Samples 1677
were stored between 2004 and 2014. Diagnosis of CPA or ABPA was taken from clinical 1678
notes. 1679
1680
All patients with ABPA at our centre are routinely screened for the development of CPA. 1681
Any patient with CPA complicating ABPA was classified as CPA. The ImmunoCAP 1682
Aspergillus-‐specific IgG assay (ThermoFisher Scientific, multi-‐national) was used for 1683
clinical diagnosis at our laboratory during this period. However the NAC accepted 1684
referrals from all over the UK and those that fulfilled the microbiological diagnostic 1685
criteria with culture growth or positive result with an alternative Aspergillus antibody 1686
assay used by the referring hospital were accepted as cases regardless of their 1687
ThermoFisher Scientific ImmunoCAP result. 1688
1689
CPA diagnostic criteria 1690
1691
CPA is diagnosed at the NAC on the basis of criteria proposed by Denning et al in 20035 1692
and later endorsed by the Infectious Diseases Society of America (IDSA)250 and 1693
European Society of Clinical Microbiology and Infectious Diseases (ESCMID)251. It 1694
required the presence of all of the following; 1 -‐ underlying disease, 2 -‐ symptoms, 3 -‐ 1695
radiological changes and 4 -‐ microbiological evidence. The latter could take the form of 1696
biopsy, repeated Aspergillus culture or PCR from sputum or broncho-‐alveolar lavage 1697
fluid. However in the vast majority of cases microbiological evidence was provided by 1698
63
raised levels of Aspergillus-‐specific IgG. As the NAC accepts patents from the whole UK a 1699
variety of Aspergillus-‐IgG assays will have been used to make the original diagnosis at 1700
referring hospitals. 1701
1702
Antibody positivity with either the ThermoFisher Scientific ImmunoCAP (cut-‐off 1703
40mg/L) or two different precipitins assays (positive at any dilution) is the commonest 1704
microbiological means of making the diagnosis at the NAC, but histological evidence 1705
from biopsy or resection, positive culture from sputum or bronchoscopy samples, 1706
Aspergillus PCR on sputum and galactomannan on bronchoalveolar lavage fluid are all 1707
also accepted as ‘microbiological evidence’ of CPA. Most patients had several positive 1708
samples, at the referring hospital and at the NAC. 1709
1710
ABPA diagnostic criteria 1711
1712
ABPA is diagnosed at the NAC in line with International Society for Human and Animal 1713
Mycology (ISHAM) diagnostic criteria4. These require the presence of raised total IgE 1714
and raised Aspergillus-‐specific IgE (or positive skin prick testing) for diagnosis. Positive 1715
precipitins or Aspergillus-‐specific IgG is one of three additional features, along with 1716
raised eosinophil count and radiological features, of which two out of three are also 1717
required to confirm the diagnosis. 1718
1719
Control samples were collected from 100 healthy Ugandan students aged 16-‐18 who 1720
were donating blood and were negative for HIV. Samples were tested for Aspergillus-‐ 1721
IgG by all methods. Diseased controls for comparison to ABPA sera were taken from 100 1722
asthmatic patients under the care of the North West Lung Centre at UHSM. These 1723
samples were stored as part of the ManRAB biobank project. Ethical clearance was 1724
granted by the ManRAB committee prior to their use in this study. 1725
1726
Assays 1727
1728
Tests were performed between January and July 2014. Aspergillus-‐specific IgG levels 1729
were measured by the author on all stored samples using the Immulite 2000 (Siemens, 1730
Germany) performed at Christie Hospital, Manchester UK. Manual plate ELISAs were 1731
64
performed by the author using kits supplied by Serion (Germany), Genesis (UK,) and 1732
Dynamiker (China) at the Mycology Reference Centre, University Hospital of South 1733
Manchester, UK. 1734
1735
Each manual plate ELISA kit followed a similar process that first required serum to be 1736
diluted using a diluent supplied by the manufacturer. A fixed volume of diluted serum 1737
was then added to wells that were pre-‐coated with Aspergillus antigens, for a fixed 1738
period of time. The Serion and Dynamiker kits required this incubation to take place in 1739
an incubator at 37oC, while the Genesis kit required incubation at room temperature. 1740
After washing with a washing solution provided with the kit a fixed volume of conjugate 1741
was added and incubated for a fixed period of time. After further washing a fixed 1742
volume of substrate was added and incubated for a fixed period of time. Finally a fixed 1743
volume of stopping solution was added. 1744
1745
Optical density was measured within five minutes of test completion on a PolarStar 1746
Omega spectrophotometer (BMG Labtech, UK), with settings for each individual kits 1747
utilized in line with manufacturers’ instructions. For the Genesis and Dynamiker assays 1748
a formula was installed on the spectrophotometer software to convert optical density to 1749
arbitrary units in line with the manufacturers’ guidelines. For Serion the optical density 1750
results were entered onto an Excell database supplied by the manufacturer with 1751
embedded formulae to convert optical density to arbitrary units. Where a result was 1752
greater than a threshold specified by the manufacturer a 1 in 10 dilution was performed 1753
and the assay was repeated. 1754
1755
Results were rejected if the manufacturers’ stated quality control criteria were not met 1756
for an individual test plate. If this occurred the tests were re-‐run on a fresh plate. The 1757
exception was the Dynamiker assay where all kits failed the same one of the three 1758
stated quality controls criteria. This was discussed with the manufacturer. After 1759
reviewing our data in detail the manufacturer concluded that they would revise the 1760
quality control guidance for this kit in line with our findings. A decision was made with 1761
the manufacturer to proceed with the use of their kits in the study, as long as the other 1762
quality control criteria were met for each plate. A summary sheet comparing the exact 1763
processes for each manufacturer is shown in appendix 1. 1764
65
The ImmunoCAP Aspergillus-‐specific IgG assay is used for routine testing at the National 1765
Aspergillosis Centre. Results of tests from CPA and ABPA patients, performed as part of 1766
routine clinical care, from the same sample that was subsequently stored, are used in 1767
the analysis. 100 control samples were sent for ThermoFisher Scientific ImmunoCAP 1768
Aspergillus-‐specific IgG testing at the Manchester Royal Infirmary immunology 1769
laboratory. These tests were performed by the staff at this laboratory. 1770
1771
Precipitation in gel (precipitins) testing was also performed by the author using the 1772
counterimmunoelectrophoresis (CIE) technique74. 10ml agarose was melted and 1773
poured onto a hydrophobic gel bond film (GE Healthcare, USA). Three mm diameter test 1774
wells were cut once the gel had set. Twenty µL sera were placed in one row of wells 1775
with 20 µL antigens (Microgen, UK) at 2 mg/ml placed in the adjacent row. The gel was 1776
placed above a CIE tank filled with veronal buffer and blotting paper wicks were used to 1777
connect either end of the gel to the buffer tanks before applying 34V for 90mins. The gel 1778
was then placed in sodium chloride tri-‐sodium citrate buffer overnight before being 1779
dried with a hair dryer. After drying the gel was placed in a Coomassie Blue stain for 15 1780
minutes, followed by 2 serial de-‐stains using methanoloic acetic acid. Each de-‐stain step 1781
lasted 10 minutes. After further drying by hair dryer the gels were read on a light box 1782
with the assistance of a magnifying glass. The presence of any precipitins bands was 1783
reported as a positive result. Neat serum was tested for all samples. Where samples 1784
were positive serial dilution to a maximum 1 in 32 dilution was produced to provide 1785
dilutional titres. 1786
1787
Where a sample produced the same result (positive or negative) on a single test by all 1788
test methods this result was accepted. Where a sample produced divergent results on 1789
different assays it was repeated twice. If the 2 new tests resulted in a different outcome 1790
(positive or negative) to the first test then the mean of these 2 new tests replaced the 1791
first test. Final results after any repeat testing are reported. 1792
1793
Statistics 1794
1795
Intra-‐assay variability (IAV) was measured for all assays except ThermoFisher Scientific 1796
ImmunoCAP. Two samples were selected for each assay, one that produced a low result 1797
66
and one that produced a high result. Each assay was repeated 20 times on each sample. 1798
When results were complete the study team (the author and his two supervisors) 1799
identified outliers. These were defined as results that were markedly different to the 1800
other tests, with a likely explanation for an error. These outliers were removed from the 1801
final analysis. Result range, mean, standard deviation and co-‐efficient of variation (CV) 1802
are reported for each assay. CV was calculated as (standard deviation / mean) X 100. 1803
Statistical analyses were performed using SPSS version 20 (IBM, USA) under license to 1804
the University of Manchester, UK. Descriptive statistics are reported for each test, 1805
including the range, mean and median results in each patient group. Receiver operating 1806
characteristic (ROC) curve analysis was performed for each test. The performance of 1807
each test is compared using the Area Under the Curve (AUC) for ROC analysis with 95% 1808
confidence intervals (95% CI). Wald statistic is used to compare the ROC AUC results 1809
from different assays. ROC curve analysis is not performed for precipitins as the semi-‐1810
quantitative nature of precipitins results are not compatible with this analysis. 1811
1812
CPA and ABPA patient results were both compared directly to healthy controls for each 1813
test. ABPA results were also compared to asthmatic diseased control sera. As CPA can 1814
develop as a complication of ABPA the results of tests in these groups are compared by 1815
ROC analysis to determine the performance of each assay for the diagnosis of CPA in 1816
patients with underlying ABPA. 1817
1818
An ideal diagnostic threshold for the best performing test was selected. This threshold 1819
had a specificity of 98% and sensitivity of 96% for the diagnosis of CPA. For ease of 1820
comparison we then selected diagnostic thresholds for the other tests that also had 98% 1821
specificity. The sensitivity and specificity for each test is reported using both the 1822
manufacturer’s suggested diagnostic threshold and the new suggested thresholds. 1823
1824
The statistical analysis of all aspects of this thesis was discussed with Julie Morris, 1825
statistician based at UHSM Academy. 1826
1827
1828
67
1829 Discussion 1830 1831
Recognition of the existence of the disease CPA 1832
1833
The work contained in this thesis is based around the syndrome chronic pulmonary 1834
aspergillosis. While the existence of pulmonary aspergilloma has been documented for 1835
centuries195,196 and complex aspergilloma recognised as a complication of cavitation 1836
lung disease for decades252 the existence of CPA in the form described above was only 1837
described in 20035. Since that time there have been major cohort descriptions 1838
published in Japan7,59,60 and Korea8 and treatment studies described in Japan253, India18 1839
and France108,254,255 in addition to further work published by colleagues at the UK 1840
National Aspergillosis Centre14,21,58,198,256,257. The total number of CPA cases recorded in 1841
these papers is 763. 1842
1843
Published evidence in the field of CPA has been summarized in recent review 1844
articles37,50,258. The Infectious Diseases Society of America has published guidelines for 1845
the diagnosis and management of CPA250 and the European Society for Clinical 1846
Microbiology and Infectious Diseases is in the process of drafting its own guidelines251. 1847
The existence of CPA is therefore now widely accepted. 1848
1849
Use of composite gold diagnostic standard for CPA 1850
1851
All these publications and professional bodies accept that CPA is diagnosed using a 1852
combination of symptoms, radiological findings and microbiological evidence. There is 1853
no single gold standard test for this condition. For many fungal infections including 1854
invasive aspergillosis biopsy is the accepted gold standard48. In the case of invasive 1855
fungal aspergillosis the presence of fungal hyphae invading lung tissue is the accepted 1856
definition of ‘proven’ disease. 1857
1858
Unfortunately biopsy cannot be used as a gold standard in CPA as there is no invasion of 1859
healthy tissue in this condition. Histological examination can demonstrate the presence 1860
of aspergilloma or Aspergillus nodules in resected lung tissue21. This sometimes 1861
represents the first diagnostic evidence of CPA, particularly in patients where the 1862
68
diagnosis of lung cancer was suspected based on CT scan appearance and resection was 1863
deemed necessary. This is clearly not an ethically acceptable method for use in an 1864
epidemiological survey, as the risk of death or morbidity secondary to resection means 1865
it cannot be performed in patients with no clear medical indication for undergoing the 1866
procedure. 1867
1868
In paper one, the sensitivity and specificity of various Aspergillus-‐specific IgG assays is 1869
described. Histological examination of resected lung tissue was not used as a gold 1870
standard to compare the assays against. This was firstly on the grounds that only a 1871
handful of cases treated by the NAC were diagnosed in this way. Given that the NAC is 1872
the largest treatment centre for CPA in the world, it is unlikely that a sufficiently large 1873
cohort of CPA patients diagnosed in this manner exists anywhere. 1874
1875
Also, patients diagnosed as a result of histological examination of resected tissue may 1876
not be representative of the CPA population as a whole. Surgery is often performed on 1877
patients with clear aspergilloma and haemoptysis15. As aspergilloma is present only in a 1878
minority of patients8 and regarded as a late complication of CPA5. Conversely, 1879
Aspergillus nodules may mimic the appearance of cancer and be resected on those 1880
grounds, but while the natural history of Aspergillus nodule disease in patients without 1881
gross immunosuppression is not entirely clear, it probably represents a less severe 1882
form of CPA259. Overall the surgical population is probably not representative of the CPA 1883
population as a whole. 1884
1885
For these reasons it was clear that the best possible standard to measure the various 1886
Aspergillus-‐specific IgG against was the composite gold standard used to diagnose 1887
clinical cases of CPA at the NAC21,198,257. The diagnosis of CPA in this cohort is never 1888
based on the presence of raised Aspergillus-‐specific IgG alone. Rather, the combination 1889
of symptoms, radiological change and microbiological evidence of infection is required. 1890
As noted above this diagnostic standard is widely accepted. It is also important to note 1891
that the cohort of patients with CPA at the NAC includes patients with microbiological 1892
evidence of Aspergillus infection from various sources. Many have raised Aspergillus-‐1893
specific IgG, but patients diagnosed on the basis of antigen tests such galactomannan, 1894
69
repeatedly positive culture from respiratory samples and histological findings from 1895
resection or biopsy are also included, regardless of their Aspergillus-‐specific IgG levels. 1896
1897
While this represents the best gold standard option available, there are difficulties with 1898
this method. It is not possible to provide positive and negative predictive values for the 1899
Aspergillus-‐specific IgG assays studied here. These figures cannot be calculated from the 1900
case control study design used in paper one. They could theoretically be calculated from 1901
the cross-‐sectional survey such as the one described in paper three. Unfortunately 1902
raised levels of Aspergillus-‐specific IgG are a mandatory criterion in the composite gold 1903
standard used to define cases of CPA in this study. It is not therefore possible to 1904
calculate positive and negative predictive values, as it would be impossible to have a 1905
false negative result for Aspergillus-‐specific IgG in relation to this study design. 1906
1907
Potential for bias due to use of Aspergillus-‐specific IgG in composite diagnostic criteria 1908
for CPA 1909
1910
It is possible that the NAC cohort is biased in favour of patients with raised Aspergillus-‐1911
specific IgG. The majority of cases in the NAC CPA cohort have raised Aspergillus-‐specific 1912
IgG74, whereas culture is positive only in a minority of patients using standard 1913
techniques260. Cases of CPA occur with positive culture or biopsy, but negative 1914
Aspergillus-‐specific IgG5,7,8,59. However measuring Aspergillus-‐specific IgG is much more 1915
convenient than other potential sources of microbiological evidence. It is therefore 1916
likely that patients with raised Aspergillus-‐specific IgG are diagnosed more frequently 1917
than ‘antibody-‐negative’ cases in the course of clinical practice. As a result the 1918
sensitivity of any Aspergillus-‐specific IgG for the diagnosis of CPA may be exaggerated 1919
by this method. 1920
1921
The ThermoFisher Scientific ImmunoCAP assay is used as the standard method to 1922
measure Aspergillus-‐specific IgG at the NAC. While patients are referred from all over 1923
the country and results of other assays at other hospitals are accepted, the large 1924
majority probably had Aspergillus-‐specific IgG measured using the ThermoFisher 1925
Scientific ImmunoCAP assay. By comparison the manufacturers of the Siemens 1926
70
Immulite, Serion, Genesis and Dynamiker kits advise us that their products are not in 1927
regular clinical use anywhere in the UK. 1928
1929
While these difficulties could not be entirely negated, the composite gold standard that 1930
was used was the best option available and the resulting study does represent the most 1931
informative assessment of the sensitivity and specificity of the various available 1932
Aspergillus-‐specific IgG assays. It is also worth noting that any potential bias in favour of 1933
ThermoFisher Scientific ImmunoCAP did not prevent the Siemens Immulite system 1934
from demonstrating equivalent sensitivity and specificity to the ThermoFisher Scientific 1935
ImmunoCAP assay. 1936
1937
Use of results from fresh and stored sera for different assays 1938
1939
The study was intended to be performed solely on stored sera. The length of time the 1940
sera had been stored for varied from patient to patient, but was up to ten years. 1941
Unfortunately ThermoFisher Scientific ImmunoCAP declined to donate kits for use in 1942
the study. Grant applications for pay for the purchase of ThermoFisher Scientific 1943
ImmunoCAP kits were not successful. ThermoFisher Scientific ImmunoCAP is the most 1944
commonly used test in the UK. The study would have been much less valuable if it was 1945
excluded from the comparison. The results of ThermoFisher Scientific ImmunoCAP 1946
testing performed on fresh samples as part of routine clinical care were therefore used 1947
in the comparison, as they were the only available source of ThermoFisher Scientific 1948
ImmunoCAP results. 1949
1950
It is conceivable that this would introduce bias. There is no published data regarding the 1951
repeatability of Aspergillus-‐specific IgG testing after prolonged storage of frozen 1952
samples. There is some evidence that galactomannan antigen testing results are altered 1953
by prolonged storage, with lower results noted in stored samples261. If this were also 1954
the case for Aspergillus-‐specific IgG testing as well then the study would be significantly 1955
biased in favour of ThermoFisher Scientific ImmunoCAP. There is, however evidence 1956
that antibody levels are unaffected by prolonged serum storage262,263. There is no 1957
reason to believe that Aspergillus-‐specific IgG would behave any differently to other 1958
71
antibodies in this respect. The use of stored sera for some assays and historical results 1959
from fresh sera for other assays is therefore unlikely to have introduced bias. 1960
1961
Number of tests performed on each serum sample 1962
1963
Ideally all the assays used in this study would have been performed in triplicate in every 1964
sample to ensure that any human error in an individual test was identified and removed 1965
from the analysis. This was not possible due to lack of time and insufficient supply of 1966
kits. The study was dependent on the kind donation of kits by the manufacturers. Many 1967
manufacturers declined to donate any kits and the ones that did donate were limited in 1968
the number of kits they could provide. There was also insufficient volume of stored sera 1969
available for such a study design in many cases. 1970
1971
It was, however possible to compare the results from each individual sample across the 1972
six assays used. While the results produced by different kits could not be directly 1973
compared as they did not all use the same units, it was possible to see if the results were 1974
reproduced across the kits in terms of positivity or negativity. Where a sample 1975
produced the same result on all seven kits this was unlikely to be the result of the same 1976
human error occurring in each separate test. These results could therefore legitimately 1977
be accepted on the basis of a single test for each kit. Where different kits produced 1978
divergent results for an individual sample then triplicate testing was performed. This 1979
method should be almost as reliable as performing triplicate testing in every case. 1980
1981
Use of single operator to perform assays 1982
1983
The author performed the almost all of tests included in the serology comparison 1984
papers himself. This has the major advantage of consistency between tests, but as a 1985
result this work cannot comment on inter-‐operator variability. The study required the 1986
performance of over 5000 individual assays, meaning the author became highly 1987
experienced in the use of these assays. The author received training on the performance 1988
of these assays by laboratory technicians experienced in their use prior to commencing 1989
the study. Assay runs used in the final analysis only went ahead after establishing that 1990
consistent results were obtained from selected test sera over several training runs. 1991
72
Selection of control groups 1992
1993
The main analysis in paper one compares results of Aspergillus-‐specific IgG testing in 1994
known cases of CPA and ABPA to healthy controls. These healthy control sera are taken 1995
from Ugandan blood donors. These may not be the ideal comparator group. Ugandan 1996
control sera were selected because the primary goal of this study was to define 1997
diagnostic cut-‐offs for assays that might subsequently be used in a prevalence study in 1998
Uganda. 1999
2000
It is conceivable that Ugandans might form different levels of antibodies than Britons, 2001
perhaps due to different levels of environmental exposure to Aspergillus spp. If this was 2002
the case then the diagnostic thresholds suggested in this study might be less accurate 2003
outside Uganda. There is, however no published evidence that the levels of Aspergillus-‐2004
specific IgG in healthy controls varies from country to country. 2005
2006
The age matching of controls and cases was sub-‐optimal. Mean age was 19 years for 2007
controls, but is 59 years for CPA cases at the NAC198. Aspergillus-‐specific antibody levels 2008
rise throughout childhood and appear to stabilize at adult levels71,72. Adolescents may 2009
still have antibody levels below that of healthy adults. Despite these potential flaws the 2010
use of control samples from healthy blood donors is common practice in research as 2011
these are the most practical healthy control samples to access. No other samples were 2012
available for use in these studies. 2013
2014
In the case of ABPA two analyses were performed in the serology comparison detailed 2015
in paper one. First the results of patients with ABPA with were compared with healthy 2016
controls and then the results of patients with ABPA were compared to asthmatic 2017
diseased controls. The reason for doing so was that ABPA very rarely occurs in non-‐2018
asthmatics. One exception is ABPA occurring secondary to cystic fibrosis, but this has 2019
been investigated elsewhere.66,78 If asthmatic patients have different levels of 2020
Aspergillus-‐specific IgG to healthy persons then a diagnostic cut off defined in a healthy 2021
population might not be applicable to the asthmatic population where the test will 2022
actually be used. 2023
2024
73
To take account of this possibility sera from asthmatic diseased controls were also 2025
tested. These came from asthmatic patients enrolled into the ManRAB biobank at the 2026
North West Lung Centre, where the NAC is based. A benefit of selecting this particular 2027
group of asthmatics is that it is that screening for ABPA is probably more likely to have 2028
occurred in a unit that specializes in aspergillosis care than might be the case in other 2029
units. On the other hand patients recruited to a biobank at a regional tertiary specialist 2030
care centre are probably not representative of the asthmatic population as a whole. 2031
2032
The ideal control group would probably consist of persons with well-‐defined asthma in 2033
a primary care setting. Unfortunately such patients are harder to access. No biobank of 2034
sera from such a group was available for use in this project and no resources were 2035
available to recruit such patients. In the absence of an unbiased asthmatic control 2036
population we must consider the possibility that healthy persons might represent a 2037
better control group than severe asthmatics referred to a tertiary hospital when 2038
considering the diagnosis of ABPA in persons with mild asthma treated in the primary 2039
care system. Diagnostic thresholds for use in ABPA have therefore been calculated 2040
against both potential control groups. 2041
2042
The same difficulty was present when the diagnostic threshold for the use of 2043
Aspergillus-‐specific IgG in the diagnosis of CPA was defined. In this case Aspergillus-‐2044
specific IgG levels in CPA cases were compared to healthy controls in paper one. 2045
However CPA occurs almost exclusively in patients with underlying diseases14. 2046
Unfortunately very little published data exists describing the levels of Aspergillus-‐2047
specific IgG found in groups of patients with these underlying diseases. It was not 2048
possible to perform these assessments within the time and financial constraints of this 2049
PhD. 2050
2051
It should be noted that Aspergillus-‐specific IgG is only one aspect of the diagnostic 2052
process for CPA. It is therefore reasonable that the threshold reached by comparison of 2053
CPA cases to healthy controls be used to define ‘abnormally high levels of Aspergillus-‐2054
specific IgG’, which form a single part of this diagnostic process. Patients with an 2055
Aspergillus-‐specific IgG above this threshold should undergo further investigation, with 2056
CPA confirmed only when all diagnostic criteria are met5. 2057
74
Other limitations 2058
2059
Finally it should be noted that the serology comparison study reported here is 2060
incomplete in several respects. Further Aspergillus-‐specific IgG kits are produced by 2061
Bio-‐Rad (France), IBL (Germany), IMMY (USA) and Bordier (Switzerland). These 2062
companies did not donate kits for use in this study and no resources were available to 2063
buy them. Precipitins were performed only with Microgen (UK) Aspergillus antigens. It 2064
is possible that the precipitins technique might produce different results if antigens 2065
from another supplier were used. 2066
2067
A single operator in a single laboratory performed all assays reported here. It therefore 2068
provides no information regarding inter-‐operator or inter-‐laboratory variability, which 2069
is crucial information that must be obtained before a test could be rolled out as part of a 2070
global screening program. Good inter-‐laboratory variation results have been published 2071
elsewhere for ThermoFisher Scientific ImmunoCAP180, but no such comparisons exist 2072
for the other assays. The study also uses tests from a single batch. Batch-‐to-‐batch 2073
variation is described in the manufacture Aspergillus antigens141 and the results of this 2074
study would not be valid if it was present in the assays used in this study. 2075
2076
Nonetheless this study represents the largest comparison of different methods of 2077
measuring Aspergillus-‐specific IgG for the diagnosis of CPA and ABPA and is the first 2078
study conducted without the bias of long-‐term antifungal therapy. As such is a major 2079
contribution to the field. 2080
2081 2082
75
Paper 2 –Aspergillus-‐specific IgG levels in patients previously treated for 2083
pulmonary tuberculosis in Gulu, Uganda 2084
2085
Ethical approval 2086
2087
Ethical approval was granted by the following bodies before the study was commenced; 2088
2089
• The University of Manchester Research Ethics Committee 1 on 7th June 2012 2090
(reference 11424). 2091
• Gulu University Faculty of Medicine Institutional Review Board (IRB) on 4th July 2092
2012 (reference GU/IRC/04/07/12) 2093
• Uganda National Council for Science and Technology (UNCST) on 20th September 2094
2012 (reference HS 1253). 2095
2096
Recruitment criteria 2097
2098
Recruitment of patients took place in Gulu, Uganda from October 2012 until January 2099
2013. Criteria for recruiting patients to the study were as follows:-‐ 2100
2101
• Patients must be aged 16 years or over. 2102
• Patients must be able to give informed consent. 2103
• Patients must have completed a full course of treatment for pulmonary 2104
tuberculosis with a treatment start date of 1st January 2005 or later. 2105
2106
A diagnosis of treated pulmonary tuberculosis (TB) was considered valid if the patient 2107
had a positive sputum smear test for acid and alcohol fast bacilli (AAFB), culture growth 2108
of mycobacterium tuberculosis species (MTB) from sputum or a positive GeneXpert 2109
polymerase chain reaction (PCR) test for MTB. A diagnosis of smear negative TB was 2110
considered acceptable only if the patient reported complete resolution of symptoms at 2111
the end of treatment. 2112
2113
76
Patients had to have documentary evidence of tuberculosis. The following was accepted 2114
as documentary evidence of TB; 2115
2116
• A completion of treatment certificate from the Uganda tuberculosis treatment 2117
program. 2118
• A tuberculosis treatment record from the Uganda tuberculosis treatment 2119
program. 2120
• Medical notes from the treating hospital documenting tuberculosis treatment. 2121
• Documentation of tuberculosis treatment on a HIV treatment card. 2122
• Documentation of the patient’s tuberculosis treatment in the central record book 2123
kept at the regional or district tuberculosis centre. 2124
2125
Recruitment targets were; 2126
• 200 HIV positive patients with previously treated TB. 2127
• 200 HIV negative patients with previously treated TB. 2128
• 100 healthy controls 2129
2130
Ugandan TB treatment protocols require HIV testing to be offered to all patients at the 2131
time of TB diagnosis. Testing is routinely performed using two different rapid test kits. 2132
If they produce different results a third brand of test kit with high sensitivity is used as a 2133
tiebreaker. The brands of rapid testing kit may have varied from site to site. The 2134
following was accepted as documentary evidence of HIV co-‐infection; 2135
2136
• HIV antibody test result form. 2137
• HIV antibody test result recorded on Uganda tuberculosis treatment program 2138
treatment card. 2139
• HIV antibody test result recorded in the tuberculosis record book at the regional 2140
or district tuberculosis centre. 2141
• HIV antibody test result documented in patients medical notes. 2142
• Treatment card stating that the patient is being treated or monitored at a HIV 2143
clinic. 2144
2145
77
Where patients presented without any of the above evidence they were offered HIV 2146
testing. Those who accepted were tested by certified laboratory assistants working with 2147
the study team and were enrolled into the study. Those who declined testing were not 2148
eligible to enter the study, although in practice no patient declined testing. 2149
2150
Recruitment strategy 2151
2152
Recruitment of patients initially took place within the Infectious Diseases clinic of Gulu 2153
Regional Referral Hospital (GRRH). Sequential recruitment of any patient attending the 2154
clinic who met the above criteria was planned. Unfortunately the patients recruited 2155
from this clinic were almost all HIV positive, as HIV-‐negative patients who have 2156
completed tuberculosis therapy are routinely discharged from active follow up. In order 2157
to recruit HIV negative patients a convenience sampling method was adopted, where 2158
patients were recruited directly from their villages with the assistance of the Gulu 2159
District Health team. 2160
2161
District health centers where patients had been treated for tuberculosis were identified 2162
from the records at the District Health team headquarters in Gulu. Recruitment was 2163
restricted to centres within a 1-‐hour drive of Gulu to allow time to transport patients to 2164
and from Gulu in addition to the time needed to perform clinical assessment and chest 2165
X-‐ray. 2166
2167
A study vehicle was sent to one district health centre daily. The centre was contacted in 2168
advance by telephone and asked to mobilise the patients with previously treated 2169
tuberculosis within their community. This was done by word of mouth with the 2170
assistance of village health workers. Radio announcements were also made inviting any 2171
patient with previous tuberculosis to present to the appropriate health centre for 2172
recruitment. A staff member (one of the two persons responsible for administering the 2173
tuberculosis program in this area) from the district health team travelled to the district 2174
health centre daily. 2175
2176
Potential recruits who brought written documentation of tuberculosis treatment were 2177
admitted to the study. Those without documentation had their names checked against 2178
78
the district health office tuberculosis record book by the staff member and were 2179
recruited to the study if they were recorded as a case of treated tuberculosis. HIV status 2180
was also copied from the district health book if the patients did not bring 2181
documentation with them. A recruitment fee of around £2 per patient was paid to the 2182
district health team. 2183
2184
Patients were transported to Gulu Regional Referral Hospital HIV clinic for assessment 2185
and venepuncture and then transported to St. Mary’s Hospital, Lacor for chest X-‐ray, 2186
after which they were taken home. Patients were provided with expenses of around £2 2187
each to buy food. They were provided with written information (in Acholi or English as 2188
preferred by the patient) and signed a consent form prior to recruitment. Copies of 2189
these documents are contained in Appendices 2 and 3. If patients could not read the 2190
information on the patient information sheet was communicated to them verbally, in 2191
Acholi, by a member of the recruitment team. Where patients could not write the 2192
consent form was stamped after discussion and verbal consent. 2193
2194
Healthy controls recruitment strategy 2195
2196
Control samples were gathered with the assistance of the Gulu district blood 2197
transfusion service. Donors gave verbal consent to having some of their blood used in 2198
the study. The majority of donors were healthy adolescents attending boarding school. 2199
Venepuncture was performed by blood transfusion service phlebotomists. Blood was 2200
stored in cool boxes and fridges for up to 24 hours before serum was separated and 2201
stored at minus 80 degrees then shipped alongside the patient samples. Details of 2202
donors’ age, gender and results of testing for HIV, syphilis and hepatitis B and C were 2203
provided by the blood transfusion service. Samples were anonymised by the blood 2204
transfusion service before being transferred to the study group. Test results were not 2205
fed back to control sample donors as the meaning of an unexpected positive test in an 2206
asymptomatic control patient is not clear. 2207
2208
2209
79
Clinical assessment process 2210 2211
During clinical assessment demographic data was recorded including patient name, age, 2212
gender, village, and telephone number together with the name and telephone number of 2213
a nominated guardian. Details of previous TB treatment were taken including; date 2214
when treatment was commenced, smear status at diagnosis, culture results, GeneXpert 2215
PCR results and whether the patients symptoms resolved completely with treatment. 2216
The TB reference number was noted. HIV status was recorded together with date of 2217
diagnosis. Where available, baseline and current CD4 count were recorded. 2218
2219
The presence and duration of symptoms including; cough, haemoptysis, fatigue, 2220
breathlessness, fevers, night sweats, and chest pain were recorded. MRC Dyspnoea scale 2221
score was recorded. The MRC dyspneoa scale is described fully in Appendix 4. Patients 2222
were provided with an Acholi translation of the scale if they could not read English. 2223
Ability to work normally or not was recorded. Lung percussion and auscultation were 2224
performed and the results recorded together with oxygen saturations. 2225
2226
Potential CPA risk factors were recorded, including living in a traditional ‘grass-‐thatch’ 2227
house in comparison to a modern brick dwelling with metal or slate roof. Other factors 2228
recorded were the reported presence of visible dampness in the patient’s house, regular 2229
exposure to wood smoke though cooking, farming or handling farm products and 2230
cigarette smoking. 2231
2232
Aspergillus-‐specific IgG testing process 2233
2234
Venepuncture was performed by trained laboratory technicians. Samples were stored in 2235
a cool box for up to 4 hours. At the end of the recruitment period they were transported 2236
to Joint Clinical Research Centre (JCRC) laboratory. Here serum was separated by 2237
trained and accredited JCRC laboratory staff. Serum was placed in labeled cryotubes and 2238
frozen in a minus 80 freezer with both diesel generator and battery back up in the event 2239
of mains power failure. The freezer had a temperature alarm system and the building 2240
was occupied at all times to ensure power failures were corrected without sample 2241
thawing. 2242
80
Samples were shipped on dry ice to the Manchester University laboratory at University 2243
Hospital of South Manchester (UHSM) at the end of the study with the assistance of DHL 2244
couriers. Shipping authorization was provided in the form of a Material Transfer 2245
Agreement (MTA) between UHSM and JCRC, plus specific written authorization from 2246
the Gulu University IRB and UNCST. 2247
2248
CD4 count testing process 2249
2250
Where HIV positive patients had a recorded CD4 count result from within the last 12 2251
months this was accepted as a current CD4 count. Where no such result was available 2252
CD4 count was performed at JCRC laboratory, Gulu. 2253
2254
Radiology process 2255
2256
All chest X-‐rays were performed at Lacor hospital radiology department by a qualified 2257
radiographer. Patients were transferred to and from Lacor hospital for chest X-‐ray after 2258
completing the clinical assessment and venipuncture. 2259
2260
Radiology reporting strategy 2261
2262
Chest X-‐ray results were reported by two radiologists. Dr Cyprian Opira is the senior 2263
radiologist at St. Mary’s Hospital, Lacor. He viewed plain films, visualized on a light box 2264
in a darkened X-‐ray reporting room with curtains. Dr Sharath Hosmane is a specialty 2265
registrar in radiology, working at UHSM. Films were photographed at St. Mary’s 2266
Hospital using a Nikon digital camera. The images were sent to Dr Hosmane by email. 2267
Where the two radiologists produced discordant reports, a decisive third report was 2268
provided by Dr Richard Sawyer, senior consultant respiratory radiologist at the North 2269
West Lung Centre, UHSM. 2270
2271
2272
81
Discussion 2273 2274
Limited nature of existing prevalence data for CPA 2275
2276
This work takes place in a neglected field. The only existing published survey 2277
measuring the prevalence of pulmonary aspergillosis in patients with treated 2278
pulmonary tuberculosis is from 1968-‐7076,197. This study took place in the UK, where 2279
pulmonary tuberculosis is now much less common than it was at that time215. No study 2280
measuring CPA prevalence has been performed in an area with currently high 2281
pulmonary tuberculosis prevalence. While the limited extent of prior research in this 2282
field results in opportunities to perform landmark original research, it also presents 2283
many challenges, which require discussion and explanation. 2284
2285
Cross-‐sectional survey design 2286
2287
Paper 2 describes a cross-‐sectional survey of the prevalence of CPA in patients with 2288
treated pulmonary tuberculosis in Gulu, Uganda. This study design was the only option 2289
available that could be completed within the three-‐year timeframe of the PhD. The total 2290
budget for this study was £50,000, which is limited and resulted in a number of 2291
constraints. It is, however the first and only survey of the prevalence of CPA in an area 2292
of current high tuberculosis prevalence. The primary goal of the study is simply to 2293
confirm that CPA exists at a measurable level in this population. This goal was achieved. 2294
As this study is unique it is appropriate to extract as much information as possible, but 2295
the interpretation of the data must take account of the restraints imposed by the study 2296
design. 2297
2298
Use of convenience sampling 2299
2300
The study used convenience sampling, with patients recruited via the village health 2301
worker network and radio announcements. This method may have introduced bias. 2302
Healthier patients may have been more likely to be recruited, as sick patients may have 2303
been unable to make the journey to the recruitment sites. Conversely motivation to 2304
enter the study might have been higher in symptomatic patients. 2305
82
Most importantly this study design takes no account of patients who might have died 2306
after completing tuberculosis treatment without entering the study. The study accepted 2307
anyone who completed tuberculosis treatment within the last seven years, a duration 2308
selected on practical grounds as this was the period after the war with the Lord’s 2309
Resistance Army ended and for which good records were available. 2310
2311
The five year mortality of CPA is up to 85%7. Many patients with tuberculosis treated in 2312
Gulu within the last seven years may therefore have developed CPA and died of it 2313
without ever entering the study. This might be especially true in HIV positive patients 2314
where pulmonary aspergillosis occurs in a subacute manner and so progresses to death 2315
quicker than CPA52,223,236. 2316
2317
This cross-‐sectional recruitment method was the only practical option as patients with 2318
tuberculosis are discharged after completing treatment in Uganda. Equal opportunity to 2319
enroll was maximized by comprehensively contacting all health centres in the study 2320
zone and using radio announcements broadcast to the whole region. Sending the study 2321
vehicle to every health centre in the study zone minimized the barrier to recruitment 2322
that patients might experience due to transport difficulties. It was not, however possible 2323
to send a vehicle to every village as many were not accessible by road. Where patients 2324
travelled to the health centre by motorcycle taxi we re-‐imbursed their fares, but 2325
inability to pay up front may have limited the ability of some poorer patients in remote 2326
villages to join the study. 2327
2328
The study design allows us to state that CPA is present in this population, but the 2329
frequency of disease obtained must be interpreted in light of the cross-‐sectional design 2330
and convenience sampling. The ideal design would be a prospective study that recruited 2331
all patients completing treatment for pulmonary tuberculosis and then followed them 2332
up to see if they develop CPA. Such a study was not possible within the time and 2333
financial constraints of this PhD. The author is, however part of a team that has 2334
designed such a study in collaboration with the Kenya Medical Research Institute 2335
(KEMRI). Grant proposals have been submitted to the Japanese government to fund this 2336
study. Preliminary results from the work described in this thesis form a key part of that 2337
application. 2338
83
Paper 3 -‐ Prevalence of chronic pulmonary aspergillosis (CPA) secondary to 2339
tuberculosis: a cross-‐sectional survey in an area of high tuberculosis prevalence 2340
2341
Ethical approval 2342
2343
Ethical approval for this re-‐survey was granted by the University of Manchester ethics 2344
board, Gulu University IRB and UNCST. In each case the approval took the form of an 2345
extension and amendment to the approvals granted for paper two. New patients were 2346
not eligible. 2347
2348
CCPA diagnostic criteria 2349
2350
In this resurvey CCPA is diagnosed in patients where all of the following are present; 1 – 2351
cough or haemoptysis for one month or longer, 2 – either new or progressive cavitation 2352
on serial chest X-‐ray OR the presence of paracavitary fibrosis or aspergilloma on CT 2353
scan, 3 – raised Siemens Immulite Aspergillus-‐specific IgG, 4 – Absence of evidence of 2354
current pulmonary tuberculosis. Chronic fibrosing pulmonary aspergillosis (CFPA) is 2355
additionally diagnosed in patients who meet the diagnostic criteria for CCPA and also 2356
have extensive lung fibrosis that progressed between the 2 chest X-‐rays. Simple 2357
aspergilloma is diagnosed in patients with aspergilloma on CT scan and raised 2358
Aspergillus-‐specific IgG, but with no chronic cough or haemoptysis. Unspecified fungal 2359
ball is diagnosed in patients with apparent aspergilloma on CT scan, but with normal 2360
Aspergillus-‐specific IgG levels. 2361
2362
The frequency of raised Aspergillus–specific IgG in the absence of other signs of 2363
pulmonary aspergillosis is also reported and the frequency of symptoms and 2364
radiological changes compared between this group and those with normal Aspergillus-‐2365
specific IgG. 2366
2367
Recruitment criteria 2368
2369
Only patients recruited to the first study in 2012 were eligible for recruitment to the re-‐2370
survey. 2371
84
Recruitment strategy 2372
2373
Mobile phone numbers were taken during the first study from all patients who had 2374
access to one. Each patient was telephoned by a member of the study team. Around two 2375
thirds of patients were reached by telephone, but the remainder either had no phone 2376
number available or did not respond when called. Radio messages were used to contact 2377
these outstanding patients. In addition the study team contacted local health workers in 2378
all areas where recruitment to the original study took place. Patients were then traced 2379
by personal visits from village health workers. Recruitment days were scheduled at 2380
each of the health centres that took part on the original survey. Patients were informed 2381
of the results from the first study. 2382
2383
Each patient was provided with expenses of around £2 to cover food costs. Any 2384
transport costs incurred by the patient to reach the study centre were re-‐imbursed. 2385
2386
Clinical assessment process 2387
2388
An identical clinical assessment to the first study was performed. 2389
2390
Aspergillus-‐specific IgG testing process 2391
2392
Venepuncture was performed by the same study assistants employed in the original 2393
study using the same methods. For this study, however serum samples were separated, 2394
labeled and stored at GRRH laboratory. The two study assistants, who are both trained 2395
laboratory technicians, performed sample processing. The laboratory was not 2396
considered suitable for storage at the time of the original survey due to inadequate back 2397
up power, but new battery and diesel generator back up was installed between the two 2398
surveys. 2399
2400
Serum was shipped to the University of Manchester laboratory at UHSM after the 2401
signing of an MTA by GRRH and UHSM and the provision of written clearance to ship by 2402
Gulu IRB and UNCST. These samples will be tested for Aspergillus-‐specific IgG on the 2403
Siemens Immulite 2000 machine at Christie Hospital, Manchester. However, these 2404
85
results are not available at the time of thesis submission. Results of Aspergillus-‐specific 2405
IgG testing from the first survey are therefore used in the analysis contained in this 2406
thesis. 2407
2408
Sputum GeneXpert tuberculosis PCR testing process 2409
2410
Where patients had a productive cough and were able to produce a sputum sample this 2411
sample underwent GeneXpert PCR testing for Mycobacterium tuberculosis (Cepheid, 2412
USA) at the GRRH lab. When the GRRH GeneXpert IV machine malfunctioned samples 2413
were taken to St. Mary’s Hospital, Lacor for testing on the GeneXpert IV machine at that 2414
laboratory. These were the only two GeneXpert machines in the region. On one occasion 2415
when both were malfunctioning sputum smear testing was performed at GRRH 2416
laboratory in place of PCR testing. All samples were tested within 72 hours of 2417
submission. 2418
2419
Radiology processes 2420
2421
When clinical assessment was complete patients were then transferred to St. Mary’s 2422
Hospital for chest X-‐ray. Trained radiographers performed the X-‐rays using the same 2423
equipment as the first study. 2424
2425
Patients who had either raised Aspergillus-‐specific IgG from samples taken during the 2426
first survey or suspected aspergilloma on chest X-‐ray from the first survey were invited 2427
to undergo CT thorax. A new patient information sheet was given to patients eligible for 2428
this test and is shown in appendix 5. Consent was then taken from those undergoing CT 2429
scan using the consent form shown in appendix 6. As in the first study these forms were 2430
made available to the patients in English or Acholi. Where patients were illiterate the 2431
contents of the sheets were read and explained to the patients by Acholi-‐speaking study 2432
staff. Where patients could not sign the form verbal consent was taken. 2433
2434
Patients who consented to CT scan were transported to Kampala and underwent CT 2435
scan at the Kampala Imaging Centre. This took place at the end of the study. A qualified 2436
radiographer employed by Kampala Imaging Centre performed the scan. Patients were 2437
86
transported by private bus in groups of 30. The journey to Kampala took 10 hours each 2438
way. Hotel accommodation was provided for patients for one night in Kampala. They 2439
also received expenses of around £10, plus re-‐imbursement of any transport costs they 2440
incurred to reach the pick-‐up point. 2441
2442
Radiological reporting strategy 2443
2444
Chest X-‐rays from 2014 will be compared to chest X-‐rays from the original 2012 survey. 2445
Progressive cavitation will be noted if present. As with the first survey all X-‐rays will be 2446
reported by two radiologists, with a deciding report produced by a third radiologist in 2447
the event of divergent reports. Radiologist reports comparing X-‐rays from the two 2448
surveys are still ongoing at the time of thesis submission. These final reports will be 2449
used in the eventual publication. The author’s reports of new or progressive cavitation 2450
are used in data analysis for this thesis. 2451
2452
CT scans will be reported by three radiologists in the same manner as the chest X-‐rays. 2453
Drs Hosmane and Sawyer will both take part, in addition to Dr Rosemary Byanyima, the 2454
senior consultant radiologist at Kampala Imaging Centre. However these reports will 2455
not be completed during the PhD period. The author’s reports are therefore used in the 2456
analysis contained in this thesis. 2457
2458
Full digital records of each CT scan are available to the author and all radiologists and 2459
were accessed using OsiriX software (Pixmeo SARL, Switzerland). 2460
2461
Statistical analysis 2462
2463
A CPA prevalence of 6% was predicted prior to the study based on published estimates 2464
of CPA prevalence after pulmonary tuberculosis11. Power calculations prior to 2465
recruitment concluded that by recruiting 400 patients the prevalence of CPA following 2466
successful tuberculosis treatment could be determined with an accuracy of ± 2.3%. 2467
2468
The frequency of symptoms, test results and diagnoses is described for all patients 2469
recruited to the resurvey. Statistical analysis is performed using SPSS v20 (IBM, USA). 2470
87
Before commencing the main analysis the characteristics of the population from the 2471
first study were compared to the population in the resurvey. The frequency of 2472
categorical variables is compared by chi-‐squared test. The frequency of continuous 2473
variables with normal distribution is compared with 2-‐sided t-‐test and the frequency of 2474
continuous variables with skewed distribution is compared by Mann Whitney U 2475
analysis. Where there is no significant difference in the characteristics of the two 2476
surveys it is concluded that the recruitment process for the resurvey did not introduce 2477
significant bias. 2478
2479
Descriptive statistics are provided for all test results. For continuous variables, mean 2480
results are reported where the results had a near-‐normal distribution and are 2481
compared in different groups using a 2-‐sided t-‐test. Where results had a skewed 2482
distribution median results are reported and the results compared using the Mann 2483
Whitney U test. 2484
2485
Rates of symptoms and diagnoses in various patient groups are compared using chi-‐2486
squared test, except for comparisons with less than 5 patients in one group, where 2487
Fisher’s exact test was used. Comparison of means in different groups was performed 2488
using 2-‐sided t-‐test. 95% confidence intervals for the prevalence of CPA and other 2489
conditions were calculated as the frequency +/-‐ (standard error of the percentage X 2490
1.96). 2491
2492
The frequency of CCPA in patients with and without various potential categorical risk 2493
factors is noted and compared using chi-‐squared test. Potential risk factors analyzed 2494
include; gender, HIV co-‐infection status, sputum smear status at time of original 2495
pulmonary tuberculosis, patient profession (subsistence farmer vs. paid employment), 2496
dwelling type (traditional grass-‐thatch dwelling vs. modern home), the presence of 2497
visible dampness inside the patients home (as reported by the patient), cigarette 2498
smoking status and frequent biomass smoke exposure status. The mean age, 2499
tuberculosis treatment date and CD4 count (in HIV positive patients) in CPA patients 2500
against other patients is compared by 2-‐sided t-‐test. The number of deaths in patients 2501
with and without CPA in the original survey is noted. 2502
2503
88
Discussion 2504
2505
Recruitment strategy issues 2506
2507
The resurvey attempted to recruit all patients recruited to the first study. 2508
Comprehensive efforts were made to contact all patients including personal telephone 2509
calls, telephone calls to patients relatives or guardians, radio announcements advising 2510
patients to self-‐present for review and direct personal contact from village health 2511
workers to request patients attend. 282 of 400 patients were successfully recruited. A 2512
further 18 died, 9 moved out of the region and 11 were contacted but declined to 2513
participate in the resurvey. 77 of 400 (19%) patients could not be contacted. This 2514
probably represents a reasonable contact rate for a resurvey that took place in a post-‐2515
war region with variable mobile phone coverage where most patients lived in villages 2516
with no electricity. 2517
2518
Contacting patients was hampered by the fact that Gulu was still in a post-‐conflict 2519
situation at the time of the first survey. While most refugee camp residents had 2520
returned to their villages prior to the start of the first survey, some patients recruited to 2521
the first survey may still have been in temporary accommodation. The rate of house 2522
move was probably higher between the two surveys than might have occurred in other 2523
circumstances. Ugandans also frequently change their telephone provider and number 2524
to take advantage of lower call rates, which may have hampered attempts to contact 2525
patients by telephone. Phones may also been turned off due to lack of reliable mains 2526
electricity for charging. 2527
2528
Recruitment to the resurvey may have introduced bias if it was unintentionally selective 2529
in nature. The resurvey team was aware of the results from the first survey. There was a 2530
risk that they may have made a greater effort to contact those with raised Aspergillus-‐2531
specific IgG. This should have been avoided, as the tracing plan was the same for all 2532
patients and the recruitment fee paid to District Health Team staff was the same for all 2533
patients. There was therefore no motive for them to concentrate on patients with 2534
positive results in the first survey. If bias was unintentionally introduced then this 2535
89
would have been expected to result in a difference in the patient characteristics 2536
between the two cohorts. No such difference was detected. 2537
2538
CCPA case definition 2539
2540
The case definition of CCPA was first produced by the author’s main supervisor in 20035 2541
and has since been accepted in several publications7,8,18,21,58,108,198,257. This definition 2542
was designed for use in trials relating to CPA in well-‐resourced health care settings such 2543
as the UK. It was necessary to adapt it in several ways for use in this study. 2544
2545
Inclusion of HIV positive patients 2546
2547
The original definition explicitly excludes patients with HIV infection as potential cases 2548
of CPA5, on the grounds that gross HIV induced immunosuppression would put the 2549
patient at risk of invasive aspergillosis. This exclusion is important for the conduct of 2550
clinical trials in well-‐resourced settings, where an unbalanced inclusion of HIV infected 2551
persons could bias the results. It is not appropriate for use in Uganda, where a large 2552
proportion of patients with treated tuberculosis are co-‐infected with HIV. Gross 2553
immunosuppression is not present in the great majority of HIV positive persons 2554
enrolled in this study, who had controlled HIV on effective therapy with good CD4 2555
counts. Indeed more recent CPA case definitions have only excluded patients with 2556
uncontrolled HIV infection91. 2557
2558
As HIV co-‐infection is so common in African tuberculosis patients it was appropriate 2559
and necessary to include patients with HIV co-‐infection in this study in order to 2560
maintain a sample that was representative of the overall population at risk. Including 2561
patients with HIV co-‐infection allows the first measurement of CPA prevalence in a HIV 2562
co-‐infected population with treated tuberculosis and allows comparison of the 2563
serological and radiological findings of CPA in those with and without HIV co-‐infection. 2564
2565
To take account of HIV co-‐infection the duration of symptoms required to define a case 2566
was reduced from 3 months to 1 month. This is in line with the case definition of CNPA / 2567
subacute invasive aspergillosis proposed by other authors6,264. This definition captures 2568
90
all cases of CPA, but would also capture cases of subacute invasive pulmonary 2569
aspergillosis that might be seen in patients with AIDS52. As this study is the first to be 2570
conducted in an area of high HIV and tuberculosis prevalence it was more important to 2571
measure the overall prevalence of pulmonary aspergillosis in this population than it 2572
was to accurately subdivide these cases into chronic and subacute pulmonary 2573
aspergillosis. 2574
2575
Definition of cough in clinical assessment 2576
2577
The original case definition of CPA5 includes productive cough as a case defining 2578
symptom. This was included in the original case definition for this survey. During the 2579
course of recruitment, however it became clear that many persons reporting productive 2580
cough did not produce sputum samples and many persons reporting non-‐productive 2581
cough did produce samples. In order to minimize the impact of language barrier on 2582
clinical assessment written patient information documents and questionnaires were 2583
produced in Acholi as well as English and distributed to all patients. However, most 2584
patients recruited to the study were illiterate and spoke no English. 2585
2586
Histories were taken through a translator as a result. This translator was a qualified 2587
laboratory assistant who spoke excellent English and native Acholi, but was not a 2588
clinician. No clinicians were available to assist with the study. It seemed likely that 2589
while the concept of cough was being translated well the differentiation into productive 2590
and non-‐productive was perhaps not being well communicated to the patients. Any 2591
patient who reported cough was therefore accepted as meeting the symptoms 2592
requirement for the case definition. 2593
2594
Absence of weight loss from clinical survey 2595
2596
The original case definition5 includes weight loss alongside cough and haemoptysis as a 2597
defining feature of CPA. It was not possible to include this in the first survey, which 2598
involved only a single assessment. Records of previous weight were not always 2599
available and where they were it was by no means certain that the scales used at local 2600
health centres were appropriately calibrated before use. It would have been possible for 2601
91
us to include weight loss in the second survey, by measuring weight during both 2602
surveys. Unfortunately recruitment of patients to the first survey took place before it 2603
was clear that a second survey was possible. As a result weight measurement was not 2604
included in that study design. 2605
2606
Absence of inflammatory markers from case definition 2607
2608
The original 2003 case definition5 includes the presence of raised inflammatory 2609
markers. Some later cohort descriptions do not include this requirement108,264. Their 2610
inclusion in the case definition of CPA is therefore a matter of debate. In practice it was 2611
not possible to perform inflammatory markers as these tests were not available at Gulu 2612
Regional Referral Hospital and no funds were available to purchase them at other 2613
locations. 2614
2615
Definition of radiological features of CPA 2616
2617
Our case definition included aspergilloma as an accepted radiological feature of CPA. 2618
The original 2003 case definition5 did not include this. The definition in this study 2619
reflects the new CPA case definition that has been drafted by ESCMID251. 2620
2621
Another difficulty with the radiological definition of CPA is that while aspergilloma is a 2622
well-‐defined radiological finding218, paracavitary fibrosis is more subjective. 2623
Progressive cavitation is also subjective, as the degree of progression necessary to meet 2624
this criterion is not defined. These flaws are inherent in the accepted case definition5. A 2625
more precise and objective definition of the radiological features of CPA has not yet 2626
been proposed in the literature. 2627
2628
The impact of subjective reporting was minimized in this study by including the reports 2629
of multiple, independent radiologists who were blinded to clinical and serological 2630
information. This is a commonly used method in radiological surveys244,265,266 where a 2631
degree of human interpretation is often required. Unfortunately radiologist reports for 2632
the resurvey are not available at the time of submitting this thesis. The radiologists 2633
involved in this study all have demanding full time clinical jobs and took two years to 2634
92
complete the reports of the first survey. Delaying the submission of the thesis to await 2635
these reports was not, therefore a good option. The resurvey results in this thesis are 2636
based on the author’s own reports. Reporting by multiple radiologists will be completed 2637
and the survey results re-‐analysed in the light of these reports prior to publication. 2638
2639
Exclusion of other conditions 2640
2641
The original 2003 CPA case definition5 includes the need for exclusion of other 2642
conditions before a diagnosis of CPA can be reached. Active pulmonary tuberculosis was 2643
excluded by performing GeneXpert PCR testing on sputum. However other conditions 2644
might also mimic CPA. Non-‐tuberculous mycobacteria are commonly present as co-‐2645
infections in patients with CPA14, but would not be detected by the GeneXpert system, 2646
however their presence does not exclude a diagnosis of CPA. The new ESCMID 2647
guidelines define some infections as acceptable co-‐infections, whereas others, such as 2648
coccidioidomycosis and histoplasmosis are differential diagnoses251. 2649
2650
Other complications of tuberculosis, such as bronchiectasis, could produce similar 2651
symptoms to those of CPA. Raised levels of Aspergillus-‐specific IgG are seen in cystic 2652
fibrosis66,78 and bronchiectasis from other sources, in the absence of CPA and may be 2653
due to Aspergillus bronchitis in these settings39. Indeed this might be the explanation for 2654
the patients identified in the study with raised Aspergillus-‐specific IgG, but no CPA. The 2655
inclusion of CT scan in all patients with raised Aspergillus-‐specific IgG allows the 2656
detection of aspergilloma or paracavitary fibrosis and so differentiates CPA from other 2657
causes of raised Aspergillus-‐specific IgG. 2658
2659
Use of an Aspergillus fumigatus specific serological technique 2660
2661
Perhaps the most relevant source of diagnostic uncertainty comes from the possibility 2662
of infection with fungi other than Aspergillus fumigatus. Other species of Aspergillus 2663
might be prevalent in Africa. A. flavus is a more common cause of human disease than A. 2664
fumigatus in India146,267 and A. niger is common in Brazil147. There is very little 2665
published information relating to the frequency of infection by various species of 2666
Aspergillus in Africa. The sole study of the frequency of fungal co-‐infection in African 2667
93
tuberculosis patients was performed in Egypt and showed two cases of A. niger and 2668
three cases of Histoplasma capsulatum213. Histoplasmosis is known to exist in Uganda268 2669
and blastomycosis elsewhere in Africa232. The frequency of these other fungal infections 2670
that might mimic the clinical and radiological presentation of CPA in Africa is not 2671
known. 2672
2673
Cross reactivity between other Aspergillus-‐specific IgG assays and Penicillium antibodies 2674
has been noted269. Cross-‐reaction with other fungi has not been studied. If cross-‐2675
reaction did occur then infections by other fungi might be misdiagnosed as CPA in this 2676
study. It is also not known whether the Siemens Immulite Aspergillus-‐specific IgG assay 2677
detects non-‐fumigatus species with the same accuracy as A. fumigatus. The sensitivity of 2678
other assays based on A. fumigatus antigens for the detection of antibodies to other 2679
Aspergillus species has been poor148. The sensitivity of the Siemens Immulite assay 2680
might therefore be poor if non-‐fumigatus species are common in Uganda and 2681
responsible for a significant proportion of CPA cases. This might explain the five cases in 2682
the study with apparent aspergilloma on CT scan, but negative serology. Sensitivity of 2683
Aspergillus-‐specific IgG assays can be improved by including extracts from multiple 2684
Aspergillus species in the antigen mixture270 2685
2686
There is therefore some legitimate doubt as to whether the study can measure the 2687
precise prevalence of CPA, as opposed to other chronic fungal lung diseases. Future 2688
studies, involving extended fungal and mycobacterial culture of sputum and broncho-‐2689
alveolar lavage fluid, together with more extensive fungal serological testing will be 2690
needed to determine if fungi other than A. fumigatus are common in this patient group. 2691
2692
This does not diminish the value of the study however, as it is likely that all the patients 2693
classified as CPA in this study have chronic fungal lung disease. This is most likely due 2694
to Aspergillus infection, as Siemens Immulite Aspergillus-‐specific IgG assay is based on A. 2695
fumigatus antigens. As there has never been a survey of fungal lung disease 2696
complicating pulmonary tuberculosis in Africa this study this study represents a 2697
significant contribution to the field, even if it cannot be stated with absolute certainty 2698
that the cases detected here are CPA, as opposed to other forms of chronic fungal lung 2699
disease. 2700
94
Issues relating to HIV infection 2701
2702
While the primary goal of the study was to measure the prevalence of CPA in all 2703
patients, an attempt was also made to compare the prevalence of CPA in patients with 2704
and without HIV infection. Ideally such a design would have included HIV testing on all 2705
patients at the time of admission to the study. Unfortunately funds were not available 2706
for this. The only source of information regarding HIV infection status was the patient 2707
notes. The vast majority of patients recruited to the study were tested for HIV at the 2708
time of entering the tuberculosis treatment program in line with national guidelines. 2709
These guidelines include the use of both screening and confirmatory point of care tests 2710
that should produce accurate results271. This does not exclude the possibility that 2711
patients might have contracted HIV infection between commencement of tuberculosis 2712
treatment and recruitment to the study and some patients classified as HIV negative 2713
were indeed on HIV treatment at the time of the resurvey. 2714
2715
While the study design used was unavoidable due to funding restrictions it could be 2716
argued that HIV status at the time of active tuberculosis infection is the more 2717
appropriate risk factor to measure as HIV infection at that time is associated with 2718
reduced residual lung cavitation225, which has itself been suggested as a determinant of 2719
the rate of CPA11. 2720
2721
It was also necessary to include historical CD4 counts in the study analysis due to lack of 2722
funds to perform new CD4 counts on all patients. It is plausible that the historical CD4 2723
count may no longer be representative in some patients. However given that almost all 2724
the HIV infected patients admitted to our study knew their status and were enrolled in 2725
treatment programs prior to recruitement it is more likely that their CD4 count will 2726
have risen with time rather than fallen272. The key assumption that the median CD4 2727
count in this cohort is high and that few patients had AIDS is likely to be valid. 2728
2729
The Aspergillus-‐specific IgG assay might be less effective in HIV infected patients, where 2730
antibody response might be impaired273,274. It should be noted, however that the 2731
median CD4 count in our patients was 424 cells/µL and that acceptable antibody 2732
response to vaccination has been recorded in patients with similar CD4 counts275. It 2733
95
therefore seems likely that most patients in the study group would produce an antibody 2734
response to Aspergillus infection as most have normal or near normal CD4 counts. 2735
2736
2737 2738
96
Paper 4 -‐ “Frequency of pulmonary aspergillosis in ‘smear-‐negative tuberculosis 2739
cases” and Paper 5 “Frequency of Aspergillus co-‐infection in patients admitted to 2740
a Ugandan hospital with pulmonary tuberculosis” 2741
2742
Ethical approval 2743
2744
The Mulago study team agreed to provide stored sera to the author for use in this study. 2745
An ethical amendment from Makerere University IRB was acquired and permission to 2746
ship samples to the UK granted by Makerere University IRB and UNCST. These sera 2747
were transported to the University of Manchester Mycology Reference Centre at 2748
University Hospital of South Manchester for further testing. 2749
2750
Recruitment Strategy 2751
2752
These two papers report the results of retrospective opportunistic testing of stored sera 2753
from a prior study undertaken by collaborators. 2754
2755
The Mulago Inpatient Noninvasive Diagnosis – International HIV Opportunistic 2756
Pneumonia (MIND-‐IHOP) study was undertaken by the respiratory research group at 2757
Makerere University, Kampala, Uganda in association with the University of California, 2758
San Francisco, USA244,266,276–278. The author had no involvement in the design or conduct 2759
of the original study. The full MIND-‐IHOP study protocol is attached in appendix 7. 2760
2761
MIND-‐IHOP was a prospective cohort study. Recruitment of patients included in this 2762
thesis took place between March 2010 and March 2011. During this period all adults 2763
admitted to the casualty department of Mulago Hospital, Kampala on weekdays, with a 2764
cough of between 2 weeks and 6 months duration were offered the chance to be 2765
admitted to the study. 2766
2767
2768
97
Sample selection criteria 2769 2770
Stored sera were available from around three quarters of patients originally recruited 2771
to the study. Sera were selected from patients meeting the following criteria; 1 – 2772
Patients with HIV infection, abnormal chest X-‐ray and no diagnosis, including no 2773
evidence of pulmonary tuberculosis, following the investigations detailed above, 2 – 2774
Patients with pulmonary tuberculosis proven by culture, GeneXpert PCR or smear 2775
testing. These two groups are the subjects of reports in the two separate papers 2776
2777
Clinical assessment and diagnostic testing processes during MIND study 2778
2779
Clinical details were recorded for each patient. Sputum samples were taken for 2780
tuberculosis AAFB smear testing, culture and tuberculosis PCR testing with the 2781
GeneXpert automated nucleic acid amplification assay. Where patients could not 2782
produce sputum spontaneously induced sputum was acquired. Blood was taken for CD4 2783
count (in HIV infected patients) and cryptococcal antigen testing. Bronchoscopy was 2784
offered to any HIV positive patient with persistent symptoms and negative sputum 2785
smear test. Bronchoalveolar lavage specimens underwent staining and culture for 2786
mycobacteria and fungi including staining for Pneumocystis jirovecii. Patients were 2787
reviewed at two months after recruitment and the mortality rate at this point was 2788
recorded. 2789
2790
Aspergillus-‐specific IgG testing process 2791
2792
Selected sera were shipped to Manchester University, UK on dry ice. Aspergillus-‐specific 2793
IgG levels were then measured in each sample using the Siemens Immulite 2000 assay, 2794
performed by the author at the laboratory at Christie Hospital, Manchester. 2795
2796
Statistical analysis 2797
2798
Results of these assays are reported in comparison to healthy Ugandan blood donor 2799
controls. This is the same control group described in papers 1 and 2. The mortality rate 2800
of patients with and without raised Aspergillus-‐specific IgG levels is described. Median 2801
98
Aspergillus-‐specific IgG levels in patients and controls and CD4 counts in those with and 2802
without raised Aspergillus-‐specific IgG levels are compared with Mann-‐Whitney U test. 2803
Mean age in those with and without raised Aspergillus-‐specific IgG levels is compared by 2804
2-‐sided t-‐test. Categorical variables are compared with Chi-‐squared test, except for 2805
comparison of number of positive Aspergillus-‐specific IgG tests in tuberculosis cases vs. 2806
healthy controls, where Fisher’s exact test is used. 2807
2808
Discussion 2809 2810
Interpretation of raised Aspergillus-‐specific IgG in these studies 2811
2812
Papers 4 and 5 describe opportunistic testing of stored sera in an attempt to estimate 2813
the prevalence of pulmonary aspergillosis in patients with presumed ‘smear negative 2814
pulmonary tuberculosis’ and proven pulmonary tuberculosis respectively. These 2815
studies do not claim to measure the prevalence of pulmonary aspergillosis. To achieve a 2816
measurement of the prevalence of ‘proven’ acute invasive pulmonary aspergillosis in 2817
line with EORTC guidelines one would need to perform biopsy on all patients48. To 2818
define patients as having ‘probable’ acute invasive pulmonary aspergillosis would 2819
require CT scan evidence of progressive cavitation with paracavitary infilatrates or 2820
aspergilloma. 2821
2822
These investigations were not performed as part of the MIND-‐IHOP study276. It is 2823
therefore not possible to claim that the investigations described here measure the 2824
prevalence of invasive pulmonary aspergillosis in this population and that claim is not 2825
made in the articles. However these definitions were designed to identify invasive 2826
aspergillosis in profoundly immunosuppressed patients. The subacute pulmonary 2827
aspergillosis seen in non-‐neutropaenic AIDS patients52,223,236 is probably closer the 2828
syndrome of chronic necrotizing pulmonary aspergillosis seen in patients with 2829
moderate immunosuppression secondary to diabetes or alcohol excess51 2830
2831
Over 200 stored sera were available for use. From these, 39 were selected from a group 2832
of patients whose presentation was as close as possible to the definition of subacute 2833
invasive pulmonary aspergillosis6,264. These were HIV positive patients with chronic 2834
99
cough with abnormal chest X-‐ray who did not have evidence of pulmonary tuberculosis 2835
after the most extensive investigation possible, including use of the highly sensitive 2836
Cepheid GeneXpert assay279. 2837
2838
Such patients are at risk of subacute invasive pulmonary aspergillosis52,223,236 and as no 2839
other diagnosis was found it is reasonable to suggest that pulmonary aspergillosis is 2840
probable where raised Aspergillus-‐specific IgG is present in this group. The fact that the 2841
diagnostic threshold for a positive test was defined in relation to Ugandan healthy 2842
controls support this claim. The 40% mortality rate noted in this group is potentially 2843
consistent with undiagnosed pulmonary aspergillosis occurring in this group. 2844
2845
Limitations in case definition for ‘smear negative tuberculosis’ 2846
2847
There are, however several potential limitations to this approach. The duration of 2848
symptoms required for the accepted case definition of subacute invasive aspergillosis is 2849
>1 month6,264. The MIND-‐IHOP study accepted patients with cough duration of 2 weeks 2850
or more276. As a result many will have had bacterial pneumonia and improved with 2851
antibiotics. Such patients were, however excluded by restricting the analysis to patients 2852
with ‘unknown’ final diagnosis. Any patients whose symptoms resolved with antibiotics 2853
were classified as probable ‘bacterial pneumonia’. Those who presented with less than 1 2854
month of symptoms, but had raised Aspergillus-‐specific IgG and did not improve with 2855
antibiotics are likely to have had acute invasive aspergillosis48, which is also often 2856
associated with raised Aspergillus-‐specific IgG90,96. The study design cannot accurately 2857
differentiate acute from subacute invasive pulmonary aspergillosis, but this 2858
differentiation is less important than simply estimating the overall prevalence of likely 2859
pulmonary aspergillosis in this group for the first time. 2860
2861
The diagnosis of subacute invasive pulmonary aspergillosis also normally requires the 2862
presence of progressive cavitation or paracavitary infiltrates6,264. These were defined on 2863
CT scan in the published cohorts of this condition. The MIND-‐IHOP study did not include 2864
CT scan276. As the interpretation of chest X-‐rays is challenging in the context of subacute 2865
respiratory disease265 and the features of invasive pulmonary aspergillosis are often 2866
non-‐specific218,280, any patient with abnormal chest X-‐ray was included in this analysis if 2867
100
other criteria were met. It is unlikely that all patients in this group would have met the 2868
diagnostic criteria for subacute invasive pulmonary aspergillosis6,48,264 if CT scan had 2869
been performed. 2870
2871
Limitations of serological testing in patients with AIDS 2872
2873
The original design of this study included testing for galactomannan, which has a higher 2874
sensitivity than Aspergillus-‐specific IgG for the diagnosis of acute invasive pulmonary 2875
aspergillosis90. Unfortunately most patients in the study received Ceftriaxone prior to 2876
sampling. This can cause false positive galactomannan results281. Galactomannan levels 2877
can also fall significantly with frozen storage261. As a result of these major flaws 2878
galactomannan was dropped from the study design, leaving Aspergillus-‐specific IgG as 2879
the sole marker of pulmonary aspergillosis. Patients with AIDS may not form a good 2880
antibody response to infection or vaccination in the context of hepatitis B and C273,274. 2881
The same problem might occur in relation to antibody response to Aspergillus. The sole 2882
use of Aspergillus-‐specific IgG as a serological marker for probable acute or subacute 2883
pulmonary aspergillosis in patients with AIDS might therefore underestimate the 2884
frequency of the disease. 2885
2886
Possible selection bias relating to stored sera volume 2887
2888
There is also a possibility of selection bias as only around three quarters of patients 2889
recruited to the MIND-‐IHOP study had stored sera available. The others did not have 2890
sufficient volume of sera left over after previous tests were complete. Although this 2891
process is not very likely to introduce bias, it is conceivable that it would remove sicker 2892
patients from the analysis, as they might be more difficult to withdraw blood from due 2893
to shock and so have lower volumes of stored sera. 2894
2895
The result described here is therefore at best an estimate of the possible prevalence of 2896
acute or subacute invasive pulmonary aspergillosis in this group. It is the best that can 2897
be achieved from opportunistic testing of sera from a study that was not specifically 2898
designed to measure the prevalence of this disease. 2899
101
While absolute accuracy is not guaranteed, such data does still have significant value. 2900
There has never been an attempt to measure the prevalence of subacute invasive 2901
aspergillosis in African AIDS patients, or indeed in any living HIV-‐infected cohort, using 2902
serological methods. The fact that this condition is noted in 2-‐3% of all AIDS 2903
autopsies,207,238,239 underlines the potential importance of this disease. 90% of these 2904
cases were undiagnosed antemortem. Prospective studies to accurately measure the 2905
prevalence of this condition would be expensive, due to the need for CT scans and 2906
biopsies. The data from opportunistic testing described here provides the evidence 2907
needed to justify such prospective studies. 2908
2909
Interpretation of raised Aspergillus-‐specific IgG in patients with proven tuberculosis 2910
2911
The population tested in paper five is much more clearly defined, in that all patients 2912
included here had proven pulmonary tuberculosis, on the basis of culture, Cepheid 2913
GeneXpert PCR testing or smear testing. The difficulty here is the interpretation of the 2914
raised levels of Aspergillus-‐specific IgG. 2915
2916
There is no published evidence that the Aspergillus-‐specific IgG cross reacts with 2917
antibodies to M. tuberculosis. It therefore seems likely that those with raised Aspergillus-‐2918
specific IgG do have Aspergillus growth in their lungs or airways. This might represent 2919
early CPA, but this diagnosis could not reasonably be confirmed until after tuberculosis 2920
treatment is complete, as CPA diagnosis requires the presence of symptoms. Any cough 2921
or haemoptysis present during active tuberculosis could well be due to the tuberculosis 2922
rather than CPA. If the patient became asymptomatic after tuberculosis treatment then 2923
the symptomatic criteria for CPA would not be met. 2924
2925
It is equally possible that Aspergillus colonization of the diseased airways at the site of 2926
tuberculosis infection is more common than in healthy persons. This might be due to 2927
the immunosuppression produced by active tuberculosis infection282. Immune function 2928
might return to normal once the tuberculosis infection is cured and Aspergillus 2929
colonisation might then resolve as a result. 2930
2931
2932
102
Selected nature of tuberculosis population in this study 2933 2934
It should also be noted that the population tested here are patients diagnosed with 2935
pulmonary tuberculosis in the course of an acute emergency hospital admission. This 2936
population is not representative of the overall population with pulmonary tuberculosis. 2937
The results presented here cannot therefore be taken as a measure of the overall 2938
prevalence of CPA co-‐existing with active pulmonary tuberculosis. It should be noted, 2939
however that the diagnostic threshold used in this survey was defined in relation to 2940
healthy Ugandan controls and that 90% of patients with treated tuberculosis tested in 2941
paper 2 had normal levels using this cut-‐off. The high frequency of raised Aspergillus-‐2942
specific IgG is therefore an unexpected and significant finding. 2943
2944
While this opportunistic study cannot define the exact nature of the Aspergillus disease 2945
in these patients it does provide valuable insight into the interaction between M. 2946
tuberculosis and Aspergillus infection. In particular it strongly suggests that the 2947
commonly held assumption that CPA develops after tuberculosis is cured may be 2948
wrong. This raises the possibility that future programs to screen and treat CPA in 2949
patients with pulmonary tuberculosis might need to focus on patients with current 2950
tuberculosis rather than past tuberculosis. The results described here are therefore of 2951
significant value to those planning future prospective studies to define natural history 2952
of CPA in relation to pulmonary tuberculosis. 2953
2954
2955
2956
103
PAPER 1 - Comparison of six Aspergillus-specific IgG assays for the diagnosis of 2957
chronic pulmonary aspergillosis (CPA) and allergic bronchopulmonary aspergillosis 2958
(ABPA) 2959
2960
Authors 2961
2962
Iain D Page -‐ Institute of Inflammation and Repair, The University of Manchester, UK, 2963
Manchester Academic Health Science Centre, UK, National Aspergillosis Center, 2964
University Hospital of South Manchester, UK 2965
2966
Malcolm Richardson -‐ Institute of Inflammation and Repair, The University of 2967
Manchester, UK, Manchester Academic Health Science Centre, UK, National Aspergillosis 2968
Center and Mycology Reference Centre, University Hospital of South Manchester, UK 2969 2970
David W Denning -‐ Institute of Inflammation and Repair, The University of Manchester, 2971
UK, Manchester Academic Health Science Centre, UK, National Aspergillosis Center, 2972
University Hospital of South Manchester, UK 2973
2974
2975
104
Abstract 2976
2977
Measurement of Aspergillus-‐specific IgG, or precipitating antibodies is a key component 2978
of diagnosis of CPA. It also has a role in ABPA, where it is one of three ‘additional 2979
features’ of ABPA recognized in recent international society for human and animal 2980
mycology (ISHAM) diagnostic guidelines. Many commercial Aspergillus-‐specific IgG 2981
assays exist, but there is limited evidence regarding sensitivity and specificity of these 2982
assays for the diagnosis of CPA or ABPA. The optimal diagnostic cut offs for CPA and 2983
ABPA are poorly defined. 2984
2985
We performed Aspergillus-‐specific IgG testing on stored sera from the following 2986
patients; 1 – CPA (n=241) 2 – ABPA (n=80) 3 – healthy controls (n=100), 4 – severe 2987
asthmatic controls (n=100). The following assays were used; ThermoFisher Scientific 2988
ImmunoCAP (multi-‐national), Siemens Immulite (Germany), Serion (Germany), Genesis 2989
(UK) and Dynamiker (China) and counterimmunoelectrophoresis (precipitins) using 2990
Microgen Aspergillus antigens. 2991
2992
Receiver operating curve area under the curve (ROC AUC) results for CPA diagnosis 2993
were as follows; ImmunoCAP 0.995 (95% CI 0.991 – 0.999), Immulite 0.991 (0.982-‐1), 2994
Serion 0.973 (0.960 – 0.987), Dynamiker 0.918 (0.89 – 0.946) and Genesis 0.902 (0.871 2995
– 0.933). 2996
2997
ROC AUC results comparing ABPA patients to healthy controls were; ImmunoCAP 0.959 2998
(0.935 – 0.987), Immulite 0.932 (95% CI 0.887 – 0.977), Serion 0.907 (0.866 – 0.949), 2999
Dynamiker 0.903 (0.859 – 0.946) and Genesis 0.73 (0.651 – 0.808). Compared to severe 3000
asthmatic patients ROC AUC results were; Immulite 0.837 (0.774 – 0.9), Serion 0.826 3001
(0.763 – 0.888), Dynamiker 0.819 (0.754 – 0.885) and Genesis 0.797 (0.728 – 0.866). 3002
3003
Optimal diagnostic cut-‐offs for CPA were; ImmunoCAP 20 mg/L (96% sensitivity, 98% 3004
specificity), Immulite 10 mg/L (96% sensitivity, 98% specificity), Serion 35 U/ml (90% 3005
sensitivity, 98% specificity), Dynamiker 65 AU/ml (77% sensitivity, 97% specificity) 3006
and Genesis 20 U/ml (75% sensitivity, 99% specificity). Precipitins were 59% sensitive 3007
and 100% specific. Sensitivity of these cut offs for diagnosis of ABPA were as follows; 3008
105
Immulite (81%), ImmuncoCAP (77%), Dynamiker (66%), Serion (62%) and Genesis 3009
(46%). Specificity for the diagnosis of ABPA was lower when compared to severe 3010
asthmatics and alternative diagnostic thresholds may be appropriate for use in this 3011
group. 3012
3013
ROC AUC results for ImmunoCAP and Immulite are both statistically significantly 3014
superior to all other assays tested for the diagnosis of both CPA and ABPA in 3015
comparison to healthy controls. The Genesis assay was statistically significantly inferior 3016
to all other assays for the diagnosis of ABPA in comparison to healthy controls. The 3017
currently accepted ImmunoCAP cut-‐off of 40 mg/L is sub-‐optimal for CPA diagnosis. 3018
Adopting the new proposed thresholds for CPA diagnosis maximizes sensitivity while 3019
maintaining specificity of 97% or higher for each assay. 3020
3021
Precipitins testing performed poorly for the diagnosis of CPA and ABPA and should be 3022
replaced by IgG ELISA in these contexts, however it performed well for identifying CPA 3023
occurring as a complication of ABPA. Most ELISA assays performed poorly in the latter 3024
context, with the exception of Immulite which had an ROC AUC of 0.863 and produced 3025
sensitivity of 71% and specificity of 91% using an optimal cut off of 100 mg/L for this 3026
purpose. 3027
3028
3029
106
Introduction 3030
3031
CPA is a serious disease that leads to severe disability and death7, but which can be 3032
treated effectively with existing drugs and surgery15,18,58,198,283. The estimated global 3033
prevalence of CPA is around 3 million cases11–13. Diagnosis of CPA requires the presence 3034
of chronic symptoms, plus appropriate radiological findings and microbiological 3035
evidence of disease5,7,8,50. The latter can be provided from biopsy or by culture of either 3036
broncho-‐alveolar lavage (BAL) fluid or sputum. Acquiring samples for these tests either 3037
requires an invasive procedure to acquire BAL or biopsy or the production of a sputum 3038
sample, which can be troublesome for patients with intermittent cough. Culture also has 3039
poor sensitivity with current methods260. As a result many CPA patients never have a 3040
positive culture or biopsy8. By comparison raised levels of Aspergillus-‐specific IgG are 3041
present in the majority of published cases and provide the sole laboratory evidence of 3042
Aspergillus infection in many cases5,8. 3043
3044
ABPA can complicate asthma and is estimated to complicate around 13% of asthma 3045
cases284 and affect around 5 million persons worldwide13. The international society for 3046
human and animal mycology (ISHAM) has recently reviewed the diagnostic criteria for 3047
ABPA4. The presence of both raised total IgE and raised Aspergillus-‐specific IgE (or 3048
positive skin prick testing) is mandatory for diagnosis. Positive precipitins or 3049
Aspergillus-‐specific IgG is one of three additional features, along with raised eosinophil 3050
count and radiological features, of which 2 out of 3 are also required to confirm the 3051
diagnosis. 3052
3053
Many specialist laboratories have developed their own ‘home-‐brew’ assays directly 3054
from fungal culture, to detect Aspergillus-‐specific IgG 18,73,130,131. However replication of 3055
such a ‘home-‐brew’ technique in other laboratories is challenging141 as the mixture of 3056
antigens produced varies in relation to factors such as strain, medium pH and length of 3057
culture98,128,134–136,138–140. The original Ouchterlony precipitation-‐in-‐gel (precipitins) 3058
technique105,123 for detection of Aspergillus-‐specific antibodies is time consuming, 3059
produces subjective results of a semi-‐quantitative nature and probably has poor 3060
sensitivity38,74. Enzyme-‐linked immunosorbent assay (ELISA) is now commonly used in 3061
its place130. 3062
107
Multiple commercial tests for Aspergillus-‐specific IgG exist, but published data 3063
comparing the performance of these tests is very limited. Research has been hampered 3064
by the lack of large cohorts of patients with clearly defined CPA. Some comparisons use 3065
small mixed populations of different types of aspergillosis, including invasive disease in 3066
addition to CPA or ABPA73. However, good intra-‐laboratory repeatability has been 3067
demonstrated for the ImmunoCAP assay (ThermoFisher Scientific, multinational) 180. 3068
3069
To our knowledge only two studies have compared the performance of commercial 3070
Aspergillus-‐specific IgG assays for the diagnosis of CPA. One study noted that the Bio-‐3071
Rad (France) and Serion (Germany) assays had respective sensitivity of 94% and 92% 3072
for the diagnosis of CPA in 51 cases, with specificity of 87% and 76% respectively38. The 3073
other study, published by our team, compared the Bio-‐Rad assay to ImmunoCAP and 3074
precipitins testing using Microgen (UK) antigens. It showed respective sensitivity of 3075
85%, 86% and 56% for the diagnosis of CPA in 116 cases 74. 3076
3077
Diagnostic cut-‐offs for ABPA in the patients with underlying cystic fibrosis have been 3078
investigated for the ImmunoCAP assay. One study of 87 patients suggests that a cut-‐off 3079
of 90 mg/L has a sensitivity of 91% and specificity of 88% for the diagnosis of ABPA78, 3080
whereas another study of 146 patients suggests that a cut-‐off of 75 mg/L66 has a 3081
sensitivity of 96% and a specificity of 90%. The ImmunoCAP assay has also been 3082
assessed in 10 ABPA patients without underlying cystic fibrosis, where a cut-‐off of 35 3083
mg/L had a sensitivity of 90% and specificity of 86% for the diagnosis of ABPA73. 3084
3085
The diagnostic cut-‐offs for other assays for ABPA have not been assessed in the 3086
published literature. Using the manufacturer’s recommended cut-‐offs the Serion and 3087
Bio-‐Rad assays had sensitivities of 84% and 92% respectively, in a study of 13 patients 3088
without cystic fibrosis38. ImmunoCAP, Bio-‐Rad and precipitins had sensitivities of 41%, 3089
47% and 15% respectively in a mixed group of 46 patients with either ABPA or Severe 3090
Asthma with Fungal Sensitization (SAFS), but no cystic fibrosis74. The Immulite assay 3091
(Siemens, Germany) has been shown to have good correlation with the ImmunoCAP 3092
assay180, but its sensitivity and specificity for the diagnosis of CPA or ABPA has not been 3093
measured directly. 3094
3095
108
These comparisons are probably too small to detect differences in test sensitivity and 3096
specificity and are potentially biased due to the presence of long-‐term antifungal 3097
therapy, which lowers Aspergillus-‐specific IgG levels, in many patients58. We are not 3098
aware of any publications describing the sensitivity and specificity of Bordier 3099
(Switzerland), Dynamiker (China), IBL (Germany) or Genesis (UK) for the diagnosis of 3100
CPA or ABPA. The optimal diagnostic thresholds for CPA and ABPA (in patients without 3101
cystic fibrosis) have never been assessed for any of the available assays, with the 3102
exception of the small ImmunoCAP ABPA study described above. 3103
3104
We have performed a retrospective comparison of six methods in cohorts of CPA 3105
patients and ABPA patients. Samples were taken when patients were not taking long-‐3106
term antifungal medication. Receiver operating characteristic (ROC) curve analysis 3107 285,286 is used to compare test performance. It is also to define optimal diagnostic cut offs 3108
for CPA and ABPA. 3109
3110
Methods 3111
3112
Patients 3113
3114
Eighty patients with ABPA and 241 patients with CPA were identified at the UK National 3115
Aspergillosis Centre (NAC). Each had a stored sample of serum taken when either off 3116
antifungal treatment or within three months of starting treatment. The median levels of 3117
Aspergillus-‐specific IgG in study patients on antifungal therapy and off antifungal 3118
therapy are described and compared. Control samples were collected from 100 healthy 3119
Ugandan blood donors 100 patients with severe asthma under the care of the North 3120
West Lung Centre, UK. Samples were tested for Aspergillus-‐specific IgG by all methods, 3121
other than asthmatic controls where ImmunoCAP was not performed as no funding was 3122
available to perform this assay in this group. 3123
3124
Diagnosis of CPA or ABPA was taken from patient records. CPA diagnosis at the NAC is 3125
based on the composite gold standard comprising symptoms, radiological changes, 3126
raised inflammatory markers and microbiological evidence of Aspergillus infection as 3127
described previously by our group5 and subsequently accepted in many 3128
109
studies7,8,18,58,108,198,253,254 and in IDSA and ESCMID guidelines250,251. Diagnosis of ABPA 3129
at the NAC is based on a composite gold standard comprising raised total IgE, positive 3130
Aspergillus-‐specific IgE or skin prick testing, abnormal radiological findings, raised 3131
Aspergillus-‐specific IgG and raised eosinophil count4,287. All patients with ABPA at our 3132
centre are routinely screened for the development of CPA, principally with regular chest 3133
imaging. Any patient with progressive lung cavities, paracavitary fibrosis or 3134
aspergilloma on imaging was classified as CPA and removed from the ABPA group. 3135
3136
Laboratory techniques 3137
3138
Tests were performed between January and July 2014. Aspergillus-‐specific IgG levels 3139
were measured on all samples using the Siemens Immulite 2000. Manual plate ELISAs 3140
were performed on all samples using kits supplied by Serion, Genesis and Dynamiker. 3141
All results for plate ELISAs were read on a PolarStar Omega spectrophotometer (BMG 3142
Labtech, UK). Optical density readings were converted to arbitrary units using the 3143
formulae or software provided by test manufacturers. Results were rejected if the 3144
manufacturers’ stated quality control criteria were not met for an individual test plate. 3145
If this occurred the tests were re-‐run on a fresh plate. Where a result was greater than a 3146
threshold specified by the manufacturer a 1 in 10 dilution was performed and the assay 3147
was repeated. 3148
3149
Precipitation in gel (precipitins) testing was also performed on all samples using the 3150
counterimmunoelectrophoresis (CIE) technique and Aspergillus antigens supplied by 3151
Microgen (UK). Briefly, 10ml agarose was melted and poured onto a hydrophobic gel 3152
bond film (GE Healthcare, USA). Three mm diameter test wells were cut once the gel 3153
had set. Twenty µL sera were placed in one row of wells with 20 µL antigens (Microgen, 3154
UK) at 2 mg/ml placed in the adjacent row. The gel was placed above a CIE tank filled 3155
with veronal buffer and blotting paper wicks were used to connect either end of the gel 3156
to the buffer tanks before applying 34V for 90mins. After running the gel was placed in 3157
sodium chloride washing solution overnight. After drying it was placed in a Coomassie 3158
Blue stain for 15 mins, followed by two serial de-‐stain solutions of 10 mins each. After 3159
further drying the gels were read on a light box with the assistance of a magnifying 3160
glass. The presence of any precipitins bands was reported as a positive result. Neat 3161
110
serum was tested for all samples. Where samples were positive serial dilution to a 3162
maximum 1 in 32 dilution was produced to provide dilutional titres. The same 3163
technician performed all the above tests. 3164
3165
Aspergillus-‐specific IgG levels were performed on the ThermoFisher Scientific 3166
ImmunoCAP system for all CPA and ABPA cases as part of routine clinical care at the 3167
time of original sampling. Where a sample produced a result of >200 mg/L a 1 in 10 3168
dilution was performed and the sample was retested. Other assays were performed on 3169
these same samples after storage. Healthy control samples were also tested. 3170
3171
Where a sample produced the same result (positive or negative) on a single test by all 3172
test methods this result was accepted. Where a sample produced divergent results on 3173
different assays it was repeated twice. If the two new tests resulted in a different 3174
outcome (positive or negative) to the first test then the mean of these two new tests 3175
replaced the first test. 3176
3177
72 (14%) samples were repeated for the Dynamiker assay with 13 (2%) results 3178
changed. 38 (7%) samples were repeated for the Serion assay with 10 (2%) results 3179
changed. 52 (10%) samples were repeated for the Genesis assay with 5 (1%) results 3180
changed. 42 (8%) samples were repeated for the Siemens assay with 2 (0.4%) results 3181
changed. 112 (21%) samples were repeated for the CIE assay with 15 (3%) results 3182
changed. ImmunoCAP samples were not repeated as results from tests performed as 3183
part of clinical care at the time of original sampling were being used and no funding was 3184
available for repeat testing. We report final results after any repeat testing for all assays. 3185
3186
Intra-‐assay variability 3187
3188
To calculate intra-‐assay variability (IAV) we selected two samples for each assay other 3189
than ImmunoCAP, with high and low results respectively. No funding was available to 3190
perform IAV on ImmunoCAP. Each assay was repeated 20 times per sample. Outliers 3191
were identified by study team consensus and removed from the final analysis. For the 3192
low level repeats four Serion samples and one Dynamiker sample with readings lower 3193
than the substrate blank were removed as outliers, as was one Genesis sample with 3194
111
apparent cross-‐well contamination. No samples were removed as outliers for the high 3195
level IAV. The Serion IAV was repeated three times with similar results each time. 3196
Arbitrary units from different assays cannot be directly compared. Result range, mean, 3197
standard deviation and co-‐efficient of variation (CV) are reported for each assay. CV was 3198
calculated as (standard deviation / mean) X 100. 3199
3200
Statistical analysis 3201
3202
Intra assay variation with co-‐efficient of variation is reported for Immulite, Serion, 3203
Genesis and Dynamiker assays. Descriptive statistics are reported for each assay in each 3204
patient group, including the frequency of positive, negative and intermediate results by 3205
manufacturers current guidelines. The median Aspergillus-‐specific IgG levels in patients 3206
on antifungal therapy and patients not on antifungal therapy are compared by Mann-‐3207
Whitney U analysis. ImmunoCAP and Immulite both produce results in mg/L. 3208
Correlation between these two assays is measured by Spearman’s rank analysis. 3209
3210
ROC curve analysis is performed for each ELISA assay and the Area Under the Curve 3211
(AUC) for ROC analysis is shown with 95% confidence intervals (95% CI). Wald’s 3212
statistic is used to compare the significance of differences in ROC AUC between assays. A 3213
significant result defined as p<0.005. 3214
3215
Optimal diagnostic cut-‐offs for each assay are calculated using Youden’s J statistic 3216
(sensitivity + specificity -‐ 1). Sensitivity and specificity are described for these cut-‐offs. 3217
These comparisons are performed for CPA cases vs. healthy controls, ABPA cases vs. 3218
healthy controls, ABPA cases vs. severe asthmatic controls and ABPA cases vs. CPA 3219
cases. Statistical analyses were performed using SPSS version 20 (IBM, USA) under 3220
license to the University of Manchester, UK. 3221
3222
Results 3223
3224
Stored samples from 241 patients with proven CPA, 80 patients with ABPA and 100 3225
patients with asthma were acquired from the UK National Aspergillosis Centre and 3226
North West Lung Centre. One hundred control samples were acquired from healthy 3227
112
blood donors in Gulu, Uganda. Patient demographics and underlying conditions for the 3228
patient groups are compared in table 1. 3229
3230
Table 1 – Patient and control characteristics 3231
3232
Characteristic
CPA n=241
ABPA n=80
Asthmatic controls n = 100
Healthy controls n=100
Female gender 101 (42%) 42 (52%) 77 (77%) 55 (55%) Mean age (years) 65 67 52 years 19 Age range (years) 23 – 92 25 -‐ 95 19 -‐ 81 17 – 39 Chronic cavitary pulmonary aspergillosis
238 (99%) 0 0 0
Aspergillus nodule disease
3 (1%) 0 0 0
HIV 0 0 0 2 (2%) Prior tuberculosis 37 (15%) 1 (1%) 1 (1%) 0 Non-‐tuberculous mycobacterial infection
28 (12%) 0 0 0
COPD 85 (35%) 3 (4%) 3 (3%) 0 Bronchiectasis 60 (25%) 43 (54%) 29 (29%) 0 ABPA 35 (15%) 80 (100%) 0 0 Sarcoidosis 9 (4%) 0 0 0 Malignancy (active or in remission)
33 (14%) 7 (9%) 1 (1%) 0
Autoimmune disease
33 (14%) 3 (4%) 2 (2%) 0
Diabetes 7 (3%) 8 (10%) 4 (4%) 0 Asthma 41 (17%) 78 (97%) 100 (100%) 0 Cystic fibrosis 0 2 (3%) 0 0 3233 HIV = human immunodeficiency virus seropositive; COPD = chronic obstructive pulmonary disease; ABPA 3234 = allergic bronchopulmonary aspergillosis 3235 3236
All kits produced results within the manufacturers’ stated quality control criteria for the 3237
tests included in the analysis, with the exception of Dynamiker. In this case the high 3238
concentration control serum was slightly below the stated range in all runs. Forty-‐two 3239
(17%) of CPA patients included in the study had received up to three months antifungal 3240
therapy at the time of sampling. The remaining 199 (83%) patients were not on any 3241
antifungal therapy. Median Aspergillus-‐specific IgG levels in CPA patients on and off 3242
antifungal therapy are shown in table 2. Nine of the 80 (11%) ABPA patients had 3243
113
received up to 3 months antifungal therapy at the time of sampling. The remaining 71 3244
(89%) were not on any antifungal therapy. Median Aspergillus-‐specific IgG levels in 3245
ABPA patients on and off antifungal therapy are shown in table 3. Median levels are 3246
compared by Mann-‐Whitney U test. 3247
3248
Table 2 – Aspergillus-‐specific IgG levels in CPA patients with and without antifungal 3249 therapy 3250 3251
3252
3253 3254 Table 3 – Aspergillus-‐specific IgG levels in ABPA patients with and without antifungal 3255 therapy 3256 3257
3258
Test Median Aspergillus-‐specific IgG level in those with < 3 months antifungals n=42
Median Aspergillus-‐specific IgG level in those not on antifungals n=199
p-‐value
ImmunoCAP 130 mg/L 125 mg/L 0.375 Immulite 533 mg/L 250 mg/L 0.051 Serion 143 U/ml 125 U/ml 0.372 Genesis 90 U/ml 47 U/ml 0.006
Dynamiker 141 AU/ml 119 AU/ml 0.230
Test Median Aspergillus-‐specific IgG level in those with < 3 months antifungals n=9
Median Aspergillus-‐specific IgG level in those not on treatment n=71
p-‐value
ImmunoCAP 68 mg/L 45 mg/L 0.337 Immulite 49 mg/L 36 mg/L 0.819 Serion 92 U/ml 57 U/ml 0.479 Genesis 18 U/ml 13 U/ml 0.849 Dynamiker 331 AU/ml 103 U/ml 0.173
114
Intra-‐assay variation for low-‐level samples is shown in table 4. Intra-‐assay variation 3259
results for high-‐level samples are shown in table 5. All low level precipitins repeats 3260
were negative and all high level repeats were positive, but with dilutional titre results as 3261
follows; neat = 1 sample, 1 in 2 = 2 samples, 1 in 4 = 11 samples, 1 in 8 = 4 samples, 1 in 3262
16 = 1 sample. 3263
3264
Table 4 – Intra-‐assay variation -‐ low 3265
3266
Test Range Mean Standard deviation
Co-‐efficient of variation
Dynamiker (AU/ml)
45.5 – 66.2 (20.7) 55.8 5.6 10.1%
Genesis (U/ml)
4.6 – 6.3 (1.6) 5.2 0.4 8.2%
Serion (U/ml)
6 -‐ 42.5 (36.5) 24 10.5 43.7%
Immulite (mg/L)
58.4 – 67.8 (9.4) 62.6 2.2 3.6%
AU = arbitrary units, U = units. Both represent arbitrary numbers and no direct comparison can be made 3267 between assays producing results in this manner. 3268 3269
Table 5 – Intra-‐assay variation -‐ high 3270
3271
Test Range Mean Standard deviation
Co-‐efficient of variation
Dynamiker (AU/ml)
240.7 – 372.7 (132) 287.5 31.9 11.1%
Genesis (U/ml)
55.4 – 96.5 (41.1) 83.4 10 12.1%
Serion (U/ml)
59.9-‐122.1 (62.2) 75 17.4 23.2%
Immulite (mg/L)
95.3 – 107 (11.7) 99.6 3.4 3.4 %
3272
Box and whisker plots with logarithmic scale compare results for cases and control 3273
groups for each assay in figures 1-‐5. Results in cases and control groups are 3274
summarized in table 6. Where manufacturers provide instructions on interpretation of 3275
results, outcomes are summarized in table 7. Dynamiker, Genesis and Serion advise 3276
115
reporting of results as positive, intermediate or negative. ImmunoCAP is interpreted 3277
with a single diagnostic cut-‐off (40 mg/L) in line with current UK practice and Immulite 3278
does not currently have a recommended diagnostic cut-‐off. 3279
3280
Precipitins tests produced the following results in CPA cases; negative = 102 cases 3281
(42%), neat = 23 cases (10%), 1 in 2 = 34 cases (14%), 1 in 4 = 29 cases (12%), 1 in 8 = 3282
26 cases (11%), 1 in 16 = 23 cases (10%), 1 in 32 = 4 cases (2%). For ABPA cases 4% of 3283
cases had positive precipitins results with neat sera only, all others were negative. 3284
3285
The correlation between ImmunoCAP and Immulite in patients with CPA, ABPA and in 3286
healthy and asthmatic controls was good (Spearman’s rank analysis 0.876, p 0.000). 3287
3288
Table 6 – Results in CPA cases and healthy controls 3289 3290 3291
ImmunoCAP testing could not be performed on sera from asthmatic patients, as no funding was available. 3292 Results expressed in U/ml or AU/ml are arbitrary and cannot be directly compared across assays. 3293 3294
ROC curves comparing CPA and ABPA to healthy controls are shown in figures 6 and 7. 3295
ROC curves comparing ABPA to asthmatic controls are shown in figure 8 and ROC 3296
curves comparing Aspergillus-‐specific IgG levels in CPA and ABPA are shown in figure 9. 3297
Results of ROC AUC analysis are presented in table 8. The ROC analyses identified 3298
optimal cut-‐offs for each situation. We report the specificity and sensitivity of each of 3299
these for the diagnosis of CPA in tables 9-‐12. The suggested optimal cut off for each 3300
assay is highlighted in bold. 3301
3302
Test
Controls range (n=100)
Asthma range (n=100)
ABPA cases range (n=80)
CPA cases range (n=241)
Controls mean
Asthma mean
ABPA mean
CPA mean
Controls median
Asthma median
ABPA median
CPA median
Dynamiker (AU/ml)
16-‐88 26 -‐ 643
27-‐5239
23-‐6118
37 93 334 341 34 46 119 124
Genesis (U/ml)
0-‐20 1 -‐ 25 0-‐362 1-‐930 7 6 33 111 6 4 14 60
Immulite (mg/L)
0–35 0 -‐ 87 0-‐149 3-‐7660 5 11 46 678 4 7 39 392
ImmunoCAP (mg/L)
2-‐36 -‐ 3-‐408 9-‐1707 6 -‐ 66 216 5 -‐ 46 126
Serion (U/ml)
0-‐40 1 -‐ 416 1-‐898 4-‐3436 10 34 112 232 6 16 57 131
116
3303 Table 7 – Frequency of positive results by manufacturers’ guidelines 3304
Test Frequency of positive results in controls (intermediate results)
Frequency of positive results in ABPA (intermediate results)
Frequency of positive results in CPA (intermediate results)
Dynamiker 6% (11%) 69% (7%) 78% (5%) Genesis 22% (13%) 59% (7%) 82% (5%) Immulite* n/a n/a n/a ImmunoCAP 0 54% 88% Serion 0 (0) 43% (14%) 74% (10%) Precipitins 0 4% 59% *Immulite do not currently provide diagnostic cut-‐offs so the number of positive results by manufacturers 3305 guidelines cannot be reported. 3306 3307
3308
Table 8– Receiver operating characteristic curve area under curve (ROC AUC) results 3309 Test CPA vs.
healthy controls
95% CI ABPA vs. healthy controls
95% CI ABPA vs. asthmatic controls
95% CI CPA vs. ABPA
95% CI
ImmunoCAP 0.996 0.992 -‐ 1
0.961 0.935 – 0.987
-‐ -‐ 0.778 0.723 – 0.834
Immulite 0.991 0.982 -‐ 1
0.932 0.887 – 0.977
0.818 0.753 – 0.883
0.863 0.825 – 0.901
Serion 0.973 0.960 – 0.987
0.907 0.866 – 0.949
0.760 0.690 – 0.831
0.698 0.629 – 0.768
Dynamiker 0.918 0.890 – 0.946
0.903 0.859 – 0.946
0.725 0.651 – 0.799
0.519 0.443 – 0.596
Genesis 0.902 0.871 – 0.933
0.73 0.651 – 0.808
0.797 0.728 – 0.866
0.759 0.701 – 0.818
3310
Nine of 241 sera from CPA cases were negative (<20 mg/L) on testing with the 3311
ImmunoCAP assay. Using the new diagnostic cut-‐offs suggested above 6 of these 9 3312
(67%) were positive on Immulite testing (mean level in positives 93 mg/mL), 3 of 9 3313
(33%) samples were positive on Serion testing (mean level in positive 217 U/ml), 1 3314
sample (11%) was positive by Genesis at 130 U/ml, 1 sample (11%) were positive by 3315
Dynamiker at 614 AU/ml and 3 (33%) were positive by CIE (two neat, one at 1 in 2). 3316
Two samples were negative on all assays. 3317
3318
117
Wald’s statistic confirmed the overall difference in ROC AUC performance for the 3319
diagnosis of CPA across the five assays is statistically significant (p<0.0001). 3320
ImmunoCAP ROC AUC is equivalent to Immulite (p=0.32). ImmunoCAP and Immulite 3321
both have significantly superior ROC AUC to the other 3 assays (p=0.0006 for Immulite 3322
vs. Serion). Serion has a superior ROC AUC to Dynamiker and Genesis (p<0.0001 for 3323
Serion vs. Dynamiker). Dynamiker and Genesis have equivalent ROC AUC (p=0.38). 3324
3325
Table 9 – Potential diagnostic cut offs for CPA 3326
Assay Diagnostic cut-‐off
Sensitivity Specificity Youden’s J statistic
ImmunoCAP 10mg/L 100% 86% 0.86 20 mg/L 96% 98% 0.94 30 mg/L 91% 99% 0.9 40 mg/L 87% 100% 0.87
Immulite 5 mg/L 99% 78% 0.77 10 mg/L 96% 98% 0.94 20 mg/L 93% 99% 0.92 30 mg/L 91% 99% 0.9
Serion 30 U/ml 91% 95% 0.86 35 U/ml 90% 98% 0.88 40 U/ml 88% 99% 0.87 45 U/ml 85% 100% 0.85
Dynamiker 60 AU/ml 78% 94% 0.72 65 U/ml 77% 97% 0.74 70 AU/ml 75% 98% 0.73 75 AU/ml 73% 99% 0.72
Genesis 10 U/ml 86% 65% 0.51 15 U/ml 79% 95% 0.74 20 U/ml 75% 99% 0.74 25 U/ml 71% 100% 0.71
Precipitins -‐ 59% 100% 0.59 3327
3328
Wald statistic confirmed an overall difference in the ROC AUC performance of the assays 3329
in the comparison of ABPA patients to healthy controls, with ImmunoCAP, Immulite, 3330
Serion and Dynamiker all demonstrating statistically significantly superior ROC AUC 3331
results to Genesis (p<0.001). There was no statistically significant difference in the 3332
performance of the other four assays in this context. ImmunoCAP ROC AUC was 3333
equivalent to Immulite (p=0.27), Siemens Immulite ROC AUC was equivalent to Serion 3334
(p=0.27) and Serion ROC AUC was equivalent to Dynamiker (p=0.21).3335
118
Table 10 –Potential diagnostic cut offs for ABPA vs. healthy controls 3336
Assay Diagnostic cut-‐off
Sensitivity Specificity Youden’s J statistic
ImmunoCAP 10mg/L 94% 86% 0.8 20 mg/L 77% 98% 0.75 30 mg/L 64% 99% 0.63 40 mg/L 54% 100% 0.54
Immulite 5 mg/L 94% 77% 0.71 10 mg/L 81% 98% 0.79 20 mg/L 60% 98% 0.58 30 mg/L 56% 99% 0.55
Serion 30 U/ml 65% 95% 0.6 35 U/ml 62% 98% 0.6 40 U/ml 57% 100% 0.57 45 U/ml 55% 100% 0.55
Dynamiker 55 AU/ml 70% 88% 0.58 60 AU/ml 69% 95% 0.64 65 AU/ml 66% 97% 0.63 70 AU/ml 64% 98% 0.62
Genesis 10 U/ml 66% 65% 0.31 15 U/ml 46% 95% 0.41 20 U/ml 37% 100% 0.37 25 U/ml 31% 100% 0.31
Precipitins -‐ 4% 100% 0.04 3337
In the comparison of assays in ABPA cases vs. asthmatics Wald’s statistic demonstrated 3338
no overall difference in the performance of the assays (p=0.1). For the comparison of 3339
CPA vs. ABPA there was a difference in the overall performance of the assays (<0.0001). 3340
Immulite ROC AUC was superior to all other assays in this context (p=0.0004 for 3341
Immulite vs. ImmunoCAP). Immulite, ImmunoCAP and Genesis ROC AUCs were all 3342
superior to Serion (p<0.0001 for Genesis vs. Serion). All assays were superior to 3343
Dynamiker (p<0.0001 for Serion vs. Dynamiker), which has no diagnostic value in this 3344
setting as the lower 95% confidence interval for ROC AUC crosses 0.5. 3345
3346
3347
119
Table 11 –Potential diagnostic cut offs for ABPA vs. severe asthmatic controls 3348 Assay Diagnostic
cut-‐off Sensitivity Specificity Youden’s J
statistic Immulite 10 mg/L 81% 67% 0.48
20 mg/L 60% 88% 0.48 30 mg/L 56% 94% 0.5 40 mg/L 50% 97% 0.47 50 mg/L 41% 97% 0.38
Serion 30 U/ml 65% 70% 0.35 50 U/ml 55% 86% 0.41 70 U/ml 41% 89% 0.3 90 U/ml 34% 93% 0.27 100 U/ml 27% 95% 0.23
Dynamiker 60 AU/ml 68% 63% 0.31 80 AU/ml 60% 74% 0.34 100 AU/ml 56% 79% 0.35 120 AU/ml 50% 80% 0.3 500 AU/ml 20% 96% 0.16
Genesis 10 U/ml 66% 83% 0.49 15 U/ml 46% 93% 0.39 20 U/ml 38% 95% 0.33 25 U/ml 31% 99% 0.3
Precipitins -‐ 4% 97% 0.01 3349
3350
120
Table 12 –Potential diagnostic cut offs for CPA vs. ABPA 3351
Assay Diagnostic cut-‐off
Sensitivity Specificity Youden’s J statistic
ImmunoCAP 20 mg/L 96% 23% 0.19 50 mg/L 84% 55% 0.39 100 mg/L 56% 78% 0.34 150 mg/L 37% 95% 0.32 200 mg/L 25% 98% 0.23
Immulite 10 mg/L 95% 19% 0.14 50 mg/L 84% 59% 0.43 100 mg/L 71% 91% 0.62 125 mg/L 65% 95% 0.6 150 mg/L 61% 100% 0.61
Serion 35 U/ml 90% 38% 0.28 50 U/ml 84% 45% 0.29 100 U/ml 60% 73% 0.33 200 U/ml 31% 84% 0.15 400 U/ml 14% 95% 0.09
Genesis 15 U/ml 80% 54% 0.34 50 U/ml 52% 88% 0.4 75 U/ml 43% 93% 0.36 100 U/ml 29% 93% 0.22 125 U/ml 24% 96% 0.2
Precipitins -‐ 59% 96% 0.54 3352
Discussion 3353
3354
This is the first study to compare the performance of Aspergillus-‐specific IgG assays in 3355
large populations of well-‐characterised patients with clinically confirmed CPA and 3356
ABPA, who were not on long term antifungal treatment at the time of sampling. The 3357
performance of both the ThermoFisher Scientific ImmunoCAP and Siemens Immulite 3358
assays was superior to the other four assays, but equivalent to each other. 3359
3360
The Serion assay has statistically superior performance to the other two manual assays 3361
produced by Genesis and Dynamiker, but suffers from poor reproducibility. The 3362
performance of the manual Dynamiker galactomannan-‐specific IgG ELISA for the 3363
diagnosis of CPA was similar to manual ELISAs using culture extract antigens. We 3364
suggest that ImmunoCAP or Immulite assays be adopted as the test of choice for CPA 3365
diagnosis wherever possible. 3366
3367
121
The worst performing assay for CPA diagnosis was precipitins. We confirmed the poor 3368
sensitivity found in earlier studies74. In all cases with positive precipitins at least one 3369
ELISA was also positive. There are other manufacturers of reagents for precipitins 3370
testing, and further studies are needed to confirm that the low sensitivity of precipitins 3371
antibody detection was intrinsic to the system and not a peculiarity of the Microgen 3372
reagents. 3373
3374
Nonetheless, we argue that Aspergillus-‐specific IgG ELISA should now replace that 3375
precipitin testing for the diagnosis of CPA as the preciptins assay has no advantages 3376
over ELISA in this context. Furthermore many treatment trials in CPA have previously 3377
specified a positive precipitins test as a mandatory diagnostic criteria18,108. This is no 3378
longer appropriate. The term ‘precipitins’ has been used in the literature to refer to both 3379
precipitation-‐in-‐gel and Aspergillus-‐specific IgG ELISA8. This inaccuracy is confusing and 3380
should be avoided given the differing performance characterstics of the two techniques. 3381
Medical practitioners will need to be educated about this shift as ‘Aspergillus preciptins’ 3382
testing remains part of the routine vocabulary of respiratory physicians in the English-‐3383
speaking world. 3384
3385
A small number of CPA sera were negative on the ImmunoCAP assay. In these cases 3386
other assays produced strongly positive results in most cases, with Immulite 3387
demonstrating the best sensitivity in this group. Each assay will have its own mix of 3388
antigens. These patients may be reacting only to certain antigens, present in some 3389
assays, but not others. Two patients had no response to any of the assays, suggesting an 3390
underlying immune deficit preventing an effective antibody response to Aspergillus 3391
infection. 3392
3393
When being used for ABPA diagnosis, the Genesis assay had inferior performance to all 3394
other ELISAs when diagnosing ABPA in relation to healthy controls. There was no 3395
significant difference between the four assays assessed for the diagnosis of ABPA 3396
against a severely asthmatic population. For diagnosis of CPA complicating ABPA the 3397
Immulite assays was statistically significantly superior to all other assays. The 3398
Dynamiker assay performed worst, with no diagnostic value for identification of CPA 3399
122
complicating ABPA. Precipitins performed well in this context, with good specificity and 3400
superior sensitivity to all ELISAs other than Immulite. 3401
3402
Manufacturers often recommend reporting results as positive, negative or intermediate, 3403
with repeat testing in intermediate cases. This may be appropriate in the dynamic 3404
context of acute invasive aspergillosis, but in chronic diseases such as CPA and ABPA 3405
any change on repeat testing is likely to simply represent the inter-‐assay variability of 3406
the test. We therefore advocate the use of a single diagnostic cut-‐off for these 3407
conditions. 3408
3409
We have confirmed the earlier finding by van Toorenenbergen180 that there is good 3410
correlation between the ImmunoCAP and Immulite assays. In our study median levels in 3411
ABPA and healthy controls were similar for both assays, however the median Immulite 3412
level was three times higher than the ImmunoCAP level in CPA cases. This should be 3413
taken into consideration when comparing results from different laboratories using 3414
different assays, especially with higher levels. Laboratories should identify the assay 3415
used when reporting test results. 3416
3417
Patients on long term antifungal therapy were excluded from the study on the grounds 3418
that such therapy is likely to reduce Aspergillus-‐specific IgG and so introduce bias to 3419
diagnostic cut-‐off calculations58. However, we included patients who had received up to 3420
three months of antifungal therapy in addition to those who were not on antifungal 3421
therapy. This was necessary as a common source of stored samples was from sera taken 3422
for drug level measurement soon after starting therapy. Most patients included were 3423
not on antifungal therapy. 3424
3425
The only significant difference in results between patients on and off antifungal therapy 3426
was found in CPA patients using the Genesis assay, where levels were higher in patients 3427
on therapy. All other assays showed no significant difference in the median Aspergillus-‐3428
specific IgG levels between CPA patients on no antifungal therapy and those on 3429
antifungal therapy for up to three months. No assay produced a significant difference in 3430
the ABPA patients. There is therefore no evidence that up to three months antifungal 3431
123
therapy produces the reduction in Aspergillus-‐specific IgG levels seen with long-‐term 3432
therapy58. 3433
3434
One of the limitations of our study was that it was limited to testing in a single 3435
laboratory with tests kits from a single batch. Published evidence of intra-‐laboratory 3436
variability currently exists only for the ImmunoCAP assay180. In addition, ImmunoCAP 3437
testing was also performed on fresh samples from CPA patients, while all other tests 3438
were performed on frozen stored samples. Long term storage does not appear to 3439
significantly reduce other antibody levels in serum262,288 and there is no reason to 3440
believe that Aspergillus-‐specific IgG will behave differently to other antibodies in this 3441
respect. However, it would have been desirable to demonstrate that this is specifically 3442
true for Aspergillus-‐specific IgG by testing stored samples with the ImmunoCAP assay 3443
and confirming that there was no significant difference between results obtained before 3444
and after storage. Unfortunately no funding was available for such a comparison. 3445
3446
We were also unable to perform intra-‐assay variability testing for the ImmunoCAP 3447
assay, as no funding was available for this. However, three existing studies from 3448
separate laboratories have shown between-‐run CV results from <5% to 23% for this 3449
assay73,78,180 and the inter-‐laboratory CV for ImmunoCAP is 7.3-‐18.1%180. 3450
3451
In most CPA cases in our cohort microbiological evidence of CPA was provided by raised 3452
Aspergillus-‐specific IgG. The ImmunoCAP and precipitins assays were routinely used for 3453
clinical testing in our reference laboratory throughout the study period, but with a 3454
diagnostic cutoff of 40 mg/L for ImmunoCAP. The Immulite, Serion, Genesis and 3455
Dynamiker assays are not used routinely at any UK diagnostic laboratory. While there 3456
may therefore be a degree of selection bias in favour of the ImmunoCAP system, this did 3457
not prevent Immulite system demonstrating equivalent ROC AUC performance to 3458
ImmunoCAP in our study. 3459
3460
Our study defines optimal cut-‐offs for Aspergillus-‐specific IgG by comparing levels in 3461
CPA cases to Ugandan healthy controls. This is appropriate as raised levels of 3462
Aspergillus-‐specific IgG form only one aspect of CPA diagnostic criteria5,250. However 3463
levels are also raised in other conditions such as Aspergillus bronchitis, Aspergillus 3464
124
rhinosinusitis or allergic bronchopulmonary aspergillosis4,39,41. The Aspergillus-‐specific 3465
IgG assay cannot therefore be used in isolation to diagnose CPA. 3466
3467
The median ImmunoCAP level in our healthy controls was 5mg/L. The median 3468
ImmunoCAP level in Dutch blood donors is 8.75 mg/L180 and in healthy female Belgian 3469
laboratory workers it is 13.75 mg/L73. These results are consistent with our proposed 3470
diagnostic cut-‐off of 20mg/L, which we suggest should now be adopted for the 3471
diagnosis of CPA. The median Serion level in pregnant French women is 20 AU/ml38. 3472
This is consistent with the diagnostic cut-‐off of 35AU/ml produced by our ROC analysis. 3473
We are not aware of any prior descriptions of Omega and Dynamiker levels in healthy 3474
controls. 3475
3476
The median Immulite value in Dutch blood donors is 13.2 mg/L180. This is higher than 3477
both the median level of 4mg/L we found in healthy Ugandan blood donors and the 3478
optimal diagnostic cut-‐off of 10 mg/L produced by our ROC analysis. This significant 3479
difference might be explained by different levels of Aspergillus exposure in these 3480
different environments or differences in the median age of these healthy control groups. 3481
3482
We chose to use healthy Ugandan blood donors as a control group as a key goal of this 3483
study was to select an assay for use in a CPA prevalence survey to be conducted in 3484
Uganda. While the proposed Immulite cut off of 10mg/L is appropriate for the diagnosis 3485
of CPA in Ugandans, further comparisons including European healthy controls may be 3486
needed before this cut off can be used with confidence in Europe. The age matching 3487
between our controls and cases was sub-‐optimal, but the use of blood donors is 3488
common practice and was the only practical option available to us to acquire a healthy 3489
Ugandan control group. 3490
3491
CPA occurs in patients with underlying diseases such as treated tuberculosis, 3492
sarcoidosis and COPD14. We have shown that levels of Aspergillus-‐specific IgG are 3493
different in asthmatics than healthy controls and levels might also be higher than 3494
healthy controls in patients with these other underlying diseases, perhaps as a result of 3495
frequent Aspergillus colonisation in these patients. 3496
125
The purpose of the Aspergillus-‐specific IgG assay in the context of CPA diagnosis is to 3497
provide evidence of Aspergillus infection. Radiological and clinical criteria then need to 3498
be met before CPA can be diagnosed. While the optimal diagnostic cut-‐offs defined here 3499
are appropriate to define ‘abnormally high levels of Aspergillus-‐specific IgG’ and provide 3500
evidence of Aspergillus infection for use as a single aspect of the CPA composite 3501
diagnostic criteria, they might frequently produce positive results in a population with 3502
underlying lung disease if Aspergillus colonisation is common in that population. 3503
Further studies are needed to compare levels of Aspergillus-‐specific IgG in CPA to those 3504
in ‘at-‐risk’ diseased controls. 3505
3506
Optimal cut-‐offs for ABPA diagnosis have been calculated for each assay, both by 3507
comparing ABPA cases to healthy controls and by comparing them to asthmatic 3508
controls. Comparison to asthmatic controls is the more valid of these two options, 3509
however there are some difficulties with this approach. The asthmatic controls used in 3510
this study are patients being treated at a regional referral centre with a sub-‐specialist 3511
interest in fungal lung disease. Patients referred to a centre of this nature are likely to 3512
have fungal sensitization or unusually severe asthma, as demonstrated by the fact that 3513
29% of patients in the asthmatic control group had evidence of bronchiectasis. 3514
3515
As such they may be at increased risk of Aspergillus colonisation and are probably not 3516
representative of the overall asthmatic population in this respect. It is also possible that 3517
a population of severe asthmatics such as this may include patients whose disease has 3518
been complicated by ABPA, which was not diagnosed at the time of sampling. Our 3519
retrospective study design did not allow the identification and exclusion of any such 3520
patients. These factors might explain why this asthmatic cohort has a higher median 3521
Aspergillus-‐IgG level than healthy controls. It is therefore possible that the cut off 3522
produced by the comparison of ABPA cases with healthy controls is more appropriate 3523
for use in the diagnosis of ABPA in the asthmatic population as a whole. 3524
3525
Ideally, further studies should be performed involving the testing of samples from a 3526
non-‐selected asthmatic population, ideally one treated at primary care level. An ideal 3527
study design would include prospective screening of patients to remove cases of ABPA 3528
from the cohort. Such a design would allow the identification of a definitive cut off for 3529
126
the diagnosis of ABPA in asthmatics. Such an optimal design was not possible within 3530
time and financial constraints associated with this study. We have therefore reported 3531
diagnostic cut offs using both the sub-‐optimal control groups available to us at the time 3532
of this study. The ideal diagnostic cut off probably lies somewhere between the levels 3533
calculated from the two groups. 3534
3535
ABPA can be complicated by the development of CPA. Aspergillus-‐specific IgG 3536
measurement might be used to identify cases of CPA complicating ABPA. This is 3537
important as the management of CPA differs from ABPA, with long-‐term antifungal 3538
therapy being indicated wherever possible and surgery often required18,21,250. We have 3539
suggested optimal cut-‐offs in terms of Youden's J statistic. However it may be more 3540
clinically useful to select a cut off with high specificity, above which CPA is likely and CT 3541
scan should be recommended to further investigate the possibility. The Immulite had 3542
reasonable sensitivity and specificity for this purpose with an optimal cut off of 100 3543
mg/L, but other assays performed poorly. Precipitins testing also performed reasonably 3544
well in this context. However, no assay performs well enough to be used alone to 3545
diagnose CPA in those with ABPA. CPA must be considered possible in any patient with 3546
ABPA and raised Aspergillus-‐specific IgG. 3547
3548
All tests were fairly labour intensive to perform. Automated platforms required manual 3549
sample loading and dilution. Plate ELISAs required over 10 pipetting steps. The 3550
precipitins test was the most labour-‐intensive requiring around 18 gels to be run for 3551
every one ELISA plate used in the comparison. We also undertook preliminary work 3552
with the ELITech (France) Aspergillus-‐specific haemagglutination assay289, which we 3553
found to be labour-‐intensive with end point reading that was highly subjective. As a 3554
result this assay was not taken forward to the main analysis. 3555
3556
The Dynamiker Aspergillus-‐specific IgG assay uses purified galactomannan as its sole 3557
antigen and performed similarly to equivalent manual plate ELISAs for the diagnosis of 3558
CPA and ABPA. This antigen might be appropriate for use in an Aspergillus-‐specific IgG 3559
lateral flow device (LFD) that would be ideal for use in resource-‐poor settings. No such 3560
LFD exists at present. 3561
3562
127
Our study did not include all available commercial Aspergillus-‐specific IgG ELISAs. Bio-‐3563
Rad produce an Aspergillus-‐specific IgG ELISA that uses recombinant antigens. This 3564
assay has been shown to have similar sensitivity and specificity to the Serion and 3565
ImmunoCAP assays in two small studies38,74, however these studies did not exclude 3566
patients on long-‐term antifungal therapy and used the sub-‐optimal diagnostic cut-‐off of 3567
40mg/L for ImmunoCAP. Other commercial assays exist, but have no published data 3568
describing their sensitivity and specificity for the diagnosis of CPA. 3569
3570
We have described the diagnostic performance of six of the most commonly used 3571
Aspergillus-‐specific IgG assays for the diagnosis of CPA and ABPA, the most common 3572
forms of pulmonary aspergillosis. The ImmunoCAP assay is currently widely used with 3573
a diagnostic cut-‐off of 40 mg/L. This is sub-‐optimal for the diagnosis of CPA and should 3574
be replaced with a diagnostic cut-‐off of 20 mg/L. Optimal diagnostic cut-‐offs for use in 3575
CPA have been defined for all assays, which can improve sensitivity for the diagnosis of 3576
CPA while maintaining excellent specificity. Further studies are now required to 3577
confirm intra-‐laboratory and batch-‐to-‐batch variation for these assays. This will 3578
hopefully then allow the roll out of routine testing of at risk patients, including those in 3579
areas of high tuberculosis prevalence, where most CPA patients are predicted to reside, 3580
but where access to Aspergillus serology is currently extremely limited. 3581
3582
Hypothesis 3583
3584
That there are clinically relevant differences in the sensitivity and specificity of different 3585
Aspergillus-‐specific IgG assays or precipitins testing in the context of diagnosis of 3586
chronic pulmonary aspergillosis (CPA) and allergic pulmonary aspergillosis (ABPA). 3587
3588
Aims 3589
3590
1 – To measure the levels of Aspergillus-‐specific IgG found in groups of patients with 3591
untreated CPA or ABPA and in healthy and asthmatic controls, using assays produced by 3592
Siemens Immulite, ThermoFisher Scientific ImmunoCAP, Serion, Genesis and 3593
Dynamiker, plus precipitins testing using Microgen antigens. 3594
3595
128
2 – To define the diagnostic performance for each of these assays for CPA and ABPA by 3596
performing receiver operating curve (ROC) area under the curve (AUC) analysis 3597
comparing patients with CPA and ABPA to healthy and diseased controls. 3598
3599
3 – To define an optimal diagnostic cut-‐off for each assay in relation to ROC AUC 3600
analysis for both CPA and ABPA in relation to both healthy and diseased controls. 3601
3602
4 – To measure intra-‐assay variability for both high and low level samples for each 3603
Aspergillus-‐specific IgG assay, within funding restrictions. 3604
3605
5 – To measure the correlation between Siemens Immulite and ThermoFisher Scientific 3606
ImmunoCAP results. 3607
3608
Ethics 3609
3610
Control samples were acquired as part the ‘Pulmonary aspergillosis in association with 3611
tuberculosis’ study. Ethical approval was granted by Gulu University IRB (ref 3612
GU/IRC/04/07/12), the Ugandan National Council for Science and Technology (ref 3613
HS1253) and the University of Manchester (ref 11424). Stored serum was taken from 3614
samples provided by CPA patients for the purpose of Aspergillus-‐specific IgG testing as 3615
part of routine care of CPA and ABPA at the National Aspergillosis Centre, Manchester, 3616
UK. Further stored serum samples were acquired from the ManRAB biobank. Ethical 3617
approval was granted by the ManRAB REC committee (ref 10/H1010/7). 3618
3619
Funding 3620
3621
Siemens, Serion, Genesis and Dynamiker all donated sufficient test kits to perform this 3622
comparison. Serion and Dynamiker each provided grant support to cover the cost of 3623
laboratory consumables. 3624
3625
Control samples were acquired as part the ‘Pulmonary aspergillosis in association with 3626
tuberculosis’ study (paper two), which was funded by a grant from the University 3627
129
Hospital of South Manchester Academy charity as part of the established Manchester-‐3628
Gulu link program. 3629
3630
Acknowledgements 3631
3632
We would like to thank the staff of Gulu Blood Transfusion service, Uganda, for their 3633
assistance in recruiting donations of serum from healthy blood donors for use as control 3634
samples. 3635
3636
We would like to thank the staff at Manchester Royal Infirmary immunology laboratory 3637
for their assistance in performing ImmunoCAP Aspergillus-‐specific IgG testing on 3638
control samples. 3639
3640
We would like to thank the staff at Christie Hospital pathology laboratory, Manchester 3641
for permitting access to their Siemens Immulite 2000 system to perform this study. 3642
3643
We would like to thank Siemens, Serion, Genesis and Dynamiker for kindly donating test 3644
kits to perform this study and for their practical assistance in installing the test kits and 3645
relevant software prior to undertaking the study. 3646
3647
We would like to thank the ManRAB biobank at University Hospital of South Manchester 3648
for providing stored sera from National Aspergillosis Centre CPA patients for use in this 3649
study. ManRAB is supported by the NIHR. 3650
3651
3652
3653
3654
3655
130
Figures 3656
3657
Figure 1 – Dynamiker results in various patient groups 3658
3659
3660
3661
131
Figure 2 – Genesis results in various patient groups 3662
3663
3664 3665
132
3666
Figure 3 – ThermoFisher Scientific ImmunoCAP results in various patient groups 3667
3668
3669 Samples from asthmatic patients were not tested with the ImmunoCAP assay, as no 3670 funding was available for this. 3671 3672
133
Figure 4 – Serion results in various patient groups 3673
3674
3675
134
Figure 5 – Siemens Immulite results in various patient groups 3676
3677
3678 3679
135
3680 Figure 6 –ROC curve for CPA cases vs. healthy controls 3681
3682
3683
3684 3685
136
3686
Figure 7 – ROC curves for ABPA cases vs. healthy controls 3687
3688
3689
3690 3691
137
3692
Figure 8 – ROC curves for ABPA cases vs. asthmatic controls 3693
3694
3695 3696
138
3697
Figure 9 – CPA cases vs. ABPA cases 3698
3699
139
PAPER 2 - Aspergillus-specific IgG levels in patients previously treated for pulmonary 3700
tuberculosis in Gulu, Uganda 3701
3702
Authors 3703
3704
Iain D Page – Institute of Inflammation and Repair, The University of Manchester, UK, 3705
Manchester Academic Health Science Centre, UK, National Aspergillosis Center, 3706
University Hospital of South Manchester, UK. 3707
3708
Nathan Onyachi – Gulu Regional Referral Hospital, Uganda. 3709
3710
Cyprian Opira – St. Mary’s Hospital, Lacor, Gulu, Uganda. 3711
3712
Sharath Hosmane – University Hospital of South Manchester, UK 3713
3714
Richard Sawyer -‐ University Hospital of South Manchester, UK 3715
3716
Malcolm Richardson – Institute of Inflammation and Repair, The University of 3717
Manchester, UK, Manchester Academic Health Science Centre, UK, National Aspergillosis 3718
Center and Mycology Reference Centre, University Hospital of South Manchester, UK. 3719
3720
David W Denning– Institute of Inflammation and Repair, The University of Manchester, 3721
UK, Manchester Academy Health Science Centre, UK, National Aspergillosis Centre, 3722
University Hospital of South Manchester, UK. 3723
3724
3725
140
Abstract 3726
3727
In 1970, 34% of 544 British patients with residual cavities after treated pulmonary 3728
tuberculosis were found to have precipitating antibodies to Aspergillus. Aspergilloma 3729
was detected in 63% of those with antibodies and was often complicated by 3730
haemoptysis. Based on this data the global 5-‐year period prevalence of chronic 3731
pulmonary aspergillosis (CPA) is estimated at 0.8 to 1.3 million cases. There are no 3732
published surveys from current areas of high tuberculosis prevalence to confirm this 3733
prediction and the impact of HIV co-‐infection on the prevalence of CPA is not known. 3734
We aimed to measure the prevalence of Aspergillus-‐specific IgG in Ugandan patients 3735
with treated pulmonary tuberculosis. 3736
3737
We conducted a cross-‐sectional survey in Gulu, Uganda. Recruitment was open to all 3738
persons aged 16 or over who had completed treatment for pulmonary tuberculosis in 3739
the last 7 years. Eligible patients were identified with the assistance of clinic staff at 3740
Gulu Hospital and the District Health Team. Radio announcements were used to 3741
encourage patients to participate. All patients underwent clinical assessment and chest 3742
X-‐ray and had Aspergillus-‐specific IgG measured by Siemens Immulite. 3743
3744
Recruitment was undertaken between October 2012 and February 2013. 400 patients 3745
were recruited. 200 (50%) were HIV positive. Median age was 42 years (range 16-‐83). 3746
39% of patients were female. Median CD4 count in those with HIV was 415 cells/µL 3747
(range 0-‐1400). 3748
3749
Raised Aspergillus-‐specific IgG was found in 10% of patients. Chronic cough was 3750
reported by 33% of patients and haemoptysis by 3% of patients. 4% of all patients had 3751
suspected fungal ball on chest x-‐ray, with cavitation present in 16% and pleural 3752
thickening in 15%. 3753
3754
This study cannot measure the prevalence of CPA, as it does not include CT scan and 3755
serial chest X-‐ray. However, we suggest that a patient who has ALL of the following 3756
should be considered to have possible CPA; 1 – chronic symptoms (over 1 month of 3757
cough or haemoptysis), 2 – Raised levels of Aspergillus-‐specific IgG and 3 – Chest X-‐ray 3758
141
findings consistent with CPA (cavities or aspergilloma). Overall 12 (3%) met these 3759
criteria. A further 2 (0.5%) had simple aspergilloma, without chronic symptoms. HIV 3760
status had no statistically significant impact on the frequency of likely CPA. 3761
3762
3763
142
Introduction 3764
3765
An estimated 9 million people developed tuberculosis in 2013215. It was associated with 3766
1.5 million deaths, of which only 210,000 were estimated to be due to multidrug 3767
resistant strains. Many of the other 1.29 million deaths will have been due to late 3768
presentation to medical care, lack of diagnosis, poor access to treatment or inadequate 3769
adherence, given that they mostly occured in resource-‐poor countries with weak health 3770
infrastructure. However, misdiagnosis may also have contributed to the problem. 3771
3772
Chronic pulmonary aspergillosis (CPA) is a condition that complicates tuberculosis14. 3773
CPA usually presents with progressive pulmonary cavitation associated with weight 3774
loss, persistent cough and haemoptysis5,7,8. It has a 5-‐year mortality of 50 – 80%6,7,264 3775
and has recently been estimated to affect around 3 million people globally11–13, 3776
including 1.3 million cases secondary to tuberculosis11. This estimate takes no account 3777
of the potential impact of HIV co-‐infection, which is present in half of the cases of 3778
suspected pulmonary tuberculosis notified in Uganda215. 3779
3780
Undiagnosed CPA could be making a substantial contribution to the observed mortality 3781
rates currently attributed to tuberculosis. Both conditions present with cavities, pleural 3782
thickening and fibrosis on chest X-‐ray266,280. Aspergillomas are distinctive, but while 3783
they are present in all cases of simple aspergilloma, they are present in only 25-‐36% of 3784
cases of CPA in developed countries8,58. Raised levels of Aspergillus-‐specific IgG are key 3785
to diagnosis of CPA5,7,8, but this test is generally unavailable in Africa220. In Uganda 34% 3786
of all notified cases of pulmonary tuberculosis are clinically diagnosed with no 3787
microbiological proof of tuberculosis infection215. Some of these cases may well be CPA 3788
that has been misdiagnosed as tuberculosis. 3789
3790
Large CPA case series have been reported in the UK, France, India, China, Korea and 3791
Japan, the majority of which are secondary to tuberculosis7,8,14,15,18,108,198. Over 180 3792
cases of CPA have been reported throughout Africa, including South Africa, Nigeria, 3793
Ivory Coast, Senegal, Central African Republic, Djibouti, Ethiopia, Tanzania and 3794
Uganda16,201–212. Over 90% of these cases were secondary to pulmonary tuberculosis. 3795
3796
143
CPA is treatable. Oral treatment with itraconazole, voriconazole or posaconazole 3797
prevents clinical and radiological progression18,58,108,198,251. Surgery is curative in 3798
selected patients with localized disease15,21 and has been safely delivered in resource-‐3799
poor settings16,54,212. 3800
3801
The prevalence of CPA was measured in 544 patients with residual lung cavities after 3802
tuberculosis treatment in the UK in 1968-‐7076,197. Precipitating antibodies to Aspergillus 3803
fumigatus were present in 34%, of whom 63% had an aspergilloma visible on chest X-‐3804
ray within 48 months of completion of tuberculosis treatment. Subsequent series have 3805
found positive Aspergillus-‐specific antibodies in 20-‐27% of patients previously treated 3806
for pulmonary tuberculosis in Japan, India and Brazil80,146,192,221. 3807
3808
CPA prevalence in areas where tuberculosis is now common might differ from the UK in 3809
1968-‐70. Rates of Aspergillus rhinitis and keratitis are higher in countries with warm 3810
climates and many subsistence farmers10. This might also be true for CPA. Biomass 3811
smoke-‐induced emphysema is common in Africa222 and might increase CPA risk14. 3812
Crucially HIV co-‐infection might either result in more CPA cases due to 3813
immunosuppression52,223,224 or fewer due to reduced the rate of residual cavitation seen 3814
in those co-‐infected with HIV225–227. 3815
3816
We conducted a cross-‐sectional survey to measure the prevalence of raised Aspergillus-‐3817
specific IgG in persons with treated pulmonary tuberculosis in Gulu, Uganda. We 3818
recorded presence of chronic symptoms, performed chest X-‐ray and measured 3819
Aspergillus-‐specific IgG using the Siemens Immulite 2000 system. This assay has 3820
specificity of 98% and sensitivity of 96% for the diagnosis of CPA (paper 1). 3821
3822
This study cannot definitively measure the prevalence of CPA as it does not include CT 3823
scan or serial chest X-‐ray. However, we have classified patients who meet all of the 3824
following conditions as ‘possible CPA’ ; 1 – chronic symptoms (over 1 month of cough or 3825
haemoptysis), 2 – Raised levels of Aspergillus-‐specific IgG and 3 – Chest X-‐ray findings 3826
consistent with CPA (cavities or fungal ball). 3827
3828
144
We also diagnosed simple aspergilloma in patients with fungal ball on chest X-‐ray and 3829
raised Aspergillus-‐specific IgG, but with no chronic cough or haemoptysis. We targeted 3830
recruitment of 50% of patients with HIV co-‐infection to measure the impact of HIV 3831
status on the frequency of raised Aspergillus-‐specific IgG and possible CPA. 3832
3833
Methods 3834
3835
Study design and participants 3836
3837
Patients aged 16 or over, who had completed a full course of treatment for pulmonary 3838
tuberculosis in 2005 or later, were recruited in Gulu, Uganda from October 2012 to 3839
February 2013. Evidence of tuberculosis treatment was taken from tuberculosis 3840
treatment cards, completion of treatment certificates, or from the District Health 3841
Tuberculosis Team’s central records. Patients with documentary evidence of fully 3842
treated smear-‐negative pulmonary tuberculosis were also accepted, but only if they 3843
reported complete resolution of all symptoms at the end of tuberculosis treatment. 3844
3845
We aimed to recruit 400 patients, of whom 200 would be HIV negative and 200 HIV 3846
positive. We calculated this would have sufficient power to measure the prevalence of 3847
CPA in this population with an accuracy of +/-‐ 2.3%. 3848
3849
Procedures 3850
3851
Convenience sampling was used. Eligible patients were identified from District Health 3852
team records and invited to join the study. Radio announcements were used to 3853
encourage patients to enroll. All patients were provided with written study information 3854
and written consent was given prior to recruitment. Illiterate patients were provided 3855
with verbal information in English or in Acholi via a translator and gave verbal consent. 3856
3857
Patients underwent structured clinical assessment. HIV status was taken from patients’ 3858
medical notes or TB treatment records. Where no such record was available patients 3859
underwent HIV testing prior to recruitment. Patients with no documented HIV status 3860
145
who declined testing were not eligible. Serum was tested for Aspergillus-‐specific IgG 3861
using the Immulite 2000 system (Siemens, Germany) in July 2014. 3862
3863
Chest X-‐ray was performed at St. Mary’s Hospital in Lacor and X-‐rays were 3864
photographed with a Nikon DSLR camera. Two radiologists reported results. Where 3865
they produced divergent reports the senior consultant respiratory radiologist at the UK 3866
National Aspergillosis Centre provided a decisive third report. All were blinded to 3867
clinical and serological findings. 3868
3869
Diagnostic criteria 3870
3871
Possible CPA was diagnosed when all three of the following criteria were met:-‐ 3872
3873
1 – Symptoms -‐ patients must have been suffering from at least one of the following 3874
symptoms for no less than 1 month. 3875
• Haemoptysis 3876
• Cough 3877
3878
2 – Radiological changes – at least one of the following features must be present on 3879
chest X-‐ray 3880
• Fungal Ball 3881
• Cavitation 3882
3883
3 – Raised Aspergillus-‐specific IgG 3884
3885
In addition, simple aspergilloma was diagnosed in patients with suspected fungal ball 3886
on chest X-‐ray and raised Aspergillus-‐specific IgG, but no chronic cough or haemoptysis. 3887
100 control sera had previously been collected from healthy Ugandan blood donors 3888
(paper 1). These were used in receiver operating characteristic curve studies to define 3889
the diagnostic threshold of 10 mg/L used in this study. 3890
3891
3892
146
3893 Statistical methods 3894
3895
Statistical analysis was performed using SPSS v20 (IBM, USA). Rates of raised 3896
Aspergillus-‐specific IgG and possible CPA in groups of patients with and without 3897
potential risk factors were compared using chi-‐squared test, except for comparisons 3898
with less than 5 patients in one group, where Fisher’s exact test was used. Comparison 3899
of means for continuous variables in different patient groups with normal distribution 3900
was performed using 2-‐sided t test. Where distribution was skewed Mann Whitney U 3901
test was used. 3902
3903
Results 3904
3905
400 patients were consented to enter the survey. One patient did not undergo chest X-‐3906
ray and one patient’s blood sample was lost, leaving 398 patients who completed the 3907
assessment process. Patient characteristics are shown in table 1. HIV status was 3908
documented on tuberculosis care records in most cases, with only a handful undergoing 3909
HIV testing on the day of recruitment. No patient declined HIV testing. 3910
3911
The overall frequency of various symptoms, X-‐ray abnormalities and Aspergillus-‐specific 3912
IgG levels are shown in table 2. Raised Aspergillus-‐specific IgG was found in 9.8% of 398 3913
patients with prior pulmonary tuberculosis and 2% of 100 healthy adult controls (p 3914
0.01) 3915
3916
147
Table 1 – Patient characteristics 3917
Characteristic Number of patients n=398
Female gender 155 (38.9%) Mean age (range) 42 years (16-‐83) Positive sputum smear at TB diagnosis
303 (76.1%)
HIV infection 199 (50%) Median 2012 CD4 count in HIV positive persons (range)
424 (14 – 1400) cells/µL
2012 CD4 count < 200 cells/µL
23 (12%#)
2012 CD4 count 200 – 499 cells/µL
94 (49.2%#)
2012 CD4 count ≥ 500 cells/µL
74 (38.7%#)
Traditional ‘grass-‐thatch’ home
371 (93.2%)
Patient reports dampness in home
119 (29.9%)
Patient is a subsistence farmer
373 (93.7%)
Patient frequently cooks on open charcoal stove
194 (48.7%)
Patient smokes tobacco 39 (9.8%) Median Aspergillus IgG 4.2 mg/L 3918
The frequency of symptoms and X-‐ray changes in patients with raised and normal levels 3919
of Aspergillus-‐specific IgG is compared in table 3. The frequency of suspected CPA and 3920
simple aspergilloma is shown in table 4, together with the number of patients with 3921
suspected fungal ball on chest X-‐ray, but normal levels of Aspergillus-‐specific IgG 3922
(unspecified fungal ball) and the number of patients with raised Aspergillus-‐specific IgG 3923
in whom the symptomatic and radiological criteria for CPA or simple aspergilloma are 3924
not met (raised Aspergillus-‐specific IgG, but no pulmonary aspergillosis). 3925
3926
Tables 5 – 9 show the frequency of symptoms, test results and diagnoses in relation to 3927
gender, prior TB smear status, HIV status, CD4 count and time since tuberculosis 3928
diagnosis respectively. 3929
3930
148
Table 2 – Symptoms and test findings 3931
Result No patients n=398
Frequency (%)
Cough* 130 32.7 Haemoptysis* 14 3.5 Fatigue* 191 48 Breathlessness* 193 48.5 Fevers* 99 24.9 Night sweats* 131 32.9 Chest pain* 214 53.8 Cavities on CXR 65 16.3 Paracavitary fibrosis on CXR
34 8.5
Pleural thickening on CXR
58 14.6
Fungal ball on CXR 15 3.8 Aspergillus IgG positive
39 9.8
*present for 1 month or longer 3932 3933
Table 3 – Symptoms and X-‐ray changes in patients with and without raised Aspergillus-‐3934 specific IgG 3935 3936 Results Raised
Aspergillus-‐specific IgG n=39
Normal Aspergillus-‐specific IgG n=359
p-‐value
Cough* 5 (12.8%) 60 (16.7%) 0.532 Haemoptysis* 6 (15.4%) 8 (2.2%) 0.000 Fatigue* 20 (51.3%) 171 (47.6%) 0.665 Breathlessness* 21 (53.8%) 172 (47.9%) 0.481 Fevers* 12 (30.8%) 87 (24.2%) 0.370 Night sweats* 11 (28.2%) 120 (33.4%) 0.510 Chest pain* 28 (71.8%) 186 (51.8%) 0.017 Cavities on CXR 18 (46.2%) 47 (13.1%) 0.000 Paracavitary fibrosis on CXR
6 (15.4%) 28 (7.8%) 0.108
Pleural thickening on CXR
15 (38.5%) 43 (14.8%) 0.000
Fungal ball on CXR
4 (12.9%) 11 (3%) 0.049**
*present for 1 month or more. **Fisher’s exact test used. 3937 3938
3939
149
3940 Table 4 – Frequency of various conditions 3941
Condition Number of cases N=398
Frequency (%) Frequency 95% confidence interval (%)
Likely CPA 12 3 1.7 -‐ 5 Likely simple aspergilloma
2 0.5 0.1 – 1.6
Unspecified fungal ball
11 2.8 1.5 – 4.7
Raised Aspergillus IgG, but no pulmonary aspergillosis
25 6.3 4.2 -‐ 9
3942
3943
Table 5 -‐ Symptoms and test results by gender 3944
Result Female n=155 Male n=243 p-‐value by chi-‐squared test
Cough* 47 (30.3%) 83 (34.2%) 0.426 Haemoptysis* 5 (3.2%) 9 (3.7%) 0.801 Fatigue* 80 (51.6%) 111 (45.7%) 0.248 Breathlessness* 69 (44.5%) 124 (51%) 0.205 Fevers* 42 (27.1%) 57 (23.5%) 0.413 Night sweats* 44 (28.4%) 87 (35.8%) 0.125 Chest pain* 85 (54.8%) 129 (53%) 0.732 Cavities on CXR 17 (11%) 48 (19.8%) 0.021 Paracavitary fibrosis on CXR
4 (2.6%) 30 (12.3%) 0.001
Pleural thickening on CXR
11 (7.1%) 47 (19.3%) 0.001
Fungal ball on CXR 3 (1.9%) 12 (4.9%) 0.125 Positive Aspergillus IgG
11 (7.1%) 28 (11.5%) 0.148
Median Aspergillus IgG level
3.84 mg/L 4.58 mg/L 0.002**
Likely CPA 3 (1.2%) 9 (3.7%) 0.381*** Likely simple aspergilloma
0 2 (0.8%) 0.523***
Unspecified fungal ball
2 (1.3%) 9 (3.7%) 0.214***
Raised Aspergillus IgG, but no pulmonary aspergillosis
8 (5.2%) 17 (7%) 0.462***
*present for 1 month or longer. ** medians compared by Mann Whitney U test, ***Fisher’s exact test 3945 3946
150
Table 6 – Symptoms and test results by prior TB smear status 3947
Results Prior smear positive pulmonary tuberculosis n= 303
Prior smear negative pulmonary tuberculosis n= 95
p-‐value by chi-‐squared test
Cough* 107 (35.3%) 23 (24.2%) 0.044 Haemoptysis* 9 (3%) 5 (5.3%) 0.029 Fatigue* 146 (48.2%) 45 (47.4%) 0.889 Breathlessness* 151 (49.8%) 42 (44.2%) 0.339 Fevers* 72 (23.8%) 27 (28.4%) 0.359 Night sweats* 103 (34%) 28 (29%) 0.413 Chest pain* 165 (54.4%) 49 (51.6%) 0.624 Cavities on CXR 55 (18.2%) 10 (10.5%) 0.079 Paracavitary fibrosis on CXR
32 (10.6%) 2 (2.1%) 0.01***
Pleural thickening on CXR
50 (16.5%) 8 (8.4%) 0.051
Fungal ball on CXR 13 (4.3%) 2 (2.1%) 0.537*** Positive Aspergillus IgG
32 (10.6%) 7 (7.4%) 0.361
Median Aspergillus IgG level
4.37 mg/L 3.91 mg/L 0.025**
Likely CPA 9 (1.7%) 3 (3.2%) 1*** Likely simple aspergilloma
1 (0.3%) 1 (1.1%) 0.421***
Unspecified fungal ball
10 (3.3%) 1 (1.1%) 0.472***
Raised Aspergillus IgG, but no pulmonary aspergillosis
22 (7.3%) 3 (3.2%) 0.224***
*present for 1 month or longer. ** medians compared by Mann Whitney U test. ***Fishers exact test used. 3948 3949
3950
151
Table 7 – Symptoms and test results by HIV status 3951
Results Positive HIV status n= 199
Negative HIV status n= 199
p-‐value by chi-‐squared test
Cough* 51 (25.6%) 79 (39.7%) 0.003 Haemoptysis* 5 (2.5%) 9 (4.5%) 0.276 Fatigue* 96 (48.2%) 95 (47.7%) 0.92 Breathlessness* 94 (47.2%) 99 (49.7%) 0.616 Fevers* 50 (25.1%) 49 (24.6%) 0.908 Night sweats* 54 (27.1%) 77 (38.7%) 0.014 Chest pain* 99 (49.7%) 115 (57.8%) 0.108 Cavities on CXR 24 (12.1%) 41 (20.6%) 0.021 Paracavitary fibrosis on CXR
14 (7.1%) 20 (20.1%) 0.282
Pleural thickening on CXR
23 (11.6%) 35 (17.6%) 0.088
Fungal ball on CXR 5 (2.5%) 10 (5%) 0.188 Positive Aspergillus IgG
12 (6%) 27 (13.6%) 0.011
Median Aspergillus IgG level
3.84 mg/L 4.65 mg/L 0.000**
Likely CPA 4 (2%) 8 (4%) 0.38*** Likely simple aspergilloma
1 (0.5%) 1 (0.5%) 1***
Unspecified fungal ball
3 (1.5%) 8 (4%) 0.22***
Raised Aspergillus IgG, but no pulmonary aspergillosis
7 (3.5%) 18 (9%) 0.023
*present for 1 month or longer. ** medians compared by Mann Whitney U test. ***Fishers exact test used. 3952 3953
3954
152
Table 8 – Symptoms and tests for HIV positive patients by CD4 count groups 3955
Results CD4 count < 200 cells/µL n= 23
CD4 count 200 – 499 cells/µL n=94
CD4 count ≥ 500 cells/µL n=74
p-‐value
Cough* 6 (26%) 26 (27.7%) 17 (23%) 0.787 Haemoptysis* 0 3 (3.2%) 2 (2.7%) 0.690 Fatigue* 9 (39.1%) 48 (51.1%) 35 (47.3%) 0.580 Breathlessness* 10 (43.5%) 44 (46.8%) 37 (50%) 0.839 Fevers* 4 (17.4%) 20 (21.3%) 24 (32.4%) 0.168 Night sweats* 5 (21.7%) 26 (27.7%) 20 (27%) 0.845 Chest pain* 10 (43.5%) 49 (52.1%) 36 (48.6%) 0.737 Cavities on CXR 2 (8.7%) 14 (14.9%) 8 (10.8%) 0.611 Paracavitary fibrosis on CXR
1 (4.3%) 7 (7.4%) 6 (8.1%) 0.832
Pleural thickening on CXR
3 (13%) 13 (13.8%) 7 (9.5%) 0.680
Fungal ball on CXR
0 4 (4.3%) 1 (1.4%) 0.355
Positive Aspergillus IgG
0 5 (5.3%) 6 (8.1%) 0.334
Median Aspergillus IgG level
3.77 mg/L 3.98 mg/L 3.65 mg/L 0.809**
Likely CPA 0 2 (2.1%) 2 (2.7%) 0.731 Likely simple aspergilloma
0 1 (1.1%) 0 0.595
Unspecified fungal ball
0 2 (2.1%) 1 (1.4%) 0.749
Raised Aspergillus IgG, but no pulmonary aspergillosis
0 2 (2.1%) 4 (5.4%) 0.315
*present for 1 month or longer. ** medians compared by independent samples median test 3956 3957
153
3958
Table 9 – Antibody levels and diagnoses by year of starting tuberculosis treatment 3959
Symptom 2012 n = 27
2011 n = 89
2010 n = 80
2009 n = 57
2008 n = 58
2007 n = 41
2006 or earlier n = 46
P value
Median Aspergillus IgG level (mg/L)
4 4.4 4.2 4.8 4.1 4.1 4 0.795*
Positive Aspergillus IgG
1 (3.7%)
7 (7.9%)
5 (6.2%)
8 (14%)
10 (17.2%)
2 (4.9%)
5 (10.9%)
0.188
Likely CPA 0 3 (3.4%)
2 (2.5%)
3 (5.3%)
3 (5.2%)
1 (2.4%)
0 0.632
Likely simple aspergilloma
0 0 0 0 1 (1.7%)
0 1 (2.2%)
0.450
Unspecified fungal ball
0 1 (1.1%)
1 (1.2%)
3 (5.2%)
2 (3.4%)
2 (4.9%)
2 (4.3%)
0.564
Raised Aspergillus IgG, but no pulmonary aspergillosis
1 (3.7%)
4 (4.5%)
3 (5%)
5 (8.8%)
6 (10.3%)
1 (2.4%)
4 (8.7%)
0.482
* medians compared by independent samples median test 3960 3961
Discussion 3962
3963
We have demonstrated that raised Aspergillus-‐specific IgG is present in 9.8% of our 3964
cohort of Ugandan adults with previously treated pulmonary tuberculosis, but only 2% 3965
of healthy controls. This difference was statistically significant with a p-‐value of 0.01. 3966
This is first study to measure this in a community-‐based survey. Previous publications 3967
reported higher frequencies of raised Aspergillus-‐specific antibodies in patients with 3968
treated pulmonary tuberculosis 80,146,192,221, but these were conducted in highly selected 3969
groups of patients attending tertiary referral hospitals for follow up due to ongoing 3970
symptomatic illness. Our community-‐based study provides a more accurate 3971
measurement of the overall frequency of this finding in this patient group. 3972
3973
Symptoms were common in the study group, with cough reported by 33% of all patients 3974
and breathlessness by 48% of all patients. However, the presence of raised Aspergillus-‐3975
specific IgG was associated with a statistically significant increase in the presence of 3976
154
both haemoptysis and chest pain. Fatigue, breathlessness and fevers were also more 3977
common in the group with raised Aspergillus-‐specific IgG, but these associations were 3978
non-‐significant. 3979
3980
All patients underwent chest X-‐ray. Radiological abnormalities were less common in the 3981
overall study population than reported symptoms, with cavities and pleural thickening 3982
both noted in around 15% of X-‐rays. There was a statistically significant increase in the 3983
frequency of cavities, pleural thickening and suspected fungal ball in patients with 3984
raised Aspergillus-‐specific IgG compared to those with normal antibody levels. There 3985
was also a non-‐significant increase in the frequency of paracavitary fibrosis associated 3986
with raised Aspergillus-‐specific IgG levels. 3987
3988
We compared levels of Aspergillus-‐specific IgG in different groups of patients with 3989
previously treated pulmonary tuberculosis. There was a statistically significant increase 3990
in the frequency of raised Aspergillus-‐specific IgG levels in HIV negative patients 3991
compared to HIV positive patients. The study was not powered to detect differences 3992
between other patient groups. There was, however a non-‐significant trend towards 3993
increased frequency of Aspergillus-‐specific IgG in HIV positive patients with higher CD4 3994
cell counts compared to those with lower CD4 cell counts. 3995
3996
There was also a statistically significant increase in the number of patients with cavities 3997
on chest X-‐ray in the HIV negative group, compared to the HIV positive group, as has 3998
been observed previously266. The presence of residual cavities after tuberculosis might 3999
result in increased vulnerability to Aspergillus infection, in which case the increased 4000
level of raised Aspergillus-‐specific IgG in the HIV negative group might reflect a 4001
genuinely higher rate of CPA in this group. 4002
4003
An alternative explanation for this finding would be that patients with HIV infection, 4004
especially those with low CD4 counts, might form a less effective antibody response to 4005
active Aspergillus infection than HIV negative patients. However, the presence of raised 4006
Aspergillus-‐specific IgG in 26% of Ugandan in patients with AIDS and sub-‐acute 4007
respiratory disease with no confirmed diagnosis (paper 4), demonstrates that patients 4008
with AIDS can mount an antibody response to Aspergillus. It therefore seems most likely 4009
155
that pulmonary aspergillosis does occur more frequently in HIV negative patients, 4010
perhaps on account of the increased frequency of pulmonary cavitation in this group. 4011
4012
It is important to note that this study only measured antibodies to Aspergillus fumigatus. 4013
This species is responsible for over 90% of aspergillosis in Europe 5,66,108. In such 4014
circumstances measuring A. fumigatus specific IgG gives a reasonably reliable measure 4015
of the overall prevalence of raised Aspergillus-‐specific IgG. However most aspergillosis 4016
in India and the Middle East is due to A. flavus10 and A. niger is common in Brazil147. A. 4017
fumigatus assays can have poor sensitivity for infection with other Aspergillus 4018
species147,148. The sole published study describing the frequency of fungal co-‐infection 4019
in African tuberculosis patients showed two cases of A. niger and two cases of 4020
histoplasmosis213. It is not therefore clear whether A. fumigatus is likely to be the 4021
dominant species of Aspergillus causing disease in humans in Uganda. If other species of 4022
Aspergillus are common in Uganda then our study might significantly underestimate the 4023
total prevalence of Aspergillus-‐specific IgG in the study population. 4024
4025
Histoplasmosis is also known to exist in Uganda268 and blastomycosis elsewhere in 4026
Africa232. These and other chronic fungal lung infections can also cause chronic cough 4027
with progressive lung cavitation and fibrosis290–292. Cross reactivity between other 4028
Aspergillus-‐specific IgG assays and Penicillium antibodies has been noted269 and might 4029
theoretically also occur with other fungi. If this does occur then in some cases of raised 4030
Aspergillus-‐specific IgG found in our study might be due to infection with another 4031
fungus. 4032
4033
It is also possible that some of the 11 (2.8%) of patients with ‘unspecified fungal ball’ on 4034
chest X-‐ray might be suffering from chronic infection with another fungus and that 4035
cross-‐reaction with the Aspergillus-‐specific IgG assay is not occurring in these cases. 4036
The alternative explanation of erroneous reporting of chest X-‐rays must also be 4037
considered. We aim to perform CT scan on all patients with reported ‘fungal ball’ on 4038
chest X-‐ray during the resurvey to confirm that a fungal ball is indeed present in each 4039
case. 4040
4041
156
This study was not designed to measure the precise prevalence of CPA. All the currently 4042
published case cohorts of CPA include CT scan rather than chest X-‐ray in their 4043
diagnostic criteria. They also include progressive lung cavitation and exclusion of 4044
chronic lung conditions as mandatory diagnostic features for CPA. None of these 4045
features are included in this study. We will later undertake a resurvey of this cohort 4046
with repeat chest X-‐ray, CT thorax and exclusion of recurrent pulmonary tuberculosis 4047
with geneXpert sputum PCR testing to permit measurement of the prevalence of CPA in 4048
this cohort. 4049
4050
While this study cannot measure the precise prevalence of CPA, it is the first survey of 4051
its kind in an area of high HIV prevalence and some useful conclusions can be drawn 4052
regarding the prevalence of CPA. 15 (3.8%) of patients had a suspected fungal ball on 4053
chest X-‐ray and 4 (1%) of all patients had a combination of fungal ball on chest X-‐ray 4054
and raised Aspergillus-‐specific IgG. The latter group is highly likely to be suffering from 4055
some form of pulmonary aspergillosis. This finding alone suggests that CPA is likely to 4056
occur with measurable frequency in Ugandan adults with treated pulmonary 4057
tuberculosis. 4058
4059
We have identified ‘likely CPA’ in 12 (3%) patients and simple aspergilloma in 2 (0.5%) 4060
patients. We cannot be certain CPA is present in these patients in the absence of 4061
progressive radiology or CT scan. However, the Siemens Aspergillus-‐specific IgG assay 4062
used has a sensitivity of 96% and specificity of 98% for the diagnosis of CPA (paper 1). 4063
The combination of chronic cough or haemoptysis, plus cavities or fungal ball on chest 4064
X-‐ray and raised Aspergillus-‐specific IgG in these patients is therefore likely to be due to 4065
CPA. Indeed the 98% specificity of the Siemens Aspergillus-‐specific IgG assay suggests 4066
that many of the 25 (6.3%) patients with raised Aspergillus-‐specific IgG in whom the 4067
symptomatic and radiological criteria for CPA have not met in this study will be 4068
confirmed as having CPA in the resurvey. 4069
4070
Our study suffered from further limitations. The convenience sampling method used is 4071
vulnerable to selection bias. Only survivors are recruited. The 5-‐year mortality of CPA is 4072
up to 80%7. By allowing recruitment of patients treated for tuberculosis up to 7 years 4073
ago we might therefore have missed patients who developed CPA soon after completing 4074
157
tuberculosis treatment and subsequently succumbed to the condition. This could be 4075
especially important in the HIV positive group, where pulmonary aspergillosis presents 4076
in subacute invasive form52,223,236. Such patients would normally die within months 4077
without treatment. 4078
4079
Despite these limitations we have demonstrated that there is a statistically significant 4080
increase in the frequency of raised Aspergillus-‐specific IgG in adults with treated 4081
pulmonary tuberculosis, in comparison to healthy controls in Uganda. We cannot 4082
definitively state the frequency of CPA in this study, but we can state with reasonable 4083
confidence that it must lie between 1% (the number of patients with fungal ball on chest 4084
X-‐ray plus raised Aspergillus-‐specific IgG) and 10% (the total number of patients with 4085
raised Aspergillus-‐specific IgG). 4086
4087
We have estimated that 3.5% of all patients with treated pulmonary tuberculosis in this 4088
cohort are suffering from CPA or simple aspergilloma. We are now conducting a 4089
resurvey with repeat chest X-‐ray and CT thorax to confirm the prevalence of CPA. If this 4090
study confirms our estimate of the prevalence of CPA this would constitute evidence of 4091
a life threatening and previously neglected complication of tuberculosis occurring with 4092
clinically relevant frequency. 4093
4094
Hypothesis 4095
4096
That treated pulmonary tuberculosis is complicated by chronic pulmonary aspergillosis 4097
in adults and that an increased frequency of raised Aspergillus-‐specific IgG levels can 4098
therefore be detected in patients with treated pulmonary tuberculosis in comparison to 4099
healthy controls. 4100
4101
Aims 4102
4103
1 – To measure levels of Aspergillus-‐specific IgG in a group of Ugandan adults with 4104
previously treated pulmonary tuberculosis. 4105
4106
158
2 – To determine the frequency of raised levels of Aspergillus-‐specific IgG in this group, 4107
using the diagnostic cut off defined in paper 1. 4108
4109
3 – To record the frequency of various chronic symptoms associated with CPA in this 4110
population using a structured questionnaire. 4111
4112
4 – To perform chest X-‐ray on all patients in this group and record the frequency of 4113
abnormalities associated with CPA. 4114
4115
5 – To determine if raised levels of Aspergillus-‐specific IgG are associated with increased 4116
frequency of symptoms or X-‐ray abnormalities. 4117
4118
6 – To estimate the prevalence of CPA in this population, as accurately as possible 4119
within the limitations of this study design. 4120
4121
7 – To compare the frequency of raised Aspergillus-‐specific IgG, symptoms, X-‐ray 4122
changes and likely CPA secondary to treated pulmonary tuberculosis in patients with 4123
and without HIV infection. 4124
4125
Ethics 4126
4127
Ethical permission for this study was granted by the University of Manchester, UK (ref 4128
11424), Gulu University IRB, Uganda (GU/IRC/04/07/12) and the Ugandan National 4129
Council for Science and Technology (ref HS1253). 4130
4131
Funding 4132
4133
Funding to conduct this study was provided by the Manchester Academy academic 4134
charity. Test kits for use in the study were kindly donated by Siemens (Germany). 4135
4136
Acknowledgements 4137
4138
We are indebted to the following persons and organisations; 4139
159
• Gulu District Health team for their substantial assistance in identifying eligible 4140
patients. 4141
• Study assistants Geoffrey Abwola and Thomas Okumu for their work throughout 4142
the study. 4143
• Gulu Regional Referral Hospital Infectious Diseases Clinic for providing us with 4144
space to review patients and assisting in identifying eligible patients from those 4145
attending clinic. 4146
• The Joint Clinical Research Centre (JCRC) Gulu laboratory for the storage and 4147
cataloging of samples and performing CD4 counts. 4148
• Brother Carlos and the radiology staff at St. Mary’s Hospital, Lacor for their 4149
assistance in performing chest X-‐ray on study patients. 4150
• Andrew Mockridge of Manchester University for his practical assistance in 4151
planning the study in Gulu. 4152
• The North West Lung Centre at University Hospital of South Manchester for 4153
providing storage of serum samples. 4154
• Department of Pathology at Christie Hospital, Manchester for allowing us to 4155
access their Siemens Immulite 2000 to test study samples. 4156
4157
160
4158
PAPER 3 - Prevalence of chronic pulmonary aspergillosis (CPA) secondary to 4159
tuberculosis: a cross-sectional survey in an area of high tuberculosis prevalence. 4160
4161
Authors 4162
4163
Iain D Page – Institute of Inflammation and Repair, The University of Manchester, UK, 4164
Manchester Academic Health Science Centre, UK, National Aspergillosis Center, 4165
University Hospital of South Manchester, UK. 4166
4167
Nathan Onyachi – Gulu Regional Referral Hospital, Uganda. 4168
4169
Cyprian Opira – St. Mary’s Hospital, Lacor, Gulu, Uganda. 4170
4171
Sharath Hosmane – University Hospital of South Manchester, UK 4172
4173
Rosemary Byanyima – Kampala Imaging Centre, Uganda 4174
4175
Richard Sawyer -‐ University Hospital of South Manchester, UK 4176
4177
Malcolm Richardson – Institute of Inflammation and Repair, The University of 4178
Manchester, UK, Manchester Academic Health Science Centre, UK, National Aspergillosis 4179
Center and Mycology Reference Centre, University Hospital of South Manchester, UK. 4180
4181
David W Denning– Institute of Inflammation and Repair, The University of Manchester, 4182
UK, Manchester Academy Health Science Centre, UK, National Aspergillosis Centre, 4183
University Hospital of South Manchester, UK. 4184
4185
4186
161
Abstract 4187
4188
Chronic cavitary pulmonary aspergillosis (CPA) is estimated to complicate 1.2 million 4189
cases of pulmonary tuberculosis worldwide. This includes both chronic cavitary 4190
pulmonary aspergillosis (CCPA) and simple aspergilloma. CPA has a 5-‐year mortality of 4191
50 – 80%, but is treatable. We measured the prevalence of CPA in an area of high 4192
tuberculosis prevalence. 4193
4194
In 2012 to 2013 we surveyed 400 adult patients with treated pulmonary tuberculosis in 4195
Gulu, Uganda. Half also had HIV infection. Between October 2014 and January 2015 we 4196
conducted a re-‐survey. Patients underwent clinical assessment and chest X-‐rays. Those 4197
with raised levels of Aspergillus-‐specific IgG or suspicion of aspergilloma had a CT chest 4198
scan. Those with productive cough submitted sputum for GeneXpert PCR. 4199
4200
CCPA was diagnosed in any patient who did not have recurrent active pulmonary 4201
tuberculosis, but had all the following; 1 – Cough or haemoptysis for one month or 4202
more, 2 – Raised Aspergillus-‐specific IgG, 3 – Progressive cavitation on serial chest X-‐ray 4203
or cavities with paracavitary infiltrates or aspergilloma on CT scan. Simple 4204
aspergilloma was diagnosed in persons with aspergilloma on CT scan and raised 4205
Aspergillus-‐specific IgG, but no chronic cough or haemoptysis. 4206
4207
282 patients were re-‐surveyed. There was no significant difference in patient 4208
characteristics between the surveys. 99 (35%) patients resurveyed had cough and 31 4209
(11%) had haemoptysis. 31 (11%) had progressive cavitation on serial chest X-‐ray. 29 4210
(10%) patients had raised Aspergillus-‐specific IgG. 43 (15%) patients had paracavitary 4211
fibrosis on CT scan and 14 (5%) had aspergilloma. 25 patients underwent GeneXpert 4212
sputum testing and 3 had confirmed active tuberculosis, none of whom had Aspergillus 4213
co-‐infection. 4214
4215
CCPA was present in 16 (6%) patients resurveyed. A further 2 (1%) patients had simple 4216
aspergilloma. CPA was diagnosed in 62% of those with raised Aspergillus-‐specific IgG. 4217
10 of the 12 patients diagnosed with likely CCPA based on chest X-‐ray findings in the 4218
162
first survey were re-‐surveyed. All 10 had CT changes consistent with CCPA. HIV status 4219
had no significant impact on CPA prevalence. 4220
4221
CPA complicates pulmonary tuberculosis with clinically relevant frequency. This survey 4222
supports the estimated global 5-‐year point prevalence of CPA secondary tuberculosis of 4223
1.3 million cases. CPA should be considered for in any patient with a background of 4224
pulmonary tuberculosis who presents with cough, haemoptysis or progressive 4225
cavitation. Access to diagnosis and treatment for CPA is almost non-‐existent in most 4226
areas with high tuberculosis prevalence. Improving this should be an urgent priority in 4227
global health. 4228
4229
4230
4231
163
Introduction 4232
4233
An estimated 9 million people developed tuberculosis in 2013215. It was associated with 4234
1.5 million deaths, of which only 210,000 were estimated to be due to multidrug 4235
resistant strains. Many of the other 1.29 million deaths will have been due to late 4236
presentation to medical care, lack of diagnosis, poor access to treatment or inadequate 4237
adherence, given that they mostly occurred in resource-‐poor countries with weak 4238
health infrastructure. However, misdiagnosis may also have contributed to the problem. 4239
4240
Chronic pulmonary aspergillosis (CPA) is a condition that complicates tuberculosis14. 4241
CPA includes both chronic cavitary pulmonary aspergillosis (CCPA) and simple 4242
aspergilloma. CCPA usually presents with progressive pulmonary cavitation associated 4243
with weight loss, persistent cough and haemoptysis5,7,8. It has a 5-‐year mortality of 50 – 4244
80%6,7,264 and has recently been estimated to affect around 3 million people globally11–4245 13, including 1.3 million cases secondary to tuberculosis11. This estimate takes no 4246
account of the potential impact of HIV co-‐infection, which is present in half of the cases 4247
of suspected pulmonary tuberculosis notified in Uganda215. 4248
4249
Undiagnosed CCPA could therefore be making a substantial contribution to the 4250
observed mortality rates attributed to tuberculosis. Both conditions present with 4251
cavities, pleural thickening and fibrosis on chest X-‐ray266,280. Aspergillomas are 4252
distinctive, but while they are present in all cases of simple aspergilloma, they are 4253
present in only 25-‐36% of cases of CCPA in developed countries8,58. Raised levels of 4254
Aspergillus-‐specific IgG are key to diagnosis of CPA5,7,8, but this test is generally 4255
unavailable in Africa220. In Uganda 34% of all notified cases of pulmonary tuberculosis 4256
are clinically diagnosed with no microbiological proof of tuberculosis infection215. Some 4257
of these cases may well be CCPA misdiagnosed as tuberculosis. 4258
4259
Large CPA case series have been reported in the UK, France, India, China, Korea and 4260
Japan and the majority of cases are secondary to tuberculosis7,8,14,15,18,108,198. Over 180 4261
cases of CPA have been reported throughout Africa, including South Africa, Nigeria, 4262
Ivory Coast, Senegal, Central African Republic, Djibouti, Ethiopia, Tanzania and 4263
Uganda16,201–212. Over 90% of African cases were secondary to pulmonary tuberculosis. 4264
164
The prevalence of CPA was measured in 544 patients with residual lung cavities after 4265
tuberculosis treatment in the UK in 1968-‐7076,197. Precipitating antibodies to Aspergillus 4266
fumigatus were present in 34%, of whom 63% had an aspergilloma visible on chest X-‐4267
ray within 48 months of completion of tuberculosis treatment. Subsequent series have 4268
found positive Aspergillus-‐specific antibodies in 20-‐27% of patients previously treated 4269
for pulmonary tuberculosis in Japan, India and Brazil80,146,192,221. 4270
4271
CPA prevalence in areas where tuberculosis is now common might differ from the UK in 4272
1968-‐70. Rates of Aspergillus rhinitis and keratitis are higher in countries with warm 4273
climates and many subsistence farmers10. This might also be true for CPA. Biomass 4274
smoke-‐induced emphysema is common in Africa222 and might increase CPA risk14. 4275
Crucially HIV co-‐infection might either result in more CPA cases due to 4276
immunosuppression52,223,224 or fewer due to reduced the rate of residual cavitation seen 4277
in those co-‐infected with HIV225–227. 4278
4279
CPA is treatable. Oral treatment with itraconazole, voriconazole or posaconazole 4280
prevents clinical and radiological progression18,58,108,198,251. Surgery is curative in 4281
selected patients with localized disease15,21 and has been safely delivered in resource-‐4282
poor settings16,54,212. 4283
4284
We conducted a cross-‐sectional survey to measure the prevalence of CPA in persons 4285
with treated pulmonary tuberculosis in Gulu, Uganda. We targeted recruitment of 50% 4286
of patients with HIV co-‐infection to measure the impact of this on CPA rates. We 4287
therefore used a case definition6,264 that would capture both CCPA and the subacute 4288
invasive aspergillosis seen in HIV5,6,52,223,236,264. We diagnosed simple aspergilloma in 4289
patients with aspergilloma and raised Aspergillus-‐specific IgG, but no chronic cough or 4290
haemoptysis. Taken together these conditions represent the total prevalence of chronic 4291
pulmonary aspergillosis. 4292
4293
An initial survey was conducted in 2012 (paper 2). It recorded presence of chronic 4294
symptoms, performed chest X-‐ray and measured Aspergillus-‐specific IgG using the 4295
Siemens Immulite 2000 system. This assay has specificity of 98% and sensitivity of 96% 4296
for the diagnosis of CPA (paper 1). 398 patients were assessed. 39 (9.8%) were found 4297
165
to have raised levels of Aspergillus-‐specific IgG and 15 (3.8%) had a suspected fungal 4298
ball on chest X-‐ray. Within the limits of this single survey it was estimated that 12 (3%) 4299
of patients were likely to be suffering from CPA as they had a combination of chronic 4300
cough or haemoptysis, plus cavitation or fungal ball on chest X-‐ray and had raised levels 4301
of Aspergillus-‐specific IgG. A further 2 (0.5%) asymptomatic patients were suspected to 4302
be suffering from simple aspergilloma as they had a combination of fungal ball on chest 4303
X-‐ray and raised levels of Aspergillus-‐specific IgG. However this first survey was limited 4304
in its ability to accurately identify cases of CPA due to the lack of CT scans and lack of 4305
exclusion of recurrent tuberculosis. It was also impossible to identify progressive 4306
cavitation on the basis of a single survey. 4307
4308
A resurvey was therefore conducted in 2014 to allow accurate measurement of the 4309
prevalence of CPA. This included repeated clinical assessment, serology and chest X-‐ray. 4310
Patients with raised Aspergillus-‐specific IgG or suspicion of fungal ball on chest X-‐ray in 4311
2012 also underwent CT chest scan. The impact of potential risk factors, including HIV 4312
co-‐infection, on the frequency of CPA was assessed. 4313
4314
Methods 4315
4316
Study design and participants 4317
4318
398 patients were recruited in 2012 as described in paper 2. Patients enrolled in the 4319
first survey were traced by District Health Tuberculosis Team staff and re-‐assessed 4320
between October 2014 and January 2015. All patients underwent repeat clinical 4321
examination and chest X-‐ray, which was reported as before. Repeat Immulite 4322
Aspergillus-‐specific IgG were measured on serum. Patient flow and recruitment 4323
outcomes are shown in figure 1. 4324
4325
CT scan (GE Duo-‐slice, USA) was performed at the Kampala Imaging Center on those 4326
with raised Aspergillus-‐specific IgG or suspicion of aspergilloma on 2012 chest X-‐ray. 4327
Digital CT scan images were saved and accessed with OsirisX software (Pixmeo SARL, 4328
Switzerland). Patient flow for those selected for CT scan is shown in figure 2. Reports 4329
were provided by three radiologists, in the same manner as chest X-‐rays. Verbal 4330
166
autopsy was performed by district health workers in those that died between 4331
surveys293. 4332
4333
Sputum was taken from all patients who were able to provide a sample and underwent 4334
GeneXpert IV (Cepheid, USA) Mycobacterium tuberculosis nucleic acid amplification 4335
testing. Patient flow for patients with productive cough is shown in figure 3. 4336
4337
Diagnostic criteria 4338
4339
CCPA was diagnosed when ALL four of the following criteria were met:-‐ 4340
4341
1 – Symptoms -‐ patients must have been suffering from at least one of the following 4342
symptoms for no less than 1 month. 4343
• Haemoptysis 4344
• Cough 4345
4346
2 – Radiological changes – at least one of the following features must be present 4347
• Fungal ball on CT scan 4348
• Cavitation with paracavitary fibrosis on CT scan 4349
• New or progressive cavitation on serial chest X-‐ray 4350
4351
3 – Raised Aspergillus-‐specific IgG 4352
4353
4 – Absence of positive GeneXpert test for M. tuberculosis 4354
4355
In addition, simple aspergilloma was diagnosed in patients with fungal ball on CT scan 4356
and raised Aspergillus-‐specific IgG, but no chronic cough or haemoptysis. 100 control 4357
sera had previously been collected from healthy Ugandan blood donors (paper 1). These 4358
were used in receiver operating characteristic curve studies to define the diagnostic 4359
threshold of 10 mg/L used in this study. 4360
4361
Chest X-‐rays and CT scans were both reported by the author on the day of testing and 4362
patients were informed of their diagnosis. Where CPA was diagnosed patients were 4363
167
provided with a written statement of the diagnosis together with an advisory treatment 4364
plan, with advice to attend the Gulu Regional Referral Hospital Infectious Diseases clinic 4365
for follow up. Patients with resectable disease were referred to the cardiothoracic 4366
surgical unit at Mulago Hospital, Kampala for surgical treatment. Where surgery was 4367
not appropriate treatment with oral itraconazole was recommended. 4368
4369
Statistical methods 4370
4371
Statistical analysis was performed using SPSS v20 (IBM, USA). Rates of CPA in groups of 4372
patients with and without potential risk factors were compared using chi-‐squared test, 4373
except for comparisons with less than 5 patients in one group, where Fisher’s exact test 4374
was used. Comparison of means for continuous variables in different groups with 4375
normal distribution was performed using 2-‐sided t test. Where distribution was skewed 4376
Mann Whitney U test was used. 4377
4378
Results 4379
4380
Patient characteristics for the original survey and re-‐survey are compared in table 1. 4381
There is no evidence that the resurvey recruitment process introduced bias, as there are 4382
no significant differences in characteristics between the groups. 4383
4384
389 patients were recruited in 2012. 282 of these patients were reviewed in the 4385
resurvey. Recruitment outcomes for all patients from the 2012 survey are shown in 4386
figure 1. 29 of these patients had raised Aspergillus-‐specific IgG in 2012 and were 4387
eligible for CT scan. A further 45 patients in the re-‐survey group had a suspicion of 4388
aspergilloma on their 2012 chest X-‐ray and also underwent CT scan, with 73 persons 4389
undergoing CT scan in total. CT scan outcomes are shown in figure 2. 4390
4391
All patients who could provide a sputum sample underwent GeneXpert PCR testing for 4392
recurrent tuberculosis. In two cases the GeneXpert machine was not functioning and 4393
smear test was performed in its place. Two cases of active pulmonary tuberculosis were 4394
identified. A third patient was diagnosed with multi-‐drug resistant tuberculosis 4395
168
between the original study and the resurvey. None of the three patients had evidence of 4396
additional CPA. A breakdown of GeneXpert test outcomes is shown in figure 3. 4397
4398
Table 1 – Patient characteristics 4399
4400
Characteristic in 2012 Original survey n=398
Re-‐survey n= 282
p-‐value*
Female gender 155 (38.9%) 99 (35.1%) 0.308 Mean age in years (range)
42 (16-‐83) 42 (16-‐82) 0.53**
Positive sputum smear at TB diagnosis
303 (76.1%) 222 (78.7%) 0.427
HIV infection 199 (50%) 134 (47.5%) 0.524 Median 2012 CD4 count in HIV positive persons (range)
424 (14 – 1400) cells/µL
424 (59 -‐ 1400) cells/µL
0.867***
2012 CD4 count < 200 cells/µL
23 (12%#) 15 (11.6%##) 0.911
2012 CD4 count 200 – 499 cells/µL
94 (49.2%#) 65 (50.4%##) 0.837
2012 CD4 count ≥ 500 cells/µL
74 (38.7%#) 49 (38%##) 0.891
Traditional ‘grass-‐thatch’ home
371 (93.2%) 267 (94.7%) 0.434
Patient reports dampness in home
119 (29.9%) 76 (27%) 0.402
Patient is a subsistence farmer
373 (93.7%) 267 (94.7%) 0.599
Patient frequently cooks on open charcoal stove
194 (48.7%) 127 (45%) 0.34
Patient smokes tobacco 39 (9.8%) 30 (10.6%) 0.721 2012 Aspergillus IgG positive
39 (9.8%) 29 (10.3%) 0.836
2012 median Aspergillus IgG
4.2 mg/L 4.1 mg/L 0.840
*Chi-‐squared test, except **where means compared by 2-‐sided t-‐test ***Mann-‐Whitney U test #fraction of 4401 those with CD4 count available in original survey n=191 ##fraction of those with CD4 count available in 4402 re-‐survey n=129. 4403 4404
Symptoms, chest X-‐ray results and Aspergillus-‐specific IgG results for re-‐survey patients 4405
are shown in table 2. Breakdowns of these results by gender, original tuberculosis 4406
smear status, HIV status and CD4 count are shown in supplementary data. CT scan 4407
findings for those who underwent that test are shown in table 3. The median level of 4408
169
Aspergillus-‐specific IgG in those who were resurveyed was 4.1 mg/L (range 0.1-‐1060 4409
mg/L). 4410
4411
Table 2 – Resurvey symptoms and test findings n=282 4412
4413
Result No patients
Frequency (%)
Cough* 99 35 Haemoptysis* 31 11 Fatigue* 88 31 Breathlessness* 83 29 Fevers* 26 9 Night sweats* 51 18 Chest pain* 76 27 New cavitation on serial CXR
16 6
Enlarged cavitation on serial CXR
14 5
New or progressive paracavitary fibrosis
15 5
New or progressive pleural thickening on CXR
4 1
*present for 1 month or longer 4414 4415
Table 3 – Chest CT scan findings (n=73) 4416
4417
Result No patients
Frequency (%)
Cavities 49 67 Paracavitary fibrosis 43 59 Aspergilloma 14 19 Pleural thickening 35 47 Nodule 37 51 4418
Frequencies of diagnoses are shown in table 4. We identified 16 cases of CCPA, of which 4419
2 had been complicated by the development of chronic fibrosing pulmonary 4420
aspergillosis (CFPA)5. A further five patients had apparent fungal ball on CT scan, but 4421
normal levels of Aspergillus-‐specific IgG and are recorded as ‘unspecified fungal ball’. 4422
170
The frequency of CCPA in patients with and without potential risk factors is shown in 4423
tables 5 and 6. 4424
4425
Table 4 – Frequency of diagnoses 4426
4427
Condition Number of cases N=282
Frequency (%) Frequency 95% confidence interval (%)
CCPA 16 5.7 3.4 – 8.8 Simple aspergilloma 2 0.7 0.1 – 2.3 All CPA 18 6.4 4 – 9.4 Unspecified fungal ball
5 1.8 0.7 – 3.8
Raised Aspergillus IgG, but no pulmonary aspergillosis
11 3.9 2.1 – 6.6
4428
Twelve patients were identified as likely cases of CCPA after the first survey on the basis 4429
of chronic cough or haemoptysis, plus cavitation on chest X-‐ray and raised Aspergillus-‐4430
specific IgG. Ten of these patients were reviewed in the resurvey. All ten had 4431
radiological features of CCPA confirmed on CT scan, although two reported that their 4432
symptoms had resolved at the re-‐survey assessment and so were not classified as CCPA 4433
in the final analysis. The sole patient with likely simple aspergilloma in 2012 and was 4434
resurveyed developed symptoms between the surveys and was confirmed as a case of 4435
CPA in 2014. 4436
4437
Pulmonary aspergillosis was confirmed in 58% of those with raised Aspergillus-‐specific 4438
IgG in 2012. Four (36%) of those with fungal ball on chest X-‐ray in 2012 had a 4439
confirmed fungal ball on resurvey in 2014, but 2 could not be classified as CPA as they 4440
had normal levels of Aspergillus-‐specific IgG. The outcomes of all patients from 2012 4441
who were re-‐surveyed in 2014 are shown in table 7. 4442
4443
Nineteen (5%) patients recruited to the first survey died before the resurvey. None of 4444
these were identified as likely CPA in the first survey. Three of the 19 had raised 4445
Aspergillus-‐specific IgG, but none of these died in a manner consistent with CPA. Four of 4446
the 16 deaths in patients with normal levels of Aspergillus-‐specific IgG were preceded 4447
171
by subacute respiratory illness, including three preceded by haemoptysis. These 4448
patients may have developed pulmonary aspergillosis after the first survey, but this 4449
cannot be confirmed. 4450
4451
Table 5 – Frequency of CCPA in categorical patient groups 4452
4453
Factor
CCPA rate where factor is present
CCPA rate where factor is absent
p-‐value
All patients n=282
16 (5.7%) -‐ -‐
Female gender n=99
5 (5.1%) 11 (6%) 0.739
HIV infection n=134
5 (3.7%) 11 (7.4%) 0.18
Positive smear status at TB diagnosis n=222
13 (5.9%) 3 (5%) 0.799
Subsistence farmer n=267
16 (6%) 0 0.329
Traditional dwelling n=267
16 (6%) 0 0.329
Dampness reported in house n=76
3 (3.9%) 13 (6.3%) 0.447
Frequently cooks on charcoal n=127
7 (5.5%) 9 (5.8%) 0.915
Tobacco smoker n=30
2 (6.7%) 14 (5.6%) 0.804
4454
Table 6 – Frequency of CCPA in relation to continuous patient variables 4455
4456
Factor Mean level in those with CCPA
Mean level in those without CCPA
p-‐value
Age in 2012 43.5 years 42.5 years 0.740 TB treatment start date
28/6/2009 23/7/2009 0.890
CD4 count in HIV positive patients
554 cells/µL 472 cells/µL 0.495
Means compared by 2-‐tailed t-‐test as each variable had a reasonably normal distribution in the whole re-‐4457 survey population. 4458 4459
4460
172
Table 7 – 2014 outcomes in relation to 2012 survey diagnosis 4461 4462
2012 diagnosis
CCPA Simple aspergilloma
Unspecified fungal ball
No aspergillosis
Died before resurvey
Normal Aspergillus-‐specific IgG n= 258
0 0 5 (2%)
248 (96%)
16 (6%)
All raised Aspergillus specific IgG n=31
16 (52%)
2 (6%)
0 11 (35%)
2 (6%)
Fungal ball on CXR n=11
2 (18%)
0 2 (18%)
5 (45%)
2 (18%)
Likely CPA n=10
8 (80%)
0 0 2 (20%)
0
Likely simple aspergilloma n=1
1 (100%)
0 0 0 0
Unspecified fungal ball n=9
0 0 2 (22%)
5 (56%)
2 (22%)
Raised Aspergillus IgG, but no pulmonary aspergillosis n=20
7 (35%)
2 (10%)
0 9 (45%)
2 (10%)
4463
Discussion 4464
4465
We have demonstrated that CPA is present in 6.4% (95% confidence interval 4 – 9.4%) 4466
of patients in our cohort of Ugandan adults with treated pulmonary tuberculosis. If our 4467
findings are applied to the 5.28 million cases of pulmonary tuberculosis estimated to 4468
occur globally215, then they would be consistent with the predicted global 5-‐year point 4469
prevalence of between 0.85 and 1.37 million cases of CPA secondary to pulmonary 4470
tuberculosis11. Further cases of CPA will occur secondary to other conditions14. It is now 4471
clear that CPA is sufficiently common to be considered a global public health problem. 4472
4473
173
The marked similarity between the radiological and clinical presentation of CCPA and 4474
tuberculosis, plus the near total absence of Aspergillus-‐specific IgG testing in Africa220 4475
means that most of these cases are probably being diagnosed as ‘smear negative 4476
tuberculosis’ at present215. Correctly identifying these misdiagnosed cases could have a 4477
significant impact on the apparent global prevalence of smear negative tuberculosis. 4478
Providing treatment with surgery or cheap generic itraconazole could extend many 4479
lives. 4480
4481
The diagnosis of ‘unspecified fungal ball’ refers to cases with apparent fungal ball on CT 4482
scan, but negative Aspergillus-‐specific IgG. These are most likely to represent CPA 4483
caused by other species of Aspergillus not detected by the A. fumigatus assay147,294. 4484
Three of the 11 patients with ‘raised Aspergillus-‐specific IgG, but no pulmonary 4485
aspergillosis’ had symptoms of chronic cough or haemoptysis, but lacked radiological 4486
evidence of CCPA. These may be cases of Aspergillus bronchitis, but this diagnosis 4487
cannot be confirmed without fungal culture39. The other eight did not have cough or 4488
haemoptysis. These cases may represent Aspergillus rhinosinusitis41 or simply 4489
colonization. It is not known whether these Aspergillus-‐associated conditions are likely 4490
to progress to CPA with time. Two (2%) of healthy controls (age 17-‐39 years) also had 4491
raised Aspergillus-‐specific IgG (paper 1). 4492
4493
No risk factors for the development of CPA were identified. There was a non-‐significant 4494
reduction in CPA prevalence in the HIV infected group. However cough was significantly 4495
less common in the HIV infected group (p=0.006). The rate of cavitation and 4496
paracavitary fibrosis in the HIV positive and negative groups was similar, but there was 4497
a statistically significant reduction in the frequency of paracavitary fibrosis in those 4498
with low CD4 count (p=0.028). CPA is associated with several forms of immune 4499
dysfunction14,50. It is conceivable that AIDS-‐associated immunosuppression would be 4500
protective against the immune dysregulation that leads to fibrosis in CPA. 4501
4502
It is notable that progressive cavitation was noted in 9% of male patients resurveyed 4503
but was not noted in any female patients resurveyed. In all female cases with CPA 4504
radiological features of CPA were identified on CT scan. This difference might be 4505
explained by overlying breast tissue obscuring subtle chest X-‐ray findings. Progressive 4506
174
cavitation on chest X-‐ray provided the sole evidence of CPA in one case, with CT scan 4507
findings diagnostic in the other 15 cases. Overall 8 of 16 (50%) CPA cases showed 4508
progressive cavitation on serial chest X-‐ray. These were all in patients with prior smear 4509
positive tuberculosis. This shows that CT scan is more sensitive than serial chest X-‐ray 4510
for the diagnosis of CPA and should be used wherever it is available, however use of 4511
serial chest X-‐ray would allow many cases to be diagnosed in areas where CT scan is not 4512
available. 4513
4514
We have shown that the screening of patients with clinical assessment, chest X-‐ray and 4515
measurement of Aspergillus-‐specific IgG levels undertaken in the first survey in 2012 4516
has some ability to predict the diagnosis of CPA. 80% of the patients diagnosed as ‘likely 4517
CPA’ in 2012 and then reviewed in 2014 had confirmed CPA on re-‐survey with CT scan. 4518
The remaining 20% had signs consistent CPA on CT scan but could not be classified as 4519
CPA cases as they reported resolution of symptoms in between surveys. If the CT scan 4520
had been undertaken immediately in 2012 and produced the same results then 100% of 4521
patients with ‘likely CPA’ 2012 would have been confirmed as cases of CPA. 4522
4523
This suggest that when a patient with treated pulmonary tuberculosis has a 4524
combination of cough or haemoptysis for 1 month or more, with raised levels of 4525
Aspergillus-‐specific IgG and either cavities or fungal ball on chest X-‐ray they are likely to 4526
have CPA and should undergo further assessment with CT scan wherever possible. 4527
However it should be noted that 56% of the patients ultimately diagnosed with CPA 4528
were not identified as ‘likely CPA’ in the 2012 survey, demonstrating that chest X-‐ray 4529
alone is insufficient to exclude CPA. 4530
4531
Measurement of Aspergillus-‐specific IgG alone performed reasonably well as a screening 4532
test for the diagnosis of CPA in this population, with 58% of those with raised levels of 4533
Aspergillus-‐specific IgG having confirmed CPA after the re-‐survey. Raised levels of 4534
Aspergillus-‐specific IgG can also occur in Aspergillus bronchitis39 or colonization38. The 4535
remaining patients with raised levels of Aspergillus-‐specific IgG who did not meet the 4536
diagnostic criteria for CPA may be suffering from one of these conditions, but this could 4537
be confirmed during our study, as facilities for fungal culture were not available at the 4538
study site. 4539
175
Our study was not without flaws. The convenience sampling method used is vulnerable 4540
to selection bias and only identifies those living with CPA. The 5-‐year mortality of CPA is 4541
up to 80%7. By allowing recruitment of patients treated for tuberculosis up to 7 years 4542
ago we might therefore have missed patients who developed CPA soon after completing 4543
tuberculosis treatment and subsequently succumbed to the condition. This could be 4544
especially important in the HIV positive group, where pulmonary aspergillosis presents 4545
in subacute invasive form52,223,236. Such patients would normally die within months 4546
without treatment. 4547
4548
Our case definition necessarily differed from previous CCPA cohorts5,7,8. We did not 4549
include weight loss as a symptom of CCPA as no records of prior weight were available. 4550
We did not include inflammatory markers, as these tests were not available in Gulu. We 4551
excluded pulmonary tuberculosis by GeneXpert PCR testing in those who provided a 4552
sputum sample, but we did not have access to induced sputum or bronchoscopy. We 4553
were also unable to include culture for atypical mycobacteria or testing for alternative 4554
fungal infections that might mimic CCPA. 4555
4556
We measured antibodies to Aspergillus fumigatus, which is responsible for most of the 4557
CPA in Europe and East Asia5–8,108. However most aspergillosis in India and the Middle 4558
East is due to A. flavus10 and A. niger is common in Brazil147. A. fumigatus assays can 4559
have poor sensitivity for other species147,148. The sole study of the frequency of fungal 4560
co-‐infection in African tuberculosis patients showed two cases of A. niger and two cases 4561
of histoplasmosis213. It is therefore not clear if A. fumigatus is the dominant species of 4562
Aspergillus in Ugandan patients. Our study may not have identified cases of CPA due to 4563
Aspergillus species other than A. fumigatus. 4564
4565
Histoplasmosis is known to exist in Uganda268 and blastomycosis elsewhere in Africa232. 4566
These and other chronic fungal lung infections can also cause chronic cough with 4567
progressive lung cavitation and fibrosis290–292. Cross reactivity between other 4568
Aspergillus-‐specific IgG assays and Penicillium antibodies has been noted269. If such 4569
cross-‐reactivity occurred between the Siemens Immulite assay used and other fungal 4570
infections potentially present in Uganda then these might be falsely classified as CPA. 4571
4572
176
Finally it should be noted that CPA also complicates other conditions such as 4573
sarcoidosis, COPD, prior pneumothorax, non-‐tuberculous mycobacterial infection and 4574
ABPA13,14. CPA secondary to these conditions is not measured in this survey and as a 4575
result the prevalence of CPA will be underestimated if based on our findings alone. 4576
4577
CPA often requires treatment with long term anti fungal drugs and surgery in selected 4578
cases15,198. Provision of such long term follow up and treatment was not feasible within 4579
the limited time frame of this cross sectional study. The optimal treatment of CPA has 4580
not been clearly established in the medical literature. Surgery is effective in selected 4581
cases with localized disease15. Some specialist units in the developed world advocate 4582
long-‐term azole therapy198, but six months of generic itraconazole at fixed dose with no 4583
drug level monitoring has recently demonstrated efficacy in a randomized controlled 4584
trial in India18. 4585
4586
The study team made every effort to maximize the opportunities for patients diagnosed 4587
with CPA to access such treatment as was available in Uganda at the time of the study. 4588
We provided presentations on CPA and its management to the Gulu District Health 4589
Team, the staff of the local Gulu Regional Referral Hospital Infectious Diseases Clinic 4590
and the staff of the national referral clinic at the Infectious Diseases Institute in Kampala 4591
to raise awareness of CPA amoungst potential care providers. 4592
4593
We agreed a patient treatment pathway with the management of Gulu Regional Referral 4594
Hospital and the cardiothoracic surgical team at Mulago Hospital, Kampala. Where 4595
patients had localized disease amenable to surgery they would be referred to the 4596
cardiothoracic team for resection. This team already performs 10-‐20 resections a year 4597
for patients with aspergilloma and severe haemoptysis. Where surgery was 4598
inappropriate due to extensive or multifocal disease patients were to be recommended 4599
treatment with itraconazole with basic monitoring provided at the Gulu Regional 4600
Referral Hospital Infectious Diseases Clinic. We established the itraconazole could 4601
normally be purchased at local pharmacies in Gulu at a cost of around US$20 / patient / 4602
month, which would be affordable to patients with a modest income. 4603
4604
177
While the results of the Indian RCT18 and surgical case series from China15 and 4605
Senegal16 demonstrate that the potential does exist to deliver safe and effective CPA 4606
treatment in resource poor settings, our study uncovered a number of practical 4607
difficulties. Where patients were candidates for surgical resection there was the 4608
potential to cure them within the constraints of the existing Ugandan health service. 4609
Unfortunately cost of transport to the cardiothoracic surgery centre in Kampala is a 4610
barrier for many patients. At the time of thesis submission the clinical director of Gulu 4611
Regional Referral Hospital is trying to make arrangements to transport all the surgical 4612
resection candidates to the cardiothoracic surgery department using hospital transport. 4613
4614
Clearly we have not been able to provide the same level of care to patients diagnosed as 4615
CPA in the course of this study as would have been provided at a referral centre in a 4616
well resourced country. We have, however provided patients enrolled in our study with 4617
a unique opportunity to be diagnosed with a serious illness and have made every effort 4618
to maximize the opportunities for our patients to access the treatment options that are 4619
available locally. This includes the possibility of surgical cure of an otherwise fatal 4620
disease for some of our patients. Entry to our study has therefore provided a potential 4621
benefit to those who chose to enter. 4622
4623
When designing the study we expected that most of our recruits would be urban 4624
dwellers with a basic income, most of whom could afford the $20/month cost of 4625
itraconazole. However, most patients actually recruited to this study were subsistence 4626
farmers who could not afford this cost. Itraconazole is not on the World Health 4627
Organization’s essential drugs list and is not provided for free by the Ugandan health 4628
service. As itraconazole is off-‐patent it should be possible to manufacture this drug 4629
sufficiently cheaply to allow national health programs in resource poor countries to 4630
purchase it and provide it at low or no cost to the poorest patients, but this a significant 4631
policy shift at global and national levels would be required to achieve this. 4632
4633
Our epidemiological study is the first to measure the prevalence of CPA in an area of 4634
currently high tuberculosis prevalence. Although it has some limitations it does 4635
demonstrate that CPA is a frequent complication of pulmonary tuberculosis in this 4636
setting. Pulmonary aspergillosis has already been shown to be a frequent complication 4637
178
of pulmonary tuberculosis in the UK76,197. It should be considered in the differential 4638
diagnosis of any patient with a prior history of tuberculosis, who later presents with 4639
chronic cough, haemoptysis, progressive lung cavitation or fibrosis. 4640
4641
Further research is now also needed to develop and validate Aspergillus serology tests 4642
suitable for large-‐scale use in resource poor settings. The Siemens Immulite assay is 4643
performed on a large automated system that requires frequent maintenance. It is not 4644
ideal for use in resource-‐poor settings. A simpler, cheaper test such as a lateral flow 4645
device (LFD) is required for large scale testing295. Such an LFD has recently been 4646
developed to detect Aspergillus antigens and has good performance characteristics for 4647
the diagnosis of invasive aspergillosis, even in patients with limited 4648
immunosuppresion187–190. Its sensitivity and specificity for the diagnosis of CPA has not 4649
been investigated, but the Platelia galactomannan antigen ELISA has poor sensitivity in 4650
serum in this context8,59. 4651
4652
The efficacy of oral antifungal treatment for CPA and the suitability of these drugs for 4653
use without regular monitoring must be confirmed in further trials. Other chronic 4654
fungal infections may well be present in resource poor countries and the frequency of 4655
these needs to be measured. Such information is a necessary pre-‐requisite for the 4656
design of new health care protocols that might allow routine diagnosis and effective 4657
treatment of chronic fungal lung diseases complicating pulmonary tuberculosis in 4658
resource poor settings. 4659
4660
We are now planning a larger, prospective multi-‐center study in Kenya that should 4661
provide a more accurate measure of the prevalence of CPA and the timing at which it 4662
occurs after pulmonary tuberculosis. This will provide a key piece of evidence to inform 4663
the design of healthcare guidelines to allow identification of those patients developing 4664
CPA after tuberculosis. 4665
4666
Hypothesis 4667
4668
That chronic pulmonary aspergillosis (CPA) complicates pulmonary tuberculosis at a 4669
frequency that is clinically significant and sufficiently large to be measured. 4670
179
Aims 4671
4672
1 – To complete a cross-‐sectional survey to measure the prevalence of CPA secondary to 4673
pulmonary tuberculosis in an area of high tuberculosis prevalence. 4674
4675
2 – To determine whether HIV co-‐infection is associated with an altered frequency of 4676
CPA secondary to pulmonary tuberculosis. 4677
4678
3 – To determine whether other postulated environmental and clinical risk factors for 4679
development of CPA are associated with altered frequency of CPA secondary to 4680
pulmonary aspergillosis. 4681
4682
Ethics 4683
4684
Ethical permission for this study was granted by the University of Manchester, UK (ref 4685
11424), Gulu University IRB, Uganda (GU/IRC/04/07/12) and the Ugandan National 4686
Council for Science and Technology (ref HS1253). 4687
4688
Funding 4689
4690
Funding to conduct this study was provided by the Manchester Academy academic 4691
charity and Astellas Pharma. Test kits for use in the study were kindly donated by 4692
Siemens (Germany). 4693
4694
Acknowledgements 4695
4696
We are indebted to the following persons and organisations; 4697
• Gulu District Health Team for their substantial assistance in identifying eligible 4698
patients. 4699
• Study assistants Geoffrey Abwola and Thomas Okumu for their work throughout 4700
the study. 4701
180
• Gulu Regional Referral Hospital Infectious Diseases Clinic for providing us with 4702
space to review patients and assisting in identifying eligible patients from those 4703
attending clinic. 4704
• Gulu Regional Referral Hospital Pathology Laboratory for providing sample 4705
storage and Cepheid GeneXpert testing on sputum samples. 4706
• Brother Carlos and the radiology staff at St. Mary’s Hospital, Lacor for their 4707
assistance in performing chest X-‐ray on study patients. 4708
• Matthew Kneale and Michael Clarke of Manchester University for their practical 4709
assistance in conducting the study in Gulu. 4710
• The North West Lung Centre at University Hospital of South Manchester for 4711
providing storage of serum samples. 4712
• Department of Pathology at Christie Hospital, Manchester for allowing us to 4713
access their Siemens Immulite 2000 to test study samples. 4714
4715
4716
181
Figures 4717
4718
Figure 1 – Recruitment outcomes 4719
4720
Patients recruited in original survey (398) 4721
4722
è Died between surveys (18) 4723
4724
è Moved out of the region (9) 4725
4726
è Declined to participate in the re-‐survey (11) 4727
4728
è Could not be traced (75) 4729
4730
ê 4731 4732
Living patients traced and consented to enter the re-‐survey (285) 4733
4734
ì CXR not performed as planned (2) 4735
èPatients removed due to incomplete assessment 4736
î Serum sample not taken as planned (1) 4737
4738 ê 4739
4740
Patients in final re-‐survey analysis (282) 4741
4742
182
Figure 2 – CT scan recruitment outcomes 4743
4744
Patients with raised Aspergillus IgG levels in original survey (39) 4745
4746
è Died between surveys (2) 4747
4748
è Could not be traced (8) 4749
4750
ê 4751
4752
Consented for CT scan (29) 4753
4754
è Died between consent and CT scan date (1) 4755
4756
ê 4757
4758
Underwent CT scan (28) Resurvey patients with suspicion of 4759
aspergilloma on 2012 chest X-‐ray, 4760
but negative Aspergillus-‐specific IgG (45) 4761
î í 4762
4763
Total number of patients undergoing CT scan (73) 4764
4765
4766
183
Figure 3 – GeneXpert testing outcomes for resurvey patients 4767
4768
Patients reporting productive cough (39) 4769
4770
è No sputum sample submitted (8) 4771
4772
è GeneXpert testing attempted, but failed (6) 4773
4774
ê 4775
4776
Samples with GeneXpert result available (25) 4777
4778
184
Supplementary data 4779
4780
S1 -‐ Symptoms and test results in resurvey by gender 4781
4782
Result Female n=99 Male n=183 p-‐value by chi-‐squared test
Cough* 31 (31.3%) 68 (37.2%) 0.326 Productive cough* 9 (9.1%) 25 (13.7%) 0.261 Haemoptysis* 11 (11.1%) 20 (10.9%) 0.963 Fatigue* 34 (34.3%) 54 (29.5%) 0.403 Breathlessness* 33 (33.3%) 50 (27.3%) 0.290 Fevers* 8 (8.1%) 18 (9.8%) 0.627 Night sweats* 15 (15.2%) 36 (19.7%) 0.346 Chest pain* 23 (23.2%) 53 (29%) 0.301 New cavitation on serial CXR
0 16 (8.7%) 0.002
Enlarged cavitation on serial CXR
5 (5.1%) 9 (4.9%) 0.961
New or progressive paracavitary fibrosis
4 (4%) 11 (6%) 0.482
New or progressive pleural thickening on CXR
1 (1%) 3 (1.6%) 0.67
2012 median Aspergillus IgG level
3.77 mg/L 4.41 mg/L 0.016**
2012 positive Aspergillus IgG
7 (7.1%) 22 (12%) 0.191
*present for 1 month or longer. ** medians compared by Mann Whitney U 4783 4784
S2 – CT findings by gender 4785
4786
Result Female n=16
Male n=57
p-‐value by chi squared test
Cavities 11 (68%) 38 (66.7%) 0.875 Paracavitary fibrosis 9 (56.3%) 34 (59.6%) 0.807 Aspergilloma 2 (12.5%) 12 (21.1%) 0.443 Pleural thickening 5 (31.3%) 30 (52.6%) 0.130 Nodule 4 (25%) 33 (57.9%) 0.02 4787
4788
185
S3 -‐ Symptoms and test results in resurvey by HIV status 4789 4790
Result HIV positive n=134
HIV negative n=148
p-‐value by chi-‐squared test
Cough* 36 (26.9%) 63 (42.6%) 0.006 Productive cough* 13 (9.7%) 21 (14.2%) 0.248 Haemoptysis* 17 (12.7%) 14 (9.5%) 0.387 Fatigue* 39 (29.1%) 49 (33.1%) 0.469 Breathlessness* 38 (28.4%) 45 (30.4%) 0.706 Fevers* 12 (9%) 14 (9.5%) 0.884 Night sweats* 22 (16.4%) 29 (19.6%) 0.489 Chest pain* 31 (23.1%) 45 (30.4%) 0.169 New cavitation on serial CXR
7 (5.2%) 9 (6.1%) 0.756
Enlarged cavitation on serial CXR
5 (3.7%) 9 (6.1%) 0.364
New or progressive paracavitary fibrosis
8 (6%) 7 (4.7%) 0.643
New or progressive pleural thickening on CXR
2 (1.5%) 2 (1.4%) 0.92
2012 median Aspergillus IgG level
3.77 mg/L 4.63 mg/L 0.001**
2012 positive Aspergillus IgG
10 (7.5%) 19 (12.8%) 0.138
*present for 1 month or longer. ** medians compared by Mann Whitney U 4791 4792
S4 – CT findings by HIV status 4793
4794
Result HIV positive n=26
HIV negative n=47
p-‐value by chi squared test
Cavities 17 (65.4%) 32 (68.1%) 0.814 Paracavitary fibrosis 13 (50%) 30 (63.8%) 0.250 Aspergilloma 5 (19.2%) 9 (19.1%) 0.993 Pleural thickening 9 (34.6%) 26 (55.3%) 0.09 Nodule 11 (42.3%) 26 (55.3%) 0.287 4795 4796
4797
186
S5 -‐ Symptoms and test results in resurvey by prior TB smear status 4798 4799
Result Smear positive n=222
Smear negative n=60
p-‐value by chi-‐squared test
Cough* 80 (36%) 19 (31.7%) 0.529 Productive cough* 31 (14%) 3 (5%) 0.058 Haemoptysis* 25 (11.3%) 6 (10%) 0.782 Fatigue* 67 (30.2%) 21 (35%) 0.475 Breathlessness* 63 (28.4%) 20 (33.3%) 0.455 Fevers* 22 (9.9%) 4 (6.7%) 0.441 Night sweats* 40 (18%) 11 (18.3%) 0.955 Chest pain* 60 (27%) 16 (26.7%) 0.955 New cavitation on serial CXR
16 (7.2%) 0 0.032
Enlarged cavitation on serial CXR
10 (4.5%) 4 (6.7%) 0.494
New or progressive paracavitary fibrosis
12 (5.4%) 3 (5%) 0.901
New or progressive pleural thickening on CXR
3 (1.4%) 1 (1.7%) 0.855
2012 median Aspergillus IgG level
4.29 mg/L 3.7 mg/L 0.16**
2012 positive Aspergillus IgG
24 (10.8%) 5 (8.3%) 0.575
*present for 1 month or longer. ** medians compared by Mann Whitney U 4800 4801
S6 – CT findings by prior TB smear status 4802
4803
Result smear positive n=60
smear negative n=13
p-‐value by chi squared test
Cavities 40 (66.7%) 9 (69.2%) 0.858 Paracavitary fibrosis 37 (61.7%) 6 (46.2%) 0.303 Aspergilloma 11 (18.3%) 3 (23.1%) 0.694 Pleural thickening 30 (50%) 5 (38.5%) 0.45 Nodule 34 (56.7%) 3 (23.1%) 0.02
4804
187
S7 -‐ Symptoms and test results in resurvey by CD4 count groups 4805
4806
Result CD4 count <
200 cells/µL
n= 15
CD4 count
200 – 499
cells/µL n=65
CD4 count ≥
500 cells/µL
n=49
p-‐value by
chi-‐squared
test
Cough* 3 (20%) 15 (23.1%) 16 (32.7%) 0.433
Productive cough*
3 (20%) 4 (6.2%) 5 (10.2%) 0.241
Haemoptysis* 2 (13.3%) 7 (10.8%) 7 (14.3%) 0.847
Fatigue* 4 (26.7%) 22 (33.8%) 13 (26.5%) 0.667
Breathlessness* 4 (26.7%) 17 (26.2%) 16 (32.7%) 0.737
Fevers* 0 4 (6.2%) 8 (16.3%) 0.075
Night sweats* 2 (13.3%) 8 (12.3%) 12 (24.5%) 0.213
Chest pain* 4 (26.7%) 13 (20%) 13 (26.5%) 0.678
New cavitation on serial CXR
1 (6.7%) 3 (4.6%) 2 (4.1%) 0.917
Enlarged cavitation on serial CXR
0 3 (4.6%) 2 (4.1%) 0.703
New or progressive paracavitary fibrosis
1 (6.7%) 4 (6.2%) 2 (2.1%) 0.867
New or progressive pleural thickening on CXR
0 2 (3.1%) 0 0.368
2012 median Aspergillus IgG level
3.7 mg/L 3.75 mg/L 4.38 mg/L 0.954**
2012 positive Aspergillus IgG
0 4 (6.2%) 5 (10.2%) 0.372
*present for 1 month or longer. ** medians compared by independent samples median test 4807 4808
4809
4810
4811
4812
4813
188
S8 – CT findings by in HIV positive persons by CD4 count groups 4814
4815
4816
4817
Result CD4 count < 200 cells/µL n= 2
CD4 count 200 – 499 cells/µL n=11
CD4 count ≥ 500 cells/µL n=11
p-‐value
Cavities 1 (50%) 6 (54%) 9 (81.8%) 0.348 Paracavitary fibrosis
0 4 (36.4%) 9 (81.8%) 0.028
Aspergilloma 0 1 (9.1%) 4 (36.4%) 0.217 Pleural thickening
1 (50%) 3 (27.3%) 5 (45.5%) 0.631
Nodule 0 5 (45.5%) 6 (54.5%) 0.363
189
PAPER 4 - An estimate of the prevalence of pulmonary aspergillosis in HIV-positive 4818
Ugandan in patients diagnosed as smear-negative pulmonary tuberculosis. 4819
4820
Authors 4821
4822
Iain D Page -‐ Institute of Inflammation and Repair, The University of Manchester, UK, 4823
Manchester Academic Health Science Centre, UK, National Aspergillosis Center, 4824
University Hospital of South Manchester, UK. 4825
4826
William Worodria – Mulago Hospital, Kampala, Uganda 4827
4828
Alfred Andama – Mulago Hospital, Kampala, Uganda 4829
4830
Irene Ayakaka – Mulago Hospital, Kampala, Uganda 4831
4832
Richard Kwizera -‐ Institute of Inflammation and Repair, The University of Manchester, 4833
UK, Manchester Academic Health Science Centre, UK, National Aspergillosis Center and 4834
Mycology Reference Centre, University Hospital of South Manchester, UK, Infectious 4835
Diseases Institute, Mulago Hospital, Kampala, Uganda 4836
4837
Lucien Davis – University of California San Francisco, USA 4838
4839
Laurence Huang -‐ University of California San Francisco, USA 4840
4841
Malcolm Richardson -‐ Institute of Inflammation and Repair, The University of 4842
Manchester, UK, Manchester Academic Health Science Centre, UK, National Aspergillosis 4843
Center and Mycology Reference Centre, University Hospital of South Manchester, UK. 4844
4845
David W Denning -‐ Institute of Inflammation and Repair, The University of Manchester, 4846
UK, Manchester Academy Health Science Centre, UK, National Aspergillosis Centre, 4847
University Hospital of South Manchester, UK. 4848
4849
190
Abstract 4850
4851
In resource-‐poor settings pulmonary tuberculosis is often diagnosed on the basis of 4852
‘smear-‐negative’ criteria. Microbiological proof of tuberculosis infection is not required. 4853
In Ugandan HIV positive in-‐patients these clinical protocols have negligible diagnostic 4854
value. Subacute invasive pulmonary aspergillosis, also known as chronic necrotizing 4855
pulmonary aspergillosis (CNPA) also occurs in HIV infected persons and is 100% fatal 4856
without treatment. Autopsy studies show that it is present in 2-‐3% of all AIDS deaths 4857
and went undiagnosed ante-‐mortem in over 90% of these cases. 4858
4859
Diagnosis of CNPA / subacute invasive pulmonary aspergillosis requires one month of 4860
cough or haemoptysis, plus radiological evidence of cavitating lung lesions with 4861
paracavitary infiltrates and raised Aspergillus-‐specific IgG. Aspergilloma is not always 4862
present. Such patients would likely be classified as ‘smear-‐negative pulmonary 4863
tuberculosis’ in the absence of specific testing for aspergillosis. We aimed to estimate 4864
the likely prevalence of pulmonary aspergillosis in an at-‐risk African in-‐patient 4865
population. 4866
4867
Stored sera were available from adult patients admitted to Mulago Hospital, Kampala 4868
with a cough of 2 weeks to 6 months duration between March 2010 and March 2011. 4869
These patients were thoroughly investigated for tuberculosis. We selected 39 sera from 4870
HIV infected persons with abnormal chest X-‐rays with no evidence of tuberculosis or 4871
other clear diagnosis after full investigation. We measured Aspergillus-‐specific IgG in 4872
these samples using the Siemens Immulite assay, which has a specificity of 98% and 4873
sensitivity of 96% for the diagnosis of chronic pulmonary aspergillosis. 100 control sera 4874
had previously been collected from healthy Ugandan blood donors. These were used in 4875
receiver operating characteristic curve studies to define the diagnostic threshold of 10 4876
mg/L used in this study. 4877
4878
The mean patient age was 35 years and 59% of patients were female. Mean CD4 count 4879
was 109 cells/µL. 44% of patients had CD4 count <50 cells/µL. Raised Aspergillus-‐4880
specific IgG was present in 2% of healthy controls, but 26% of patients (95% CI 14 – 4881
41%, p 0.000). 40% of those with a positive test died within 2 months of sampling. 4882
191
While this study does not conclusively prove the existence of subacute invasive 4883
aspergillosis in this population, it is likely that the majority of these patients with raised 4884
Aspergillus-‐specific IgG had either chronic or subacute pulmonary aspergillosis. 4885
Pulmonary aspergillosis is probably being misdiagnosed as ‘smear-‐negative 4886
tuberculosis’ in many patients with HIV infection. Further prospective studies with CT 4887
scanning, plus effective fungal culture and serology should be performed to investigate 4888
this possibility. 4889
4890
4891
192
Introduction 4892
4893
In 2012 4.4 million patients were diagnosed with pulmonary tuberculosis following a 4894
sputum smear test. In 1.9 million of these cases the smear test was negative215. In the 4895
resource-‐poor settings where tuberculosis and HIV are common, pulmonary 4896
tuberculosis is often diagnosed on the basis of World Health Organization (WHO) 4897
approved ‘smear-‐negative’ criteria242. Microbiological proof of tuberculosis infection is 4898
not required241. Currently 54% of Ugandan HIV positive out-‐patients commencing 4899
tuberculosis therapy are smear-‐negative. Tuberculosis is only confirmed in 35% of 4900
these patients when sputum is cultured243. In Ugandan HIV positive in-‐patients these 4901
clinical diagnostic protocols have negligible diagnostic value for tuberculosis244. 4902
4903
The 2-‐year mortality of smear-‐negative TB is 34%296 . The hazard ratio for mortality in 4904
smear-‐negative tuberculosis against smear-‐positive tuberculosis is 1.49 for 2-‐month 4905
mortality in HIV positive cases in DR Congo245. 4906
4907
The WHO diagnostic criteria for smear negative tuberculosis require all of the following; 4908
HIV infection, cough for two weeks or more, two negative sputum acid alcohol fast 4909
bacilli (AAFB) smear tests, no response to broad-‐spectrum antibiotics and radiological 4910
changes potentially consistent with tuberculosis241. 4911
4912
Fungal lung infections would also meet these criteria. Chronic pulmonary aspergillosis 4913
(CPA) presents with cough of at least 3 months duration, plus haemoptysis, weight loss 4914
and fatigue5,7,8. Subacute pulmonary aspergillosis, also known as chronic necrotizing 4915
pulmonary aspergillosis (CNPA), has similar symptoms, but is more rapidly progressive, 4916
with a duration of illness of only one month6,49,51,297. 4917
4918
Cavities, pleural thickening and fibrosis are found on chest X-‐ray in both tuberculosis 4919
and aspergillosis266,280. Aspergilloma are distinctive, but are present in only 36% of 4920
cases of CPA8. Raised levels of Aspergillus-‐specific IgG differentiate CPA from similar 4921
conditions5,7,8, but this test is essentially unavailable in Africa220. 4922
4923
193
CPA has a global distribution, with large case series reported in the UK, France, India, 4924
China, Korea and Japan7,8,15,18,108,198. Over 180 cases of CPA have been reported 4925
throughout Africa, including South Africa, Nigeria, Ivory Coast, Senegal, Central African 4926
Republic, Djibouti, Ethiopia, Tanzania and Uganda16,201–212. A recent survey in Uganda 4927
demonstrated that CPA complicates 6% of all treated pulmonary tuberculosis cases 4928
(paper 3). The global burden of CPA secondary to tuberculosis has recently been 4929
estimated at around 1.2 million cases11. 4930
4931
Invasive aspergillosis can complicate AIDS in the absence of tuberculosis. It is 4932
associated with corticosteroid therapy and pulmonary infection, including Pneumocystis 4933
jirovecii or bacterial pneumonia. Drug-‐induced neutropaenia is present in most cases, 4934
however 44% of AIDS related cases occur in patients with normal neutrophil counts, 4935
but CD4 counts below 100 cells/μL235. Advanced AIDS is associated with impaired 4936
neutrophil function237, which might result in increased susceptibility to acute 4937
aspergillosis. In patients who are not profoundly neutropaenic, a subacute course of one 4938
to two months duration is typical52,223,224,235. Survivors who receive effective HIV 4939
treatment can develop CPA223. 4940
4941
Autopsy studies from Italy, India and Uganda have demonstrated that aspergillosis is 4942
present in 3-‐11% of all AIDS related deaths and that only 10% of these cases were 4943
diagnosed antemortem207,238–240. A recent study identified Aspergillus fumigatus growth 4944
in BAL samples from 6% of patients admitted to hospital in Uganda with subacute 4945
respiratory disease, the majority of whom were HIV positive277. 4946
4947
The mortality of untreated invasive aspergillosis is 100% within weeks19,248,249. CPA has 4948
5-‐year mortality rate of 40-‐85%7,8,264. If cases of pulmonary aspergillosis occur 4949
commonly and are being misdiagnosed as smear-‐negative tuberculosis then this could 4950
be making a substantial contribution to the excess mortality observed in this group. 4951
4952
Effective treatment is available. Most patients survive invasive aspergillosis if treated 4953
with voriconazole or amphotericin247 and oral itraconazole prevents disease 4954
progression in CPA18,198. Surgical resection of individual lesions by lobectomy can be 4955
safely delivered in resource-‐poor settings16,54,212 and is potentially curative15,21. 4956
194
We aimed to estimate the prevalence of subacute invasive pulmonary aspergillosis in 4957
HIV positive patients treated as ‘smear-‐negative tuberculosis’. We performed 4958
opportunistic testing of stored sera at Mulago Hospital, Kampala. These were acquired 4959
during an earlier study to measure the frequency of different conditions in patients 4960
admitted with chronic cough276,277. 4961
4962
Stored sera were selected from adult in-‐patients who met the diagnostic criteria for 4963
smear-‐negative tuberculosis and who had no evidence of tuberculosis after extensive 4964
investigation including smear testing, GeneXpert PCR testing and culture on sputum 4965
and/or broncho-‐alveolar lavage fluid. Samples were tested for Aspergillus-‐specific IgG 4966
using the Siemens Immulite assay, which has recently been shown to have a sensitivity 4967
of 96% and specificity of 98% for the diagnosis of CPA (paper 1). 4968
4969
Methods 4970
4971
The Mulago Inpatient Noninvasive Diagnosis – International HIV Opportunistic 4972
Pneumonia (MIND-‐IHOP) study recruited patients between March 2010 and March 4973
2011. During this period all adults admitted to the casualty department of Mulago 4974
Hospital, Kampala on weekdays, with a cough of between 2 weeks and 6 months 4975
duration were offered admission to the study. Clinical details were recorded and 4976
sputum samples taken for acid and alcohol fast bacteria (AAFB) smear testing, culture 4977
and GeneXpert automated nucleic acid amplification assay (Cepheid, USA). Induced 4978
sputum was acquired if necessary. 4979
4980
CD4 count (in HIV infected patients) and cryptococcal antigen testing (Imuno 4981
Mycologics, USA) were performed on blood. Bronchoscopy was offered to any HIV 4982
positive patient with persistent symptoms and negative sputum smear test. Broncho-‐4983
alveolar lavage (BAL) specimens underwent culture and staining for mycobacteria and 4984
fungi including Pneumocystis jirovecii. Patients were reviewed at two months after 4985
recruitment. The mortality rate at this point was recorded. 4986
4987
Stored sera were available from around three quarters of patients originally recruited 4988
to the study. Sera were retrospectively selected from patients meeting the following 4989
195
criteria; 1 -‐ HIV infection, 2 -‐ abnormal chest X-‐ray, 3 -‐ No diagnosis made after all 4990
investigations complete, including no evidence of pulmonary tuberculosis. All patients 4991
were treated with a broad-‐spectrum antibiotic, normally ceftriaxone, on admission. 4992
Those who responded were diagnosed as ‘likely bacterial pneumonia’ and so were not 4993
included in the ‘no diagnosis’ group. 100 control sera had previously been collected 4994
from healthy Ugandan blood donors. These were used in receiver operating 4995
characteristic curve studies to define the diagnostic threshold of 10 mg/L used in this 4996
study (paper 1). 4997
4998
Levels of Aspergillus-‐specific IgG were measured in each selected sample by Immulite 4999
2000 (Siemens, Germany) assay. Samples with a level greater than 200mg/L underwent 5000
a 1 in 10 dilution and were repeated. 5001
5002
Statistical analysis 5003
5004
Median Aspergillus-‐specific IgG levels and CD4 counts are compared with Mann-‐5005
Whitney U test. Mean ages are compared by 2-‐sided t-‐test. Categorical results are 5006
compared by 2-‐sided Fisher’s exact test. 5007
5008
Results 5009
5010
Sera from 39 patients that met the stated criteria were identified. 23 (59%) patients 5011
were female and the mean age was 35 years (range 21-‐54). Mean CD4 count was 109 5012
cells/µL (range 3 -‐399 cells/µL). 23 (62%) patients had CD4 count < 100 cells/µL and 5013
17 (44%) patients had CD4 count <50 cells/µL. Thirteen (33%) patients underwent 5014
bronchoscopy as part of their investigations. Chest X-‐ray showed infiltrates in 34 (87%) 5015
of cases, military appearance in one (3%) case and “likely tuberculosis”, with no further 5016
details in 4 (10%) cases. 5017
5018
Results of Aspergillus-‐specific IgG are shown in table 1. 100 control samples from 5019
healthy Ugandan blood donors were also tested (paper 1). The frequency of raised 5020
Aspergillus-‐specific IgG in controls was 2% (paper 1). The frequency of raised 5021
196
Aspergillus-‐specific IgG levels in patients meeting the study criteria was 26% (95% 5022
confidence interval 14% -‐ 41%). 5023
5024
Table 1 – Aspergillus-‐specific IgG testing in patients and controls 5025
5026
Result Healthy controls n=100
Study patients n=39
p-‐value
Median Aspergillus-‐specific IgG level
4 mg/L 7 mg/L 0.000
Aspergillus-‐specific IgG range
0-‐35 mg/L 2–26 mg/L -‐
Number of positive tests
2 (2%) 10 (26%) 0.000
5027
Table 2 – Characteristics of patients with and without raised Aspergillus-‐specific IgG 5028
5029
Result Normal Aspergillus-‐specific IgG n=29
Raised Aspergillus-‐specific IgG n=10
p-‐value
Female gender 15 (52%) 8 (80%) 0.105 Mean age 34 years 37 years 0.052 Median CD4 count 59 cells/µL 59 cells/µL -‐ CD4 <100 cells/µL 16 (55%) 7 (70%) 0.48 2 month mortality 8 (27%) 4 (40%) 0.463 5030
5031
5032
197
Discussion 5033
5034
These results suggest that subacute invasive or chronic pulmonary aspergillosis are 5035
important differential diagnoses in Ugandan in-‐patients with AIDS who are currently 5036
diagnosed and managed as ‘smear-‐negative tuberculosis’. As the mortality rate of 5037
pulmonary aspergillosis is very high this is likely contribute the excess mortality 5038
observed in this population. 5039
5040
We are unable to differentiate acute, subacute invasive and chronic pulmonary 5041
aspergillosis in patients diagnosed with ‘smear-‐negative tuberculosis’ due to differences 5042
in case definition. CPA requires chronic cough, defined as three months or more5,250,251, 5043
whereas subacute invasive pulmonary aspergillosis requires one month of cough6,264. 5044
Our cohort includes patients with cough for two weeks or more. This would capture 5045
both conditions and acute invasive disease. 5046
5047
Progressive cavitation, paracavitary fibrosis or aspergilloma are also required for the 5048
diagnosis of pulmonary aspergillosis. Radiological information was only available from 5049
chest X-‐ray, which is unreliable for the diagnosis of pulmonary tuberculosis, especially 5050
when patients are co-‐infected with HIV265. The findings in acute pulmonary aspergillosis 5051
are often non-‐specific218,280. We excluded patients with normal chest X-‐rays, but 5052
accepted those with any chest X-‐ray abnormality. While any chest X-‐ray abnormality is 5053
potentially consistent with pulmonary aspergillosis a CT scan would have been required 5054
to state that radiological features of pulmonary aspergillosis were definitely present. 5055
5056
Raised levels of Aspergillus-‐specific IgG are consistent with pulmonary aspergillosis, but 5057
can also occur in colonization38, Aspergillus bronchitis39 or tracheobronchitis43. Active 5058
pulmonary aspergillosis is, however the most likely of these options in our patients as 5059
they all had persistent respiratory symptoms and abnormal chest X-‐ray. 5060
5061
While fungal culture on BAL samples did not identify any cases of aspergillosis in this 5062
group, the sensitivity of culture for Aspergillus is very low with standard techniques260 5063
and so the absence of culture growth does not exclude pulmonary aspergillosis. Testing 5064
198
BAL for galactomannan might also have provided confirmation of pulmonary 5065
aspergillosis48, but no stored samples were available. 5066
5067
We measured antibodies to Aspergillus fumigatus, which is responsible for the vast 5068
majority of CPA in Europe and East Asia 5–8,108. However, most aspergillosis in India and 5069
the Middle East is due to A. flavus10 and A. niger is common in Brazil147. The dominant 5070
species of Aspergillus in Africa is not known. A. fumigatus based assays can have poor 5071
sensitivity for CPA due to other Aspergillus species147,148, potentially resulting in false 5072
negative results. Profound immunosuppression dampens antibody responses in 5073
AIDS273,274 and the performance of Aspergillus-‐specific IgG in a group with low CD4 5074
counts is not well-‐described. These factors may lead to false negative results. 5075
5076
The Aspergillus-‐specific IgG assay can cross-‐react with Penicillium-‐specific antibodies269. 5077
Little is known about its cross-‐reactivity with other fungal infections. Histoplasmosis is 5078
present in Uganda268 and blastomycosis elsewhere in Africa232. These fungal infections 5079
are among those known to complicate HIV/AIDS292,298. False positives might occur if 5080
other fungal infections are present in our patients and they cross-‐reacted with the 5081
Siemens Immulite assay. 5082
5083
This study was conducted in an opportunistic manner, using stored sera from a prior 5084
study. Not all sera had sufficient volume to allow Aspergillus serology. The sickest 5085
patients may have been excluded from our study as obtaining large volumes of blood 5086
can be difficult in these cases due to shock. This might lead to an underestimate of the 5087
prevalence of a rapidly fatal condition such as pulmonary aspergillosis. 5088
5089
To our knowledge this is the first attempt to estimate the prevalence of pulmonary 5090
aspergillosis in an African cohort with AIDS and ‘smear-‐negative tuberculosis’. While 5091
our study design does not include all the standard diagnostic tests for pulmonary 5092
aspergillosis in highly immunocompromised patients, the Siemens Immulite assay used 5093
has a specificity of 98% for the diagnosis of CPA (paper 1). It is therefore likely that 5094
pulmonary aspergillosis, or other chronic fungal lung disease, is present in many of the 5095
patients in this cohort. 5096
5097
199
Given the combination of high mortality and good response to treatment in these 5098
conditions, there is now an urgent need to perform thorough prospective studies in this 5099
population including CT scan, effective fungal culture with optimal techniques260, 5100
extensive fungal serology and ideally biopsy to definitively measure the prevalence of 5101
fungal lung disease in Africans with AIDS and subacute respiratory infection. 5102
5103
Hypothesis 5104
5105
That a proportion of patients who presented to hospital with AIDS and sub-‐acute 5106
respiratory disease in an area of high tuberculosis prevalence, but who had no evidence 5107
of tuberculosis after thorough investigation were suffering from undiagnosed primary 5108
sub-‐acute invasive pulmonary aspergillosis. 5109
5110
Aims 5111
5112
1 – To measure the levels of Aspergillus-‐specific IgG in stored sera from HIV infected 5113
patients admitted to hospital with sub-‐acute respiratory disease in an area of high 5114
tuberculosis prevalence, but who had no evidence of tuberculosis after thorough 5115
investigation. 5116
5117
2 – To compare these levels of Aspergillus-‐specific IgG to those found in healthy controls 5118
from the same country. 5119
5120
3 – To compare two-‐month mortality outcomes in HIV infected patients with sub-‐acute 5121
respiratory disease in an area of high tuberculosis prevalence with no evidence of 5122
tuberculosis after thorough investigation, who had either raised or normal levels of 5123
Aspergillus-‐specific IgG. 5124
5125
Ethics 5126
5127
Ethical permission for this study was granted by the University of Manchester, UK (ref 5128
11424), Makerere University, Kampala, Uganda (ref 2006-‐017) and the Ugandan 5129
National Council for Science and Technology (ref – HS259). 5130
200
5131
Funding 5132
5133
Funding to transport samples to the UK for analysis was provided by the Manchester 5134
Academy academic charity. Test kits for use in this study were kindly donated by 5135
Siemens. 5136
5137
Acknowledgements 5138
5139
We would like to thank all those involved in the MIND-‐IHOP study group for their kind 5140
decision to share serum samples for use in this collaborative study. Thanks to the North 5141
West Lung Centre, University Hospital of South Manchester for storage of samples. 5142
Thanks to the pathology laboratory staff at Christie Hospital, Manchester, UK for 5143
allowing the study group access to their Siemens Immulite 2000. 5144
5145
201
PAPER 5 - Aspergillus co-infection may be common in Africans with active pulmonary 5146
tuberculosis 5147
5148
Authors 5149
5150
Iain D Page -‐ Institute of Inflammation and Repair, The University of Manchester, UK, 5151
Manchester Academic Health Science Centre, UK, National Aspergillosis Center, 5152
University Hospital of South Manchester, UK. 5153
5154
William Worodria – Mulago Hospital, Kampala, Uganda 5155
5156
Alfred Andama – Mulago Hospital, Kampala, Uganda 5157
5158
Irene Ayakaka – Mulago Hospital, Kampala, Uganda 5159
5160
Richard Kwizera -‐ Institute of Inflammation and Repair, The University of Manchester, 5161
UK, Manchester Academic Health Science Centre, UK, National Aspergillosis Center and 5162
Mycology Reference Centre, University Hospital of South Manchester, UK, Infectious 5163
Diseases Institute, Mulago Hospital, Kampala, Uganda 5164
5165
Lucien Davis – University of California San Francisco, USA 5166
5167
Laurence Huang -‐ University of California San Francisco, USA 5168
5169
Malcolm Richardson -‐ Institute of Inflammation and Repair, The University of 5170
Manchester, UK, Manchester Academic Health Science Centre, UK, National Aspergillosis 5171
Center and Mycology Reference Centre, University Hospital of South Manchester, UK. 5172
5173
David W Denning -‐ Institute of Inflammation and Repair, The University of Manchester, 5174
UK, Manchester Academy Health Science Centre, UK, National Aspergillosis Centre, 5175
University Hospital of South Manchester, UK. 5176
5177
202
Abstract 5178
5179
CPA is estimated to affect 3 million people globally11–13. A recent survey demonstrated 5180
that chronic pulmonary aspergillosis (CPA) is present in 6.5% of Ugandan adults with 5181
previously treated pulmonary tuberculosis and raised Aspergillus-‐specific IgG present in 5182
10% (papers 2 and 3). These cases occurred in patients who no longer had active 5183
tuberculosis infection. However, active co-‐infection with atypical mycobacteria 5184
frequently occurs in CPA and co-‐infection with active Mycobacterium tuberculosis and 5185
Aspergillus has been described in several case reports. We aimed to estimate the 5186
prevalence of this problem in an area of high tuberculosis prevalence. 5187
Stored sera were available from 57 adult patients admitted to Mulago Hospital, Kampala 5188
between March 2010 and March 2011. All patients had between 2 weeks and 6 months 5189
cough and were diagnosed with pulmonary tuberculosis based on culture or GeneXpert 5190
PCR testing of sputum and/or broncho-‐alveolar fluid. We measured Aspergillus-‐specific 5191
IgG in these samples using the Siemens Immulite assay, which has a specificity of 98% 5192
and sensitivity of 96% for the diagnosis of chronic pulmonary aspergillosis. 5193
46 (81%) patients were HIV positive. Mean CD4 count in those with HIV was 99 5194
cells/µL (range 2 -‐ 581 cells/µL). 35 (61%) patients had CD4 count < 100 cells/µL and 5195
24 (42%) patients had CD4 count <50 cells/µL. 5196
Aspergillus-‐specific IgG levels were raised in 2 (2%) of controls and 27 (47%) 5197
tuberculosis patients. 3 (11%) of those with raised Aspergillus-‐specific IgG died within 2 5198
months of sampling. 5199
This is a select group of patients requiring emergency hospital admission and may not 5200
be representative of all patients with pulmonary tuberculosis. False positive IgG results 5201
might occur due to cross-‐reaction with other fungi and false negative IgG tests might 5202
occur in patients with CPA caused Aspergillus species other than A. fumigatus. 5203
However, given the diagnostic accuracy of the Siemens Immulite assay it is likely that 5204
active Aspergillus co-‐infection is present in many of those with positive results. This 5205
possibility should be considered in patients who fail to improve or clinically relapse in 5206
203
spite of appropriate tuberculosis therapy. Prospective studies are needed to record the 5207
outcome of patients with pulmonary tuberculosis and raised Aspergillus-‐specific IgG and 5208
define the prevalence of pulmonary aspergillosis in this group. 5209
5210
5211
204
Introduction 5212
5213
An estimated 9 million people developed tuberculosis in 2013215. It was associated with 5214
1.5 million deaths, of which only 210,000 were estimated to be due to multidrug 5215
resistant strains. Many of the other 1.29 million deaths will have been due to late 5216
presentation, lack of diagnosis, poor access to treatment or inadequate compliance, 5217
given that they mostly occurred in resource-‐poor countries with weak health 5218
infrastructure. However, other factors may also be have been present. 5219
5220
Chronic pulmonary aspergillosis (CPA) is an important sequel of pulmonary 5221
tuberculosis14. It presents with progressive pulmonary cavitation associated with 5222
weight loss, persistent cough and haemoptysis5,7,8. It has a 5-‐year mortality of 50 – 5223
80%6,7,264 and has recently been estimated to affect around 3 million people globally11–5224 13, including 1.2 million cases secondary to tuberculosis11. 5225
5226
Large CPA case series have been reported in the UK, France, India, China, Korea and 5227
Japan, the majority of which are secondary to tuberculosis7,8,14,15,18,108,198. Over 180 5228
cases of CPA have been reported throughout Africa, including South Africa, Nigeria, 5229
Ivory Coast, Senegal, Central African Republic, Djibouti, Ethiopia, Tanzania and 5230
Uganda16,201–212. Over 90% of these cases were secondary to pulmonary tuberculosis. A 5231
recent survey confirmed that CPA is present in 6% of Ugandan adults with previously 5232
treated pulmonary tuberculosis (paper 3). 5233
5234
CPA is treatable. Oral itraconazole, voriconazole or posaconazole all prevent clinical and 5235
radiological progression18,58,108,198. Surgery is curative in selected patients with localized 5236
disease15,21 and has been safely undertaken in resource-‐poor settings16,54,212. 5237
5238
While it is now clear that CPA frequently follows tuberculosis, the natural history of CPA 5239
is not well established. Published CPA cohort studies are all from countries where 5240
tuberculosis is now uncommon. Atypical mycobacteria, however, commonly co-‐infect 5241
persons with CPA, in addition to their role as an antecedent condition7,8,14. Evidence 5242
from countries with high tuberculosis prevalence is limited to case reports, but co-‐5243
205
infection with active pulmonary tuberculosis and pulmonary aspergillosis has been 5244
documented in India, Tunisia and Egypt213,214,299–301. A recent study identified 5245
Aspergillus fumigatus growth in BAL samples from 6% of patients admitted to Mulago 5246
Hospital, Kampala with suspected tuberculosis277. This may well be an underestimate 5247
as standard culture techniques have very poor sensitivity for Aspergillus260. 5248
5249
Co-‐infection with Aspergillus at the time of active pulmonary tuberculosis might also 5250
result in subacute invasive aspergillosis. This condition occurs in patients with mild to 5251
moderate immunosuppression and has been noted in a wide range of conditions 5252
including HIV infection, diabetes, alcohol abuse and COPD6,49,52,236,264. It presents with 5253
progressive pulmonary cavitation associated with weight loss, persistent cough and 5254
haemoptysis and is associated with 50% mortality within a few months. 5255
5256
Infection with Mycobacterium tuberculosis results in impaired immunity and decreased 5257
macrophage function282. This might well place a patient at risk of subacute invasive 5258
pulmonary aspergillosis. As the clinical and radiological presentation of this condition is 5259
essentially identical to pulmonary tuberculosis itself it would be very difficult to detect 5260
it without performing specific Aspergillus serological testing. 5261
We aimed to estimate the frequency of Aspergillus co-‐infection in patients recently 5262
diagnosed with active pulmonary tuberculosis. We performed opportunistic testing of 5263
stored sera at Mulago Hospital, Kampala. These were acquired during an earlier study 5264
to measure the frequency of different conditions in patients admitted with cough276,277. 5265
Mycobacterium tuberculosis infection was proven in all cases on the basis of smear 5266
testing, GeneXpert nucleic amplification or culture. Samples were tested for Aspergillus-‐5267
specific-‐IgG using the Siemens Immulite assay, which has recently been shown to have a 5268
sensitivity of 96% and specificity of 98% for the diagnosis of CPA (paper 1). 5269
5270
Methods 5271
5272
The Mulago Inpatient Noninvasive Diagnosis – International HIV Opportunistic 5273
Pneumonia (MIND-‐IHOP) Study recruited patients relevant to this study between March 5274
2010 and March 2011. During this period all adults admitted to the casualty department 5275
206
of Mulago Hospital, Kampala on weekdays, with a cough of between 2 weeks and 6 5276
months duration were offered admission to the study. All patients submitted sputum, 5277
on which smear testing for acid alcohol fast bacilli, nucleic acid amplification 5278
(GeneXpert, Cepheid, USA) and culture for Mycobacterium tuberculosis were performed. 5279
Bronchoscopy was also performed in selected patients. 5280
5281
Stored sera were available from around three quarters of patients originally recruited 5282
to the study. Sera were retrospectively selected from patients with proven pulmonary 5283
tuberculosis following the above investigations. 100 control sera had previously been 5284
collected from healthy Ugandan blood donors. These were used in receiver operating 5285
characteristic curve analysis to define the diagnostic threshold of 10 mg/L used in this 5286
study (paper 1). 5287
5288
Levels of Aspergillus-‐specific IgG were measured in each selected sample (Siemens 5289
Immulite 2000, Germany). Samples with a level greater than 200mg/L underwent a 1 in 5290
10 dilution and were repeated. 5291
5292
Statistical analysis 5293
5294
Median Aspergillus-‐specific IgG levels in patients and controls and CD4 counts in those 5295
with and without raised Aspergillus-‐specific IgG levels are compared with Mann-‐5296
Whitney U test. Mean age in those with and without raised Aspergillus-‐specific IgG levels 5297
is compared by 2-‐sided t-‐test. Categorical variables are compared with Chi-‐squared test, 5298
except for comparison of number of positive Aspergillus-‐specific IgG tests in 5299
tuberculosis cases vs. healthy controls, where Fisher’s exact test is used. 5300
5301
Results 5302
5303
Fifty-‐seven sera that met the stated criteria were identified. 29 (51%) patients were 5304
female. Mean age was 35 years (range 18 – 79 years). 46 (81%) patients were HIV 5305
positive. Mean CD4 count in those with HIV was 99 cells/µL (range 2 -‐581 cells/µL). 35 5306
(61%) patients had CD4 count < 100 cells/µL and 24 (42%) patients had CD4 count <50 5307
207
cells/µL. All patients had a chest X-‐ray that was reported as abnormal and potentially 5308
consistent with pulmonary tuberculosis. 5309
5310
100 control samples were acquired from healthy blood donors in Gulu, Uganda. The 5311
frequency of raised Aspergillus-‐specific IgG in controls was 2% (paper 1). The frequency 5312
of raised Aspergillus-‐specific IgG levels in patients with proven pulmonary tuberculosis 5313
was 47% (95% confidence interval 35% -‐ 60%). 5314
5315
Table 1 – Aspergillus-‐specific IgG testing in patients and controls 5316
Result Healthy controls n=100
Pulmonary tuberculosis n=57
p-‐value
Mean Aspergillus-‐specific IgG level
5 mg/L 11 mg/L 0.000
Aspergillus-‐specific IgG range
0-‐35 mg/L 4 -‐ 36mg/L -‐
Number of positive tests
2 (2%) 27 (47%) 0.000
5317
Table 2 – Characteristics of patients with and without raised Aspergillus-‐specific IgG 5318
Result Normal Aspergillus-‐specific IgG n=30
Raised Aspergillus-‐specific IgG n=27
p-‐value
Female gender 12 (40%) 17 (63%) 0.08 Mean age 38 years 38 years -‐ HIV positive 24 (80%) 22 (81%) 0.887 Median CD4 count in those with HIV
49 cells/µL 46 cells/µL 0.560
CD4 <100 cells/µL in those with HIV
17 (71%) 18 (82%) 0.761
2 month mortality 5 (17%) 3 (11%) 0.547 5319 5320
5321
208
Discussion 5322 5323
Overall, 27 (47%) of patients had raised levels of Aspergillus-‐specific IgG during 5324
admission for microbiologically confirmed pulmonary tuberculosis. These results add to 5325
the growing body of evidence that pulmonary aspergillosis is a common complication of 5326
pulmonary tuberculosis. They inference is that CPA may well begin when active 5327
tuberculosis infection is still present. 5328
5329
We cannot, however state that all patients with raised Aspergillus-‐specific IgG definitely 5330
have CPA. The diagnosis of CPA also requires all of the following in addition to raised 5331
Aspergillus-‐specific IgG; 1 – productive cough or haemoptysis of at least 3 months 5332
duration, 2 – radiological findings of either progressive cavitation, paracavitary fibrosis 5333
or aspergilloma, 3 -‐ exclusion of conditions with a similar presentation5,7,8,250. 5334
5335
The MIND-‐IHOP study allowed recruitment of patients with only 2 weeks of cough. The 5336
patients in this group also clearly have another condition confirmed, rather than 5337
excluded. If their symptoms resolve entirely with tuberculosis treatment then they 5338
could not reasonably be considered cases of CPA. However the possibility of sub-‐clinical 5339
CPA, that might cause symptoms months or years later, cannot be excluded without 5340
prolonged follow up. 5341
5342
Given that CPA is only found in 6% of patients who have completed treatment for 5343
pulmonary tuberculosis, it is possible that many of those with raised Aspergillus-‐specific 5344
IgG at the time of active pulmonary tuberculosis are simply simply colonized with 5345
Aspergillus and that this colonization frequently resolves after the tuberculosis is 5346
treated, without developing CPA. 5347
5348
Alternatively it may be that co-‐infection with both HIV and Aspergillus results in a worse 5349
clinical course in pulmonary tuberculosis, perhaps through the development of invasive 5350
pulmonary aspergillosis. If this were the case then higher rates of hospitalization would 5351
be seen in those with active Aspergillus co-‐infection than would be seen in the 5352
tuberculosis population as a whole. This would also explain the unexpectedly high rate 5353
of raised Aspergillus-‐specific IgG seen in this in patient population. 5354
209
Raised levels of Aspergillus-‐specific IgG are consistent with CPA, but can also occur in 5355
colonization38, Aspergillus bronchitis39 or tracheobronchitis43. The radiological features 5356
of CPA are normally confirmed on CT scan, which was not included in this study. While 5357
every patient in this study had an abnormal chest X-‐ray, the only radiological finding 5358
that differentiates CPA from tuberculosis is aspergilloma, which is absent in the 5359
majority of cases of CPA8 and harder to detect with chest X-‐ray than CT scan (paper 3). 5360
We cannot therefore differentiate CPA from these other conditions in patients with 5361
active tuberculosis and no aspergilloma, as it is unclear whether M. tuberculosis or 5362
Aspergillus is primarily responsible for the abnormal radiological findings. Repeat 5363
imaging after tuberculosis treatment is complete would be required to identify definite 5364
CPA cases. 5365
5366
We measured antibodies to Aspergillus fumigatus, which is responsible for the vast 5367
majority of CPA in Europe and East Asia 5–8,108. However most aspergillosis in India and 5368
the Middle East is due to A. flavus10 and A. niger is common in Brazil147. The dominant 5369
species of Aspergillus in Africa is not known. A. fumigatus based assays can have poor 5370
sensitivity for other species147,148, potentially resulting in false negative results. 5371
Antibody responses are also generally poor in AIDS273,274, which affected a large number 5372
of patients in our cohort, although we noted in an earlier study that levels of Aspergillus-‐5373
specific IgG are often raised in patients with AIDS and sub-‐acute respiratory disease 5374
(paper 4). 5375
5376
The Aspergillus-‐specific IgG assay can cross-‐react with Penicillium antibodies269. Little is 5377
known about its cross-‐reactivity with other fungal infections. Histoplasmosis is present 5378
in Uganda268 and blastomycosis elsewhere in Africa232. All these fungal infections are 5379
among those known to complicate HIV/AIDS292,298. False positives might occur if other 5380
fungal infections are present in our patients and they cross-‐reacted with the Siemens 5381
Immulite assay. 5382
5383
This study was performed in a population diagnosed with pulmonary tuberculosis 5384
during acute admission to hospital. The rate of HIV co-‐infection in this group is higher 5385
than the overall frequency of HIV co-‐infection seen in Ugandan tuberculosis patients215. 5386
Pulmonary aspergillosis might well be common in this study population, which has 5387
210
severe disease and unusually severe immunosuppression. The study group is therefore 5388
not representative of all newly diagnosed pulmonary tuberculosis. 5389
5390
Our study does not therefore definitively measure the prevalence of CPA in patients 5391
with active pulmonary tuberculosis. However the Siemens Immulite assay has good 5392
sensitivity and specificity for the diagnosis of CPA (paper 1). It is therefore likely that 5393
many of the patients identified here are suffering from some form of pulmonary 5394
aspergillosis, or other fungal lung disease. This possibility should be actively considered 5395
in any patient with pulmonary tuberculosis who is failing to respond to appropriate 5396
therapy or who has symptomatic relapse after initial response to tuberculosis therapy. 5397
5398
Prospective studies including CT thorax, fungal serology and fungal culture using 5399
sensitive high volume techniques260 are now needed to confirm the frequency of 5400
Aspergillus co-‐infection in pulmonary tuberculosis. Follow up is required to identify if 5401
and when these patients develop CPA and the optimal treatment strategy for them. 5402
Given the recently confirmed high prevalence of CPA complicating tuberculosis (paper 5403
3) and the high mortality rate of pulmonary aspergillosis6,7 these studies should be 5404
performed urgently. 5405
5406
Hypothesis 5407
5408
That chronic pulmonary aspergillosis (CPA) begins to develop during active infection 5409
with pulmonary tuberculosis. 5410
5411
Aims 5412
5413
1 – To measure the levels of Aspergillus-‐specific IgG in stored sera from HIV infected 5414
patients admitted to hospital with proven active pulmonary tuberculosis. 5415
5416
2 – To compare these levels of Aspergillus-‐specific IgG to those found in healthy controls 5417
from the same country. 5418
5419
211
3 – To compare two-‐month mortality outcomes in HIV infected patients admitted to 5420
hospital with active pulmonary tuberculosis with either raised or normal levels of 5421
Aspergillus-‐specific IgG. 5422
5423
Ethics 5424
5425
Ethical permission for this study was granted by the University of Manchester, UK (ref 5426
11424), Makerere University, Kampala, Uganda (ref 2006-‐017) and the Ugandan 5427
National Council for Science and Technology (ref – HS259). 5428
5429
Funding 5430
5431
Funding to transport samples to the UK for analysis was provided by the Manchester 5432
Academy academic charity. Test kits for use in this study were kindly donated by 5433
Siemens. 5434
5435
Acknowledgements 5436
5437
We would like to thank all those involved in the MIND study group for their kind 5438
decision to share serum samples for use in this collaborative study. Thanks to the North 5439
West Lung Centre, University Hospital of South Manchester for storage of samples. 5440
Thanks to the pathology laboratory staff at Christie Hospital, Manchester, UK for 5441
allowing the study group access to their Siemens Immulite 2000. 5442
5443
212
SUMMARY 5444
5445
The study of pulmonary aspergillosis in persons without gross immunosuppression has 5446
been neglected. Although the existence of pulmonary aspergillosis in non-‐5447
immunosuppressed persons has been documented for over 200 years, the clinical 5448
syndrome of chronic pulmonary aspergillosis was only properly defined 12 years ago. 5449
Since then significant studies have been published that describe cohorts of patients with 5450
this condition in several countries in Europe and Asia. The central importance of 5451
Aspergillus-‐specific IgG measurement to the diagnosis of CPA has been established in 5452
these studies. The link between CPA and many underlying conditions has been 5453
established and the dominance of tuberculosis as the most common underlying cause of 5454
CPA on a global scale is now clear. 5455
5456
Recent cohort studies have demonstrated that CPA is associated with a high mortality 5457
rate over the course of a few years. Fortunately the response of CPA to treatment with 5458
itraconazole has been also established in a randomized controlled trial and the efficacy 5459
and safety of surgical treatment (in suitable cases) has been demonstrated in large 5460
cohort descriptions. The potential for intervention to prolong the lives of the estimated 5461
3 million persons living with CPA therefore exists. 5462
5463
Unfortunately, it is likely that the majority of persons with CPA are currently going 5464
undiagnosed and untreated. The major barriers to progress in this area are lack of 5465
confirmation of the predicted prevalence of CPA in areas with currently high 5466
tuberculosis prevalence and lack of validation of tests for the diagnosis of CPA. 5467
5468
This work provides a substantial contribution to the field by answering three major 5469
questions in relation to chronic pulmonary aspergillosis. 5470
5471
First the optimal diagnostic cut-‐offs for CPA have been defined for the first time for five 5472
of the available commercial Aspergillus-‐specific IgG ELISAs, including the assay most 5473
commonly used in the UK. While diagnostic cut-‐offs were provided by most 5474
213
manufacturers these were defined in relation to tiny numbers of patients with CPA, 5475
often pooled with patients with invasive or allergic aspergillosis. 5476
5477
Recent studies had shown the cut-‐offs in common use were sub-‐optimal for the 5478
diagnosis of ABPA in cystic fibrosis patients. There was no certainty that they were 5479
appropriate for CPA. Lack of clearly validated CPA diagnostic cut-‐offs for Aspergillus-‐5480
specific IgG was a major barrier to any attempts to improve access to CPA diagnosis. 5481
Defining these cut-‐offs was also a pre-‐requisite for any measurement of the prevalence 5482
of CPA. 5483
5484
By accessing stored sera from the world’s largest CPA cohort it was possible to identify 5485
a suitably large number of sera to perform a meaningful analysis. Crucially these sera 5486
were taken from patients not on antifungal treatment. Such sera are representative of 5487
patients being diagnosed with CPA for the first time. As antifungal treatment lowers 5488
Aspergillus-‐specific IgG levels any cut-‐off defined in relation to CPA patients on 5489
treatment may not be applicable to those being tested for initial diagnosis. This was 5490
probably a major methodological flaw in the limited number of prior studies in this 5491
area. 5492
5493
This study identified optimal diagnostic thresholds by performing ROC analysis of 5494
results obtained from this unique cohort of untreated CPA patients and healthy 5495
controls. In the case of several assays, including ThermoFisher Scientific ImmunoCAP, 5496
the assay currently in use in most of the UK, the existing cut-‐offs were shown to be too 5497
high. By lowering the cut-‐offs to optimal levels it is possible to markedly increase 5498
sensitivity, while maintaining high specificity. These results will change practice at the 5499
UK National Aspergillosis Centre and are highly likely to inform changes to guidelines. 5500
This will allow those units with access to testing to correctly identify around 10% more 5501
cases than was previously possible. 5502
5503
The second major contribution of this work to the field is to define the sensitivity and 5504
specificity of the five Aspergillus-‐specific IgG ELISAs, plus precipitins, for the diagnosis 5505
of CPA. The large cohort descriptions for CPA all suggested that Aspergillus-‐specific IgG 5506
has excellent sensitivity for CPA, but these studies all used a single Aspergillus antibody 5507
214
assay as their main serological test for aspergillosis. While a few small studies have 5508
recently compared the sensitivity and specificity of different tests, these have not 5509
directly compared the ELISAs included in this study in patients with CPA. These studies 5510
were also potentially flawed on account of sera being taken from patients who were on 5511
antifungal therapy. Defining the sensitivity and specificity of these assays was required 5512
before any assay could be selected for use in a CPA prevalence study. 5513
5514
The performance of each ELISA was described in terms of ROC area under the curve. A 5515
sufficiently large number of cases were assessed to allow the detection of statistically 5516
significant differences in the diagnostic performance of the various assays. No previous 5517
study had achieved this. It was confirmed that the assay currently in regular use in the 5518
UK (ThermoFisher Scientific ImmunoCAP) does indeed have good sensitivity and 5519
specificity for the diagnosis of CPA. 5520
5521
While there was some unavoidable bias in our cohort, due to the fact that Aspergillus-‐5522
specific IgG serology, including the ThermoFisher Scientific ImmunoCAP assay forms an 5523
integral part of the diagnostic process for patients at our unit, this study is still the 5524
definitive work in this field. The Siemens Immulite assay was shown to have equivalent 5525
sensitivity and specificity in spite of this potential bias in favour of ThermoFisher 5526
Scientific ImmunoCAP. Other assays performed less well. It was then possible to use the 5527
donated Siemens Immulite assay with confidence in a survey to measure the prevalence 5528
of CPA in an area of high tuberculosis prevalence. 5529
5530
The optimal diagnostic cut offs and comparative sensitivity and specificity of the assays 5531
for the diagnosis of ABPA was also defined. This was performed in relation to both 5532
healthy controls and to asthmatics. The appropriate cut offs for Aspergillus-‐specific IgG 5533
for the diagnosis of ABPA in patients with cystic fibrosis and the appropriate cut-‐offs for 5534
both total and Aspergillus-‐specific IgE for the diagnosis of ABPA in general have both 5535
been assessed in recent studies. This is, however the first study to define optimal cut-‐5536
offs for Aspergillus-‐specific IgG in relation to the diagnosis of ABPA. It is also the first 5537
comparison of the sensitivity and specificity of the five Aspergillus-‐specific IgG serology 5538
assays for this purpose. The analysis defining the appropriate cut-‐offs for Aspergillus-‐5539
specific IgG to diagnose CPA complicating ABPA is unique. 5540
215
The third and perhaps most important contribution to the field is the first measure of 5541
the prevalence of CPA in an area of high tuberculosis prevalence. The cross-‐sectional 5542
study was a major undertaking that required two surveys two years apart and the 5543
transportation of patients for 700km for CT scan. The author spent a total of 14 months 5544
in Uganda undertaking the study. 5545
5546
The study demonstrated the presence of CPA in 6% of all patients with previously 5547
treated pulmonary tuberculosis. A cross-‐sectional study with convenience sampling was 5548
the only realistic option in light of the financial and time constraints in place. A degree 5549
of selection bias may exist with this method. However, the study does provide the first 5550
clear evidence that CPA is a sufficiently common problem in an area of high tuberculosis 5551
prevalence to be considered a public health issue and provides the first validation of the 5552
predicted global prevalence of 3 million cases. 5553
5554
Evidence from opportunistic testing of stored samples provided by collaborators at 5555
Mulago Hospital, Kampala is also presented. It suggests that Aspergillus infection is 5556
probably present in many patients with current active pulmonary tuberculosis and that 5557
subacute invasive pulmonary aspergillosis is probably the correct diagnosis in a 5558
significant proportion of HIV positive patients currently labeled as ‘smear-‐negative 5559
tuberculosis’. 5560
5561
These results raise major questions about the appropriateness of the diagnostic and 5562
management protocols currently in place for pulmonary tuberculosis in resource poor 5563
settings. They suggest that further studies to accurately define the prevalence of fungal 5564
lung diseases in patients presenting with suspected tuberculosis are now urgently 5565
required. The frequency of raised Aspergillus-‐specific IgG found in each Ugandan patient 5566
group is shown in table 1 below. 5567
5568
5569
5570
5571
5572
5573
216
Table 1 –Frequency of raised Aspergillus disease in Ugandan patient groups 5574
5575
Patient group Frequency of raised Aspergillus-‐specific IgG
Prevalence of CPA
Ugandan healthy controls n = 100
2% Not measured
Ugandans with previously treated pulmonary tuberculosis n = 282
10% 6.5%
Ugandans admitted to hospital with HIV and sub-‐acute lung disease, but no evidence of tuberculosis n = 39
26% Not measured
Ugandans admitted to hospital with proven active pulmonary tuberculosis n = 57
47% Not measured
5576
The results from this thesis have been presented to policy leaders at large global health 5577
institutions. They suggest that CPA is an important neglected disease in global health 5578
terms, due to the number of persons likely to be affected and the high morbidity and 5579
mortality associated with the illness. The author is part of a team planning of a new, 5580
larger, prospective multi-‐centre study to confirm the prevalence of CPA in Kenya. 5581
Further studies will be needed to confirm the prevalence of CPA in other countries 5582
around the world. If these confirm the prevalence of CPA described here it will be 5583
necessary to amend Global policies relating to the investigation and treatment of 5584
tuberculosis to include diagnosis and treatment of CPA. This process could ultimately 5585
result in large-‐scale roll out of testing and treatment of CPA and potentially extend 5586
millions of lives. 5587
5588
217
REFERENCES 5589 5590 1. Soubani AO, Chandrasekar PH. The clinical spectrum of pulmonary aspergillosis. 5591
Chest. 2002;121(6):1988–99. doi:10.1378/chest.121.6.1988. 5592
2. Hope WW, Walsh TJ, Denning DW. The invasive and saprophytic syndromes due 5593
to Aspergillus spp. Med Mycol. 2005;43 Suppl 1:S207–38. 5594
3. Hogan C, Denning DW. Allergic bronchopulmonary aspergillosis and related 5595
allergic syndromes. Semin Respir Crit Care Med. 2011;32(6):682–92. 5596
doi:10.1055/s-‐0031-‐1295716. 5597
4. Agarwal R, Chakrabarti A, Shah A, et al. Allergic bronchopulmonary aspergillosis: 5598
review of literature and proposal of new diagnostic and classification criteria. Clin 5599
Exp Allergy. 2013;43(8):850–73. doi:10.1111/cea.12141. 5600
5. Denning DW, Riniotis K, Dobrashian R, Sambatakou H. Chronic cavitary and 5601
fibrosing pulmonary and pleural aspergillosis: case series, proposed 5602
nomenclature change, and review. Clin Infect Dis. 2003;37 Suppl 3:S265–80. 5603
doi:10.1086/376526. 5604
6. Nam H-‐S, Jeon K, Um S-‐W, et al. Clinical characteristics and treatment outcomes of 5605
chronic necrotizing pulmonary aspergillosis: a review of 43 cases. Int J Infect Dis. 5606
2010;14(6):e479–82. doi:10.1016/j.ijid.2009.07.011. 5607
7. Ohba H, Miwa S, Shirai M, et al. Clinical characteristics and prognosis of chronic 5608
pulmonary aspergillosis. Respir Med. 2012;106(5):724–9. 5609
doi:10.1016/j.rmed.2012.01.014. 5610
8. Jhun BW, Jeon K, Eom JS, et al. Clinical characteristics and treatment outcomes of 5611
chronic pulmonary aspergillosis. Med Mycol. 2013;51(8):811–7. 5612
doi:10.3109/13693786.2013.806826. 5613
9. Agarwal R, Nath A, Aggarwal AN, Gupta D, Chakrabarti A. Aspergillus 5614
hypersensitivity and allergic bronchopulmonary aspergillosis in patients with 5615
acute severe asthma in a respiratory intensive care unit in North India. Mycoses. 5616
2010;53(2):138–43. doi:10.1111/j.1439-‐0507.2008.01680.x. 5617
10. Chakrabarti A, Singh R. The emerging epidemiology of mould infections in 5618
developing countries. Curr Opin Infect Dis. 2011;24(6):521–6. 5619
doi:10.1097/QCO.0b013e32834ab21e. 5620
218
11. Denning D, Pleuvry A, Cole D. Global burden of chronic pulmonary aspergillosis as 5621
a sequel to pulmonary tuberculosis. Bull World Health Organ. 2011;89(12):864–5622
872. doi:10.2471/BLT.11.089441. 5623
12. Denning DW, Pleuvry A, Cole DC. Global burden of chronic pulmonary 5624
aspergillosis complicating sarcoidosis. Eur Respir J. 2013;41(3):621–6. 5625
doi:10.1183/09031936.00226911. 5626
13. Denning DW, Pleuvry A, Cole DC. Global burden of allergic bronchopulmonary 5627
aspergillosis with asthma and its complication chronic pulmonary aspergillosis in 5628
adults. Med Mycol. 2013;51(4):361–70. doi:10.3109/13693786.2012.738312. 5629
14. Smith NL, Denning DW. Underlying conditions in chronic pulmonary aspergillosis 5630
including simple aspergilloma. Eur Respir J. 2011;37(4):865–72. 5631
doi:10.1183/09031936.00054810. 5632
15. Chen Q-‐K, Jiang G-‐N, Ding J-‐A. Surgical treatment for pulmonary aspergilloma: a 5633
35-‐year experience in the Chinese population. Interact Cardiovasc Thorac Surg. 5634
2012;15(1):77–80. doi:10.1093/icvts/ivs130. 5635
16. Ba M, Ciss G, Diarra O, Ndiaye M, Kane O. [Surgical aspects of pulmonary 5636
aspergilloma in 24 patients]. Dakar médical. 2000;45(2):144–6. 5637
17. Wark PAB, Gibson PG, Wilson AJ. Azoles for allergic bronchopulmonary 5638
aspergillosis associated with asthma. Cochrane database Syst Rev. 5639
2004;(3):CD001108. doi:10.1002/14651858.CD001108.pub2. 5640
18. Agarwal R, Vishwanath G, Aggarwal AN, Garg M, Gupta D, Chakrabarti A. 5641
Itraconazole in chronic cavitary pulmonary aspergillosis: a randomised 5642
controlled trial and systematic review of literature. Mycoses. 2013;56(5):559–70. 5643
doi:10.1111/myc.12075. 5644
19. von Eiff M, Roos N, Schulten R, Hesse M, Zühlsdorf M, van de Loo J. Pulmonary 5645
aspergillosis: early diagnosis improves survival. Respiration. 1995;62(6):341–7. 5646
20. Herbrecht R, Denning DW, Patterson TF, et al. Voriconazole versus amphotericin 5647
B for primary therapy of invasive aspergillosis. N Engl J Med. 2002;347(6):408–5648
15. doi:10.1056/NEJMoa020191. 5649
21. Farid S, Mohamed S, Devbhandari M, et al. Results of surgery for chronic 5650
pulmonary aspergillosis, optimal antifungal therapy and proposed high risk 5651
factors for recurrence -‐ a national centre’s experience. J Cardiothorac Surg. 5652
2013;8(1):180. doi:10.1186/1749-‐8090-‐8-‐180. 5653
219
22. Agarwal R. Allergic bronchopulmonary aspergillosis. Chest. 2009;135(3):805–26. 5654
doi:10.1378/chest.08-‐2586. 5655
23. Hope WW, Walsh TJ, Denning DW. Laboratory diagnosis of invasive aspergillosis. 5656
Lancet Infect Dis. 2005;5(10):609–22. doi:10.1016/S1473-‐3099(05)70238-‐3. 5657
24. Pfeiffer CD, Fine JP, Safdar N. Diagnosis of invasive aspergillosis using a 5658
galactomannan assay: a meta-‐analysis. Clin Infect Dis. 2006;42(10):1417–27. 5659
doi:10.1086/503427. 5660
25. Meersseman W, van de Casteele SJ, Wilmer A, Verbeken E, Peetermans WE, van 5661
Wijngaerden E. Invasive aspergillosis in critically ill patients without malignancy. 5662
Am J Respir Crit Care Med. 2004;170(6):621–5. doi:10.1164/rccm.200401-‐093OC. 5663
26. Lu Y, Chen Y-‐Q, Guo Y-‐L, Qin S-‐M, Wu C, Wang K. Diagnosis of invasive fungal 5664
disease using serum (1→3)-‐β-‐D-‐glucan: a bivariate meta-‐analysis. Intern Med. 5665
2011;50(22):2783–91. 5666
27. Thornton CR. Detection of invasive aspergillosis. Adv Appl Microbiol. 5667
2010;70:187–216. doi:10.1016/S0065-‐2164(10)70006-‐X. 5668
28. Arvanitis M, Anagnostou T, Fuchs BB, Caliendo AM, Mylonakis E. Molecular and 5669
nonmolecular diagnostic methods for invasive fungal infections. Clin Microbiol 5670
Rev. 2014;27(3):490–526. doi:10.1128/CMR.00091-‐13. 5671
29. European Centre for Disease Prevention and Control. Risk assessment on the 5672
impact of environmental usage of triazoles on the development and spread of 5673
resistance to medical triazoles in Aspergillus species. Stockholm; 5674
:http://www.ecdc.europa.eu/en/publications/publicat. Accessed on 19/3/15. 5675
30. van den Bergh MF, Verweij PE, Voss A. Epidemiology of nosocomial fungal 5676
infections: invasive aspergillosis and the environment. Diagn Microbiol Infect Dis. 5677
1999;34(3):221–7. 5678
31. Guinea J, Peláez T, Alcalá L, Bouza E. Outdoor environmental levels of Aspergillus 5679
spp. conidia over a wide geographical area. Med Mycol. 2006;44(4):349–56. 5680
doi:10.1080/13693780500488939. 5681
32. Tudela JLR, Sorrell TC. GAFFI Fact Sheet Fungal Keratitis. Glob Action Fund Fungal 5682
Infect. Available at: http://gaffi.org/wp-‐content/uploads/GAFFI-‐Fungal-‐Keratitis-‐5683
briefing-‐2.pdf. Accessed on 15/3/15. 5684
220
33. Thomas PA, Kaliamurthy J. Mycotic keratitis: epidemiology, diagnosis and 5685
management. Clin Microbiol Infect. 2013;19(3):210–20. doi:10.1111/1469-‐5686
0691.12126. 5687
34. Tarazi AE, Al-‐Tawfiq JA, Abdi RF. Fungal malignant otitis externa: pitfalls, 5688
diagnosis, and treatment. Otol Neurotol. 2012;33(5):769–73. 5689
doi:10.1097/MAO.0b013e3182565b46. 5690
35. Bonifaz A, Cruz-‐Aguilar P, Ponce RM. Onychomycosis by molds. Report of 78 5691
cases. Eur J Dermatol. 2007;17(1):70–2. doi:10.1684/ejd.2007.0092. 5692
36. Gianni C, Cerri A, Crosti C. Non-‐dermatophytic onychomycosis . An 5693
underestimated entity ? A study of 51 cases. Mycoses. 2000;43(1-‐2):29–33. 5694
37. Tunnicliffe G, Schomberg L, Walsh S, Tinwell B, Harrison T, Chua F. Airway and 5695
parenchymal manifestations of pulmonary aspergillosis. Respir Med. 5696
2013;107(8):1113–23. doi:10.1016/j.rmed.2013.03.016. 5697
38. Guitard J, Sendid B, Thorez S, Gits M, Hennequin C. Evaluation of a recombinant 5698
antigen-‐based enzyme immunoassay for the diagnosis of noninvasive 5699
aspergillosis. J Clin Microbiol. 2012;50(3):762–5. doi:10.1128/JCM.01257-‐11. 5700
39. Chrdle A, Mustakim S, Bright-‐Thomas RJ, Baxter CG, Felton T, Denning DW. 5701
Aspergillus bronchitis without significant immunocompromise. Ann N Y Acad Sci. 5702
2012;1272(1):73–85. doi:10.1111/j.1749-‐6632.2012.06816.x. 5703
40. Shoseyov D, Brownlee KG, Conway SP, Kerem E. Aspergillus bronchitis in cystic 5704
fibrosis. Chest. 2006;130(1):222–6. doi:10.1378/chest.130.1.222. 5705
41. Chakrabarti A, Denning DW, Ferguson BJ, et al. Fungal rhinosinusitis: a 5706
categorization and definitional schema addressing current controversies. 5707
Laryngoscope. 2009;119(9):1809–18. doi:10.1002/lary.20520. 5708
42. Chakrabarti A, Chatterjee SS, Das A, Shivaprakash MR. Invasive aspergillosis in 5709
developing countries. Med Mycol. 2011;49 Suppl 1:S35–47. 5710
doi:10.3109/13693786.2010.505206. 5711
43. Kemper CA, Hostetler JS, Follansbee SE, et al. Ulcerative and plaque-‐like 5712
tracheobronchitis due to infection with Aspergillus in patients with AIDS. Clin 5713
Infect Dis. 1993;17(3):344–52. 5714
44. Cornet M, Fleury L, Maslo C, Bernard J-‐F, Brücker G. Epidemiology of invasive 5715
aspergillosis in France: a six-‐year multicentric survey in the Greater Paris area. J 5716
Hosp Infect. 2002;51(4):288–96. 5717
221
45. Segal BH, Walsh TJ. Current approaches to diagnosis and treatment of invasive 5718
aspergillosis. Am J Respir Crit Care Med. 2006;173(7):707–17. 5719
doi:10.1164/rccm.200505-‐727SO. 5720
46. Marr KA, Carter RA, Boeckh M, Martin P, Corey L. Invasive aspergillosis in 5721
allogeneic stem cell transplant recipients: changes in epidemiology and risk 5722
factors. Blood. 2002;100(13):4358–66. doi:10.1182/blood-‐2002-‐05-‐1496. 5723
47. Mehrad B, Paciocco G, Martinez FJ, Ojo TC, Iannettoni MD, Lynch JP. Spectrum of 5724
Aspergillus infection in lung transplant recipients: case series and review of the 5725
literature. Chest. 2001;119(1):169–75. 5726
48. de Pauw B, Walsh TJ, Donnelly JP, et al. Revised definitions of invasive fungal 5727
disease from the European Organization for Research and Treatment of 5728
Cancer/Invasive Fungal Infections Cooperative Group and the National Institute 5729
of Allergy and Infectious Diseases Mycoses Study Group (EORTC/MSG). Clin Infect 5730
Dis. 2008;46(12):1813–21. doi:10.1086/588660. 5731
49. Gefter WB, Weingrad TR, Epstein DM, Ochs RH, Miller WT. “Semi-‐invasive” 5732
pulmonary aspergillosis: a new look at the spectrum of aspergillus infections of 5733
the lung. Radiology. 1981;140(2):313–21. 5734
50. Schweer KE, Bangard C, Hekmat K, Cornely O. Chronic pulmonary aspergillosis. 5735
Mycoses. 2013;57:257–70. doi:10.1111/myc.12152. 5736
51. Binder RE, Faling LJ, Pugatch RD, Mahasaen C, Snider GL. Chronic necrotizing 5737
pulmonary aspergillosis: a discrete clinical entity. Medicine (Baltimore). 5738
1982;61(2):109–24. 5739
52. Denning DW, Follansbee SE, Scolaro M, Norris S, Edelstein H, Stevens DA. 5740
Pulmonary aspergillosis in the acquired immunodeficiency syndrome. N Engl J 5741
Med. 1991;324(10):654–62. doi:10.1056/NEJM199103073241003. 5742
53. Hagiwara E, Sekine A, Sato T, et al. [Clinical features of chronic necrotizing 5743
pulmonary aspergillosis treated with voriconazole in patients with chronic 5744
respiratory disease]. Nihon Kokyuki Gakkai Zasshi. 2008;46(11):864–9. 5745
54. Pratap H, Dewan RK, Singh L, Gill S, Vaddadi S. Surgical treatment of pulmonary 5746
aspergilloma: a series of 72 cases. Indian J Chest Dis Allied Sci. 2002;49(1):23–7. 5747
55. Kim YT, Kang MC, Sung SW, Kim JH. Good long-‐term outcomes after surgical 5748
treatment of simple and complex pulmonary aspergilloma. Ann Thorac Surg. 5749
2005;79(1):294–8. doi:10.1016/j.athoracsur.2004.05.050. 5750
222
56. Sagan D, Goździuk K. Surgery for pulmonary aspergilloma in immunocompetent 5751
patients: no benefit from adjuvant antifungal pharmacotherapy. Ann Thorac Surg. 5752
2010;89(5):1603–10. doi:10.1016/j.athoracsur.2010.02.037. 5753
57. Leading International Fungal Education. LIFE chronic pulmonary aspergillosis 5754
factsheet. :http://www.life–worldwide.org/chronic–pulmonary–as. Available at: 5755
http://www.life-‐worldwide.org/chronic-‐pulmonary-‐aspergillosis1/. Accessed on 5756
19/3/15. 5757
58. Felton TW, Baxter C, Moore CB, Roberts SA, Hope WW, Denning DW. Efficacy and 5758
safety of posaconazole for chronic pulmonary aspergillosis. Clin Infect Dis. 5759
2010;51(12):1383–91. doi:10.1086/657306. 5760
59. Shin B, Koh W-‐J, Jeong B-‐H, et al. Serum galactomannan antigen test for the 5761
diagnosis of chronic pulmonary aspergillosis. The Journal of Infection 68, 494–499 5762
(2014). doi:10.1016/j.jinf.2014.01.005. 5763
60. Kitasato Y, Tao Y, Hoshino T, et al. Comparison of Aspergillus galactomannan 5764
antigen testing with a new cut-‐off index and Aspergillus precipitating antibody 5765
testing for the diagnosis of chronic pulmonary aspergillosis. Respirology. 5766
2009;14(5):701–8. doi:10.1111/j.1440-‐1843.2009.01548.x. 5767
61. Denning DW, Park S, Lass-‐Florl C, et al. High-‐frequency triazole resistance found 5768
in nonculturable Aspergillus fumigatus from lungs of patients with chronic fungal 5769
disease. Clin Infect Dis. 2011;52(9):1123–9. doi:10.1093/cid/cir179. 5770
62. Agarwal R, Maskey D, Aggarwal AN, Saikia B, Garg M. Diagnostic performance of 5771
various tests and criteria employed in allergic bronchopulmonary aspergillosis : A 5772
latent class analysis. PLoS One. 2013;8(4):1–7. 5773
doi:10.1371/journal.pone.0061105. 5774
63. Zureik M, Neukirch C, Leynaert B, Liard R, Bousquet J, Neukirch F. Sensitisation to 5775
airborne moulds and severity of asthma: cross sectional study from European 5776
Community respiratory health survey. BMJ. 2002;325(7361):411–4. 5777
64. Denning DW, O’Driscoll BR, Hogaboam CM, Bowyer P, Niven RM. The link 5778
between fungi and severe asthma: a summary of the evidence. Eur Respir J. 5779
2006;27(3):615–26. doi:10.1183/09031936.06.00074705. 5780
65. Denning DW, O’Driscoll BR, Powell G, et al. Randomized controlled trial of oral 5781
antifungal treatment for severe asthma with fungal sensitization: The Fungal 5782
223
Asthma Sensitization Trial (FAST) study. Am J Respir Crit Care Med. 5783
2009;179(1):11–8. doi:10.1164/rccm.200805-‐737OC. 5784
66. Baxter CG, Dunn G, Jones AM, et al. Novel immunologic classification of 5785
aspergillosis in adult cystic fibrosis. J Allergy Clin Immunol. 2013;132:560–66. 5786
doi:10.1016/j.jaci.2013.04.007. 5787
67. Agarwal R, Gupta D, Aggarwal AN, Saxena AK, Chakrabarti A, Jindal SK. Clinical 5788
significance of hyperattenuating mucoid impaction in allergic bronchopulmonary 5789
aspergillosis: an analysis of 155 patients. Chest. 2007;132(4):1183–90. 5790
doi:10.1378/chest.07-‐0808. 5791
68. Aimanianda V, Bayry J, Bozza S, et al. Surface hydrophobin prevents immune 5792
recognition of airborne fungal spores. Nature. 2009;460(7259):1117–21. 5793
doi:10.1038/nature08264. 5794
69. Lass-‐Flörl C, Salzer GM, Schmid T, Rabl W, Ulmer H, Dierichi MP. Pulmonary 5795
Aspergillus colonization in humans and its impact on management of critically ill 5796
patients. Br J Haematol. 1999;104(4):745–7. 5797
70. Park SJ, Mehrad B. Innate immunity to Aspergillus species. Clin Microbiol Rev. 5798
2009;22(4):535–51. doi:10.1128/CMR.00014-‐09. 5799
71. Bardana EJ. Measurement of humoral antibodies to aspergilli. Ann N Y Acad Sci. 5800
1974;221:64–75. 5801
72. Schønheyder H, Andersen P. An indirect immunofluorescence study of antibodies 5802
to Aspergillus fumigatus in sera from children and adults without aspergillosis. 5803
Sabouraudia. 1982;20(1):41–50. 5804
73. van Hoeyveld E, Dupont L, Bossuyt X. Quantification of IgG antibodies to 5805
Aspergillus fumigatus and pigeon antigens by ImmunoCAP technology: an 5806
alternative to the precipitation technique? Clin Chem. 2006;52(9):1785–93. 5807
doi:10.1373/clinchem.2006.067546. 5808
74. Baxter CG, Denning DW, Jones AM, Todd A, Moore CB, Richardson MD. 5809
Performance of two Aspergillus IgG EIA assays compared with the precipitin test 5810
in chronic and allergic aspergillosis. Clin Microbiol Infect. 2013;19(4):E197–204 5811
DOI: 10.1111/1469–0691.12133. doi:10.1111/1469-‐0691.12133. 5812
75. Iwata H, Miwa T, Takagi K. [Tuberculosis sequelae: secondary fungal infections]. 5813
Kekkaku. 1990;65(12):867–71. 5814
224
76. British Tuberculosis Association. Aspergilloma and residual tuberculous cavities -‐ 5815
The results of a resurvey. Tubercle. 1970;51:227–245. 5816
77. Chu C-‐M, Woo PCY, Chong KTK, Leung W-‐S, Chan VL, Yuen K-‐Y. Association of 5817
presence of Aspergillus antibodies with hemoptysis in patients with old 5818
tuberculosis or bronchiectasis but no radiologically visible mycetoma. J Clin 5819
Microbiol. 2004;42(2):665–9. 5820
78. Barton RC, Hobson RP, Denton M, et al. Serologic diagnosis of allergic 5821
bronchopulmonary aspergillosis in patients with cystic fibrosis through the 5822
detection of immunoglobulin G to Aspergillus fumigatus. Diagn Microbiol Infect 5823
Dis. 2008;62(3):287–91. doi:10.1016/j.diagmicrobio.2008.06.018. 5824
79. Sharma GL, Bhatnagar PK, Chattopadhya D, Sarma PU. Analysis of HIV 5825
seropositive thalassemic children for antibodies specific to Aspergillus fumigatus 5826
by luminescent immunoassay. J Clin Lab Anal. 1997;11(6):343–5. 5827
80. Ferreira-‐Da-‐Cruz MF, Wanke B, Pirmez C, Galvão-‐Castro B. Aspergillus fumigatus 5828
fungus ball in hospitalized patients with chronic pulmonary disease. Usefulness of 5829
double immunodiffusion test as a screening procedure. Mem Inst Oswaldo Cruz. 5830
1988;83(3):357–60. 5831
81. Young RC, Bennett JE. Invasive aspergillosis. Absence of detectable antibody 5832
response. Am Rev Respir Dis. 1971;104(5):710–6. 5833
82. Holmberg K, Berdischewsky M, Young LS. Serologic immunodiagnosis of invasive 5834
aspergillosis. J Infect Dis. 1980;141(5):656–64. 5835
83. Matthews R, Burnie JP, Fox A, Tabaqchali S. Immunoblot analysis of serological 5836
responses in invasive aspergillosis. J Clin Pathol. 1985;38(11):1300–3. 5837
84. Kappe R, Schulze-‐Berge A, Sonntag HG. Evaluation of eight antibody tests and one 5838
antigen test for the diagnosis of invasive aspergillosis. Mycoses. 1996;39(1-‐2):13–5839
23. 5840
85. Manso E, Montillo M, De Sio G, D’Amico S, Discepoli G, Leoni P. Value of antigen 5841
and antibody detection in the serological diagnosis of invasive aspergillosis in 5842
patients with hematological malignancies. Eur J Clin Microbiol Infect Dis. 5843
1994;13(9):756–60. 5844
86. Kappe R, Rimek D. [Antibody detection in patients with invasive aspergillosis]. 5845
Mycoses. 2004;47 Suppl 1:55–9. doi:10.1111/j.1439-‐0507.2004.01035.x. 5846
225
87. Mishra SK, Falkenberg S, Masihi KN. Efficacy of enzyme-‐linked immunosorbent 5847
assay in serodiagnosis of aspergillosis. J Clin Microbiol. 1983;17(4):708–10. 5848
88. Weig M, Frosch M, Tintelnot K, et al. Use of recombinant mitogillin for improved 5849
serodiagnosis of Aspergillus fumigatus-‐associated diseases. J Clin Microbiol. 5850
2001;39(5):1721–30. doi:10.1128/JCM.39.5.1721-‐1730.2001. 5851
89. Du C, Wingard JR, Cheng S, Nguyen MH, Clancy CJ. Serum IgG responses against 5852
Aspergillus proteins before hematopoietic stem cell transplantation or 5853
chemotherapy identify patients who develop invasive aspergillosis. Biol Blood 5854
Marrow Transplant. 2012;18(12):1927–34. doi:10.1016/j.bbmt.2012.07.013. 5855
90. Cornillet A, Camus C, Nimubona S, et al. Comparison of epidemiological, clinical, 5856
and biological features of invasive aspergillosis in neutropenic and 5857
nonneutropenic patients: a 6-‐year survey. Clin Infect Dis. 2006;43(5):577–84. 5858
doi:10.1086/505870. 5859
91. Felton TW, Roberts SA, Isalska B, et al. Isolation of Aspergillus species from the 5860
airway of lung transplant recipients is associated with excess mortality. J Infect. 5861
2012;65(4):350–6. doi:10.1016/j.jinf.2012.07.008. 5862
92. Trull AK, Parker J, Warren RE. IgG enzyme linked immunosorbent assay for 5863
diagnosis of invasive aspergillosis: retrospective study over 15 years of 5864
transplant recipients. J Clin Pathol. 1985;38(9):1045–51. 5865
93. Latgé JP. Aspergillus fumigatus and aspergillosis. Clin Microbiol Rev. 5866
1999;12(2):310–50. 5867
94. Centeno-‐Lima S, de Lacerda JM, do Carmo JA, Abecasis M, Casimiro C, Exposto F. 5868
Follow-‐up of anti-‐Aspergillus IgG and IgA antibodies in bone marrow transplanted 5869
patients with invasive aspergillosis. J Clin Lab Anal. 2002;16(3):156–62. 5870
doi:10.1002/jcla.10035. 5871
95. Tomee JF, Mannes GP, van der Bij W, et al. Serodiagnosis and monitoring of 5872
Aspergillus infections after lung transplantation. Ann Intern Med. 5873
1996;125(3):197–201. 5874
96. Herbrecht R, Letscher-‐Bru V, Oprea C, et al. Aspergillus galactomannan detection 5875
in the diagnosis of invasive aspergillosis in cancer patients. J Clin Oncol. 5876
2002;20(7):1898–906. 5877
226
97. Pinel C, Fricker-‐Hidalgo H, Lebeau B, et al. Detection of circulating Aspergillus 5878
fumigatus galactomannan: value and limits of the Platelia test for diagnosing 5879
invasive aspergillosis. J Clin Microbiol. 2003;41(5):2184–6. 5880
98. Longbottom JL, Pepys J. Pulmonary aspergillosis: diagnostic and immunological 5881
significance of antigens and c-‐substance in Aspergillus fumigatus. J Pathol 5882
Bacteriol. 1964;88:141–51. 5883
99. Kostiala AI, Stenius-‐Aarniala B, Alanko K. Analysis of antibodies to Aspergillus 5884
fumigatus antigens by class-‐specific enzyme-‐linked immunosorbent assay in 5885
patients with pulmonary aspergillosis. Diagn Microbiol Infect Dis. 1984;2(1):37–5886
49. 5887
100. Kauffman HF, van der Heide S, Beaumont F, Blok H, de Vries K. Class-‐specific 5888
antibody determination against Aspergillus fumigatus by means of the enzyme-‐5889
linked immunosorbent assay. III. Comparative study: IgG, IgA, IgM ELISA titers, 5890
precipitating antibodies and IgE binding after fractionation of the antigen. Int 5891
Arch Allergy Appl Immunol. 1986;80(3):300–6. 5892
101. Yamamoto S, Toida I, Wada M, Hosojima S, Kudou S. [Serological diagnosis of 5893
pulmonary aspergillosis by ELISA]. Kekkaku. 1989;64(1):15–24. 5894
102. Ninomiya H, Harada S, Harada Y, Takamoto M, Ishibashi T, Shinoda A. [Serological 5895
diagnosis of pulmonary aspergillosis-‐-‐measurement of IgG-‐, IgM-‐ and IgA-‐ 5896
antibodies against Aspergillus fumigatus by means of ELISA]. Kekkaku. 5897
1990;65(4):263–72. 5898
103. Bozza S, Clavaud C, Giovannini G, et al. Immune sensing of Aspergillus fumigatus 5899
proteins, glycolipids, and polysaccharides and the impact on Th immunity and 5900
vaccination. J Immunol. 2009;183(4):2407–14. doi:10.4049/jimmunol.0900961. 5901
104. Brouwer J. Detection of antibodies against Aspergillus fumigatus: comparison 5902
between double immunodiffusion, ELISA and immunoblot analysis. Int Arch 5903
Allergy Appl Immunol. 1988;85(2):244–9. 5904
105. Longbottom JL, Pepys J, Clive FT. Diagnostic precipitin test in Aspergillus 5905
pulmonary mycetoma. Lancet. 1964;1(7333):588–9. 5906
106. Tomee JF, van der Werf TS, Latge JP, Koeter GH, Dubois AE, Kauffman HF. 5907
Serologic monitoring of disease and treatment in a patient with pulmonary 5908
aspergilloma. Am J Respir Crit Care Med. 1995;151(1):199–204. 5909
227
107. Jain LR, Denning DW. The efficacy and tolerability of voriconazole in the 5910
treatment of chronic cavitary pulmonary aspergillosis. J Infect. 2006;52(5):e133–5911
7 DOI: 10.1016/j.jinf.2005.08.022. doi:10.1016/j.jinf.2005.08.022. 5912
108. Camuset J, Nunes H, Dombret M-‐C, et al. Treatment of chronic pulmonary 5913
aspergillosis by voriconazole in nonimmunocompromised patients. Chest. 5914
2007;131(5):1435–41. doi:10.1378/chest.06-‐2441. 5915
109. Cadranel J, Philippe B, Hennequin C, et al. Voriconazole for chronic pulmonary 5916
aspergillosis: a prospective multicenter trial. Eur J Clin Microbiol Infect Dis. 5917
2012;31(11):3231–9. doi:10.1007/s10096-‐012-‐1690-‐y. 5918
110. Chishimba L, Niven RM, Cooley J, Denning DW. Voriconazole and posaconazole 5919
improve asthma severity in allergic bronchopulmonary aspergillosis and severe 5920
asthma with fungal sensitization. J Asthma. 2012;49(4):423–33. 5921
doi:10.3109/02770903.2012.662568. 5922
111. Ryan MW, Marple BF. Allergic fungal rhinosinusitis: diagnosis and management. 5923
Curr Opin Otolaryngol Head Neck Surg. 2007;15(1):18–22. 5924
doi:10.1097/MOO.0b013e328013dbd9. 5925
112. Schubert MS. Allergic fungal sinusitis: pathophysiology, diagnosis and 5926
management. Med Mycol. 2009;47 Suppl 1:S324–30. 5927
doi:10.1080/13693780802314809. 5928
113. Wang JL, Patterson R, Rosenberg M, Roberts M, Cooper BJ. Serum IgE and IgG 5929
antibody activity against Aspergillus fumigatus as a diagnostic aid in allergic 5930
bronchopulmonary aspergillosis. Am Rev Respir Dis. 1978;117(5):917–27. 5931
114. Greenberger PA, Patterson R. Application of enzyme-‐linked immunosorbent assay 5932
(ELISA) in diagnosis of allergic bronchopulmonary aspergillosis. J Lab Clin Med. 5933
1982;99(2):288–93. 5934
115. Stevens DA, Schwartz HJ, Lee JY, et al. A randomized trial of itraconazole in 5935
allergic bronchopulmonary aspergillosis. N Engl J Med. 2000;342(11):756–62. 5936
doi:10.1056/NEJM200003163421102. 5937
116. Wark PAB, Hensley MJ, Saltos N, et al. Anti-‐inflammatory effect of itraconazole in 5938
stable allergic bronchopulmonary aspergillosis: a randomized controlled trial. J 5939
Allergy Clin Immunol. 2003;111(5):952–7. 5940
117. McCarthy DS, Simon G, Hargreave FE. The radiological appearances in allergic 5941
broncho-‐pulmonary aspergillosis. Clin Radiol. 1970;21(4):366–75. 5942
228
118. Menon MP, Das AK. Allergic bronchopulmonary aspergillosis (radiological 5943
aspects). Indian J Chest Dis Allied Sci. 1977;19(4):157–69. 5944
119. Kurup VP, Resnick A, Kalbfleish J, Fink JN. Antibody isotype responses in 5945
Aspergillus-‐induced diseases. J Lab Clin Med. 1990;115(3):298–303. 5946
120. Rosenberg M, Patterson R, Roberts M, Wang J. The assessment of immunologic 5947
and clinical changes occurring during corticosteroid therapy for allergic 5948
bronchopulmonary aspergillosis. Am J Med. 1978;64(4):599–606. 5949
121. Wang JL, Patterson R, Roberts M, Ghory AC. The management of allergic 5950
bronchopulmonary aspergillosis. Am Rev Respir Dis. 1979;120(1):87–92. 5951
122. Denning DW, van Wye JE, Lewiston NJ, Stevens DA. Adjunctive therapy of allergic 5952
bronchopulmonary aspergillosis with itraconazole. Chest. 1991;100(3):813–9. 5953
123. Ouchterlony O. Antigen-‐antibody reactions in gels. Acta Pathol Microbiol Scand. 5954
1949;26(4):507–15. 5955
124. Pepys J, Riddell R, Clayton Y. Human precipitins against common pathogenic and 5956
non-‐pathogenic fungi. Nature. 1959;183:296. 5957
125. Racine R, Winslow GM. IgM in microbial infections: taken for granted? Immunol 5958
Lett. 2009;125(2):79–85. doi:10.1016/j.imlet.2009.06.003. 5959
126. MacKenzie DW. Serological tests. In: Evans E, Richardson MD, eds. Medical 5960
Mycology: a practical approach. 1st ed. Oxford: IRL Press; 1989:201–234. 5961
127. Ikemoto H, Shibata S. Indirect haemagglutination in pulmonary aspergilloma 5962
diagnosis. Sabouraudia. 1973;11(2):167–70. 5963
128. Tönder O, Rödsaethier M. Indirect haemagglutination for demonstration of 5964
antibodies to Aspergillus fumigatus. Acta Pathol Microbiol Scand B Microbiol 5965
Immunol. 1974;82(6):871–8. 5966
129. Smith CE, Saito MT, Beard RR, Kepp RM, Clark RW, Eddie BU. Serological tests in 5967
the diagnosis and prognosis of coccidioidomycosis. Am J Hyg. 1950;52(1):1–21. 5968
130. Sepulveda R, Longbottom JL, Pepys J. Enzyme linked immunosorbent assay 5969
(ELISA) for IgG and IgE antibodies to protein and polysaccharide antigens of 5970
Aspergillus fumigatus. Clin Allergy. 1979;9(4):359–71. 5971
131. Richardson MD, Stubbins JM, Warnock DW. Rapid enzyme-‐linked immunosorbent 5972
assay (ELISA) for Aspergillus fumigatus antibodies. J Clin Pathol. 5973
1982;35(10):1134–7. 5974
229
132. Kim SJ, Chaparas SD, Brown TM, Anderson MC. Characterization of antigens from 5975
Aspergillus fumigatus. II. Fractionation and electrophoretic, immunologic, and 5976
biologic activity. Am Rev Respir Dis. 1978;118(3):553–60. 5977
133. Piechura JE, Huang CJ, Cohen SH, Kidd JM, Kurup VP, Calvanico NJ. Antigens of 5978
Aspergillus fumigatus. II. Electrophoretic and clinical studies. Immunology. 5979
1983;49(4):657–65. 5980
134. Kurup VP, Fink JN, Scribner GH, Falk MJ. Antigenic variability of Aspergillus 5981
fumigatus strains. Microbios. 1977;19(77-‐78):191–204. 5982
135. Kim SJ, Chaparas SD, Buckley HR. Characterization of antigens from Aspergillus 5983
fumigatus. IV. Evaluation of commercial and experimental preparations and 5984
fractions in the detection of antibody in aspergillosis. Am Rev Respir Dis. 5985
1979;120(6):1305–11. 5986
136. Thurston JR, Richard JL, McMillen S. Cultural and serological comparison of ten 5987
strains of Aspergillus fumigatus Fresenius. Mycopathol Mycol Appl. 5988
1973;51(4):327–35. 5989
137. Tran-‐van-‐Ky P, Torck C, Vaucelle T, Floc’h F. [Comparative study on 5990
immunoelectrophoregrams of enzymes of the antigenic extract of Aspergillus 5991
fumigatus revealed by experimental serums and serums of patients with 5992
aspergillosis]. Sabouraudia. 1969;7(2):73–84. 5993
138. Wallenbeck I, Aukrust L, Einarsson R. Antigenic variability of different strains of 5994
Aspergillus fumigatus. Int Arch Allergy Appl Immunol. 1984;73(2):166–72. 5995
139. Blanco JL, García ME, Kurup VP. Immunoreactivity of antigen extracts of 5996
Aspergillus fumigatus isolated from different sources. Rev Iberoam Micol. 5997
1997;14(2):60–2. 5998
140. Kurup V, Kumar A. Immunodiagnosis of aspergillosis. Clin Microbiol Rev. 5999
1991;4(4):439–456. 6000
141. Longbottom JL, Austwick PK. Antigens and allergens of Aspergillus fumigatus. I. 6001
Characterization by quantitative immunoelectrophoretic techniques. J Allergy Clin 6002
Immunol. 1986;78(1 Pt 1):9–17. 6003
142. Kurup VP, Resnick a, Scribner GH, Gunasekaran M, Fink JN. Enzyme profile and 6004
immunochemical characterization of Aspergillus fumigatus antigens. J Allergy Clin 6005
Immunol. 1986;78(6):1166–73. 6006
230
143. Kurup VP, Resnick A, Scribner GH, Kalbfleisch JH, Fink JN. Comparison of antigens 6007
and serological methods in Aspergillus fumigatus antibody detection. Mykosen. 6008
1984;27(1):43–50. 6009
144. Bardana EJ, McClatchy JK, Farr RS, Minden P. The primary interaction of antibody 6010
to components of aspergilli. II. Antibodies in sera from normal persons and from 6011
patients with aspergillosis. J Allergy Clin Immunol. 1972;50(4):222–34. 6012
145. Bardana EJ. The clinical spectrum of aspergillosis-‐-‐part 2: classification and 6013
description of saprophytic, allergic, and invasive variants of human disease. Crit 6014
Rev Clin Lab Sci. 1981;13(2):85–159. 6015
146. Shahid M, Malik A, Bhargava R. Prevalence of aspergillosis in chronic lung 6016
diseases. Indian J Med Microbiol. 2001;19(4):201–5. 6017
147. Severo LC, Geyer GR, Porto NDS, Wagner MB, Londero AT. Pulmonary Aspergillus 6018
niger intracavitary colonization. Report of 23 cases and a review of the literature. 6019
Rev Iberoam Micol. 1997;14(3):104–10. 6020
148. Chaparas ST, Kaufman L, Kim SJ, McLaughlin DW. Characterization of antigens 6021
from Aspergillus fumigatus. V. Reactivity in immunodiffusion tests with serums 6022
from patients with aspergillosis caused by Aspergillus flavus, A. niger, and A 6023
fumigatus. Am Rev Respir Dis. 1980;122(4):647–50. 6024
149. Hearn VM, Wilson E V, Proctor AG, Mackenzie DW. Preparation of Aspergillus 6025
fumigatus antigens and their analysis by two-‐dimensional 6026
immunoelectrophoresis. J Med Microbiol. 1980;13(3):451–8. 6027
150. Singh B, Oellerich M, Kumar R, et al. Immuno-‐reactive molecules identified from 6028
the secreted proteome of Aspergillus fumigatus. J Proteome Res. 2010;9(11):5517–6029
29. doi:10.1021/pr100604x. 6030
151. Kurup VP. Aspergillus antigens: which are important? Med Mycol. 2005;43 Suppl 6031
1:S189–96. 6032
152. Bowyer P, Denning DW. Genomic analysis of allergen genes in Aspergillus spp: the 6033
relevance of genomics to everyday research. Med Mycol. 2007;45(1):17–26. 6034
doi:10.1080/13693780600972907. 6035
153. Chan C-‐M, Woo PCY, Leung ASP, et al. Detection of antibodies specific to an 6036
antigenic cell wall galactomannoprotein for serodiagnosis of Aspergillus 6037
fumigatus aspergillosis. J Clin Microbiol. 2002;40(6):2041–5. 6038
231
154. Sarfati J, Monod M, Recco P, et al. Recombinant antigens as diagnostic markers for 6039
aspergillosis. Diagn Microbiol Infect Dis. 2006;55(4):279–91. 6040
doi:10.1016/j.diagmicrobio.2006.02.002. 6041
155. Latgé JP, Kobayashi H, Debeaupuis JP, et al. Chemical and immunological 6042
characterization of the extracellular galactomannan of Aspergillus fumigatus. 6043
Infect Immun. 1994;62(12):5424–33. 6044
156. Crameri R, Hemmann S, Ismail C, Menz G, Blaser K. Disease-‐specific recombinant 6045
allergens for the diagnosis of allergic bronchopulmonary aspergillosis. Int 6046
Immunol. 1998;10(8):1211–6. 6047
157. Kurup VP, Banerjee B, Hemmann S, Greenberger PA, Blaser K, Crameri R. Selected 6048
recombinant Aspergillus fumigatus allergens bind specifically to IgE in ABPA. Clin 6049
Exp Allergy. 2000;30(7):988–93. 6050
158. International Union of Immunology Societies. IUIS Allergen Nomenclature Home 6051
Page. :http://www.allergen.org/index.php. Available at: 6052
http://www.allergen.org/index.php. Accessed October 6, 2013. 6053
159. Kodzius R, Rhyner C, Konthur Z, et al. Rapid identification of allergen-‐encoding 6054
cDNA clones by phage display and high-‐density arrays. Comb Chem High 6055
Throughput Screen. 2003;6(2):147–54. 6056
160. Crameri R, Lidholm J, Grönlund H, Stüber D, Blaser K, Menz G. Automated specific 6057
IgE assay with recombinant allergens: evaluation of the recombinant Aspergillus 6058
fumigatus allergen I in the Pharmacia Cap System. Clin Exp Allergy. 6059
1996;26(12):1411–9. 6060
161. Arruda LK, Platts-‐Mills TA, Longbottom JL, El-‐Dahr JM, Chapman MD. Aspergillus 6061
fumigatus: identification of 16, 18, and 45 kd antigens recognized by human IgG 6062
and IgE antibodies and murine monoclonal antibodies. J Allergy Clin Immunol. 6063
1992;89(6):1166–76. 6064
162. Bowyer P, Fraczek M, Denning DW. Comparative genomics of fungal allergens and 6065
epitopes shows widespread distribution of closely related allergen and epitope 6066
orthologues. BMC Genomics. 2006;7:251. doi:10.1186/1471-‐2164-‐7-‐251. 6067
163. Nikolaizik WH, Moser M, Crameri R, et al. Identification of allergic 6068
bronchopulmonary aspergillosis in cystic fibrosis patients by recombinant 6069
Aspergillus fumigatus I/a-‐specific serology. Am J Respir Crit Care Med. 6070
1995;152(2):634–9. doi:10.1164/ajrccm.152.2.7633719. 6071
232
164. Hemmann S, Nikolaizik WH, Schöni MH, Blaser K, Crameri R. Differential IgE 6072
recognition of recombinant Aspergillus fumigatus allergens by cystic fibrosis 6073
patients with allergic bronchopulmonary aspergillosis or Aspergillus allergy. Eur J 6074
Immunol. 1998;28(4):1155–60. 6075
165. Nikolaizik WH, Weichel M, Blaser K, Crameri R. Intracutaneous tests with 6076
recombinant allergens in cystic fibrosis patients with allergic bronchopulmonary 6077
aspergillosis and Aspergillus allergy. Am J Respir Crit Care Med. 2002;165(7):916–6078
21. doi:10.1164/ajrccm.165.7.2109008. 6079
166. Kurup VP, Knutsen AP, Moss RB, Bansal NK. Specific antibodies to recombinant 6080
allergens of Aspergillus fumigatus in cystic fibrosis patients with ABPA. Clin Mol 6081
allergy. 2006;4(1):11. doi:10.1186/1476-‐7961-‐4-‐11. 6082
167. Denikus N, Orfaniotou F, Wulf G, Lehmann PF, Monod M, Reichard U. Fungal 6083
antigens expressed during invasive aspergillosis. Infect Immun. 2005;73(8):4704–6084
13. doi:10.1128/IAI.73.8.4704-‐4713.2005. 6085
168. Dee TH. Detection of Aspergillus fumigatus serum precipitins by 6086
counterimmunoelectrophoresis. J Clin Microbiol. 1975;2(6):482–5. 6087
169. Kauffman HF, Beaumont F, Meurs H, van der Heide S, de Vries K. Comparison of 6088
antibody measurements against Aspergillus fumigatus by means of double-‐6089
diffusion and enzyme-‐linked immunosorbent assay (ELISA). J Allergy Clin 6090
Immunol. 1983;72(3):255–61. 6091
170. Kurup VP, Fink JN. Evaluation of methods to detect antibodies against Aspergillus 6092
fumigatus. Am J Clin Pathol. 1978;69(4):414–7. 6093
171. Warnock DW. Detection of Aspergillus fumigatus precipitins: a comparison of 6094
counter immunoelectrophoresis and double diffusion. J Clin Pathol. 6095
1977;30(4):388–9. 6096
172. Faux JA, Shale DJ, Lane DJ. Precipitins and specific IgG antibody to Aspergillus 6097
fumigatus in a chest unit population. Thorax. 1992;47(1):48–52. 6098
173. Gugnani HC, Reijula KE, Kurup VP, Fink JN. Detection of IgG and IgE antibodies to 6099
Aspergillus fumigatus in human sera by immunogold assay. Mycopathologia. 6100
1990;109(1):33–40. 6101
174. Schønheyder H, Andersen P, Stenderup A. Serum antibodies to Aspergillus 6102
fumigatus in patients with pulmonary aspergillosis detected by 6103
immunofluorescence. Acta Pathol Microbiol Immunol Scand B. 1982;90(4):273–9. 6104
233
175. Avila R. Immunological study of pulmonary aspergilloma. Thorax. 6105
1968;23(2):144–52. 6106
176. Warren CP, Tai E, Batten JC, Hutchcroft BJ, Pepys J. Cystic fibrosis-‐-‐immunological 6107
reactions to A. fumigatus and common allergens. Clin Allergy. 1975;5(1):1–12. 6108
177. Mackenzie DW, Philpot CM. Counterimmunoelectrophoresis as a routine 6109
mycoserological procedure. Mycopathologia. 1975;57(1):1–7. 6110
178. Malo JL, Longbottom J, Mitchell J, Hawkins R, Pepys J. Studies in chronic allergic 6111
bronchopulmonary aspergillosis 3 Immunological findings. Thorax. 1977;32:269–6112
274. 6113
179. Mehta SK, Sandhu RS. Efficacy of counterimmunoelectrophoresis in the detection 6114
of fungal antibodies in allergic bronchopulmonary mycoses. Zentralbl Bakteriol A. 6115
1980;247(4):537–42. 6116
180. van Toorenenbergen AW. Between-‐laboratory quality control of automated 6117
analysis of IgG antibodies against Aspergillus fumigatus. Diagn Microbiol Infect Dis. 6118
2012;74(3):278–81. doi:10.1016/j.diagmicrobio.2012.07.002. 6119
181. Longbottom JL. Immunological aspects of infection and allergy due to Aspergillus 6120
species. Mykosen Suppl. 1978;1:207–17. 6121
182. Smits WL, Letz KL, Evans TS, Giese JK. Evaluating the response of patients 6122
undergoing both allergy skin testing and in vitro allergy testing with the 6123
ImmunoCAP Technology System. J Am Acad Nurse Pract. 2003;15(9):415–23. 6124
183. O’Driscoll BR, Powell G, Chew F, et al. Comparison of skin prick tests with specific 6125
serum immunoglobulin E in the diagnosis of fungal sensitization in patients with 6126
severe asthma. Clin Exp Allergy. 2009;39(11):1677–83. doi:10.1111/j.1365-‐6127
2222.2009.03339.x. 6128
184. Wood RA, Segall N, Ahlstedt S, Williams PB. Accuracy of IgE antibody laboratory 6129
results. Ann Allergy Asthma Immunol. 2007;99(1):34–41. doi:10.1016/S1081-‐6130
1206(10)60618-‐7. 6131
185. Greenwald JL, Burstein GR, Pincus J, Branson B. A rapid review of rapid HIV 6132
antibody tests. Curr Infect Dis Rep. 2006;8(2):125–131. doi:10.1007/s11908-‐006-‐6133
0008-‐6. 6134
186. Murray CK, Gasser RA, Magill AJ, Miller RS. Update on rapid diagnostic testing for 6135
malaria. Clin Microbiol Rev. 2008;21(1):97–110. doi:10.1128/CMR.00035-‐07. 6136
234
187. Thornton CR. Development of an immunochromatographic lateral-‐flow device for 6137
rapid serodiagnosis of invasive aspergillosis. Clin vaccine Immunol. 6138
2008;15(7):1095–105. doi:10.1128/CVI.00068-‐08. 6139
188. Held J, Schmidt T, Thornton CR, Kotter E, Bertz H. Comparison of a novel 6140
Aspergillus lateral-‐flow device and the Platelia(®) galactomannan assay for the 6141
diagnosis of invasive aspergillosis following haematopoietic stem cell 6142
transplantation. Infection. 2013;41(6):1163–9. doi:10.1007/s15010-‐013-‐0472-‐5. 6143
189. Wiederhold NP, Thornton CR, Najvar LK, Kirkpatrick WR, Bocanegra R, Patterson 6144
TF. Comparison of lateral flow technology and galactomannan and (1-‐>3)-‐beta-‐D-‐6145
glucan assays for detection of invasive pulmonary aspergillosis. Clin Vaccine 6146
Immunol. 2009;16(12):1844–6. doi:10.1128/CVI.00268-‐09. 6147
190. Prattes J, Flick H, Prüller F, et al. Novel tests for diagnosis of invasive aspergillosis 6148
in patients with underlying respiratory diseases. Am J Respir Crit Care Med. 6149
2014:1–27. doi:10.1164/rccm.201407-‐1275OC. 6150
191. Baddley JW, Marr KA, Andes DR, et al. Patterns of susceptibility of Aspergillus 6151
isolates recovered from patients enrolled in the Transplant-‐Associated Infection 6152
Surveillance Network. J Clin Microbiol. 2009;47(10):3271–5. 6153
doi:10.1128/JCM.00854-‐09. 6154
192. Kurhade AM, Deshmukh JM, Fule RP, Chande C, Akulwar S. Mycological and 6155
serological study of pulmonary aspergillosis in central India. Indian J Med 6156
Microbiol. 2002;20(3):141–4. 6157
193. Michael RC, Michael JS, Ashbee RH, Mathews MS. Mycological profile of fungal 6158
sinusitis: An audit of specimens over a 7-‐year period in a tertiary care hospital in 6159
Tamil Nadu. Indian J Pathol Microbiol. 2008;51(4):493–6. 6160
194. Prateek S, Banerjee G, Gupta P, Singh M, Goel MM, Verma V. Fungal rhinosinusitis: 6161
a prospective study in a university hospital of Uttar Pradesh. Indian J Med 6162
Microbiol. 2013;31(3):266–9. doi:10.4103/0255-‐0857.115634. 6163
195. Plaignaud M. Observation sur fungus de sinus maxillaire. J chirology. 1791;1:11–6164
116. 6165
196. Bennett JH. On the parasitic vegetable structures found growing in living animals. 6166
1843; Transactions of the Royal society of Edinburgh. 1842:15;277-‐294. 6167
197. British Tuberculosis Association. Aspergillus in persistent lung cavities after 6168
tuberculosis. Tubercle. 1968;49:1. 6169
235
198. Al-‐Shair K, Atherton GT, Harris C, Ratcliffe L, Newton PJ, Denning DW. Long-‐term 6170
antifungal treatment improves health status in patients with chronic pulmonary 6171
aspergillosis: A longitudinal analysis. Clin Infect Dis. 2013;57(6):828–35. 6172
doi:10.1093/cid/cit411. 6173
199. Miloshev B, Davidson CM, Gentles JC, Sandison AT. Aspergilloma of paranasal 6174
sinuses and orbit in Northern Sudanese. Lancet. 1966;1(7440):746–7. 6175
200. Fasunla J, Ibekwe T, Onakoya P. Otomycosis in western Nigeria. Mycoses. 6176
2008;51(1):67–70. doi:10.1111/j.1439-‐0507.2007.01441.x. 6177
201. Jacobs P, Friedberg EC, Neikirk V. Pulmonary aspergillosis in a white South 6178
African farmer. a case report. Cent Afr J Med. 1965;11:155–60. 6179
202. Benatar SR. Aspergillus infection in the Western Cape. South African Med J. 6180
1977;51(10):297–305. 6181
203. Tiendrebeogo H, Sangare SI, Roudaut M, Schmidt D, Assale N. [One hundred and 6182
one cases of pulmonary aspergillosis in Ivory Coast (author’s transl)]. Médecine 6183
Trop. 1982;42(1):47–52. 6184
204. Domoua K, N’Dhatz M, Coulibaly G, et al. [Hemoptysis: main etiologies observed in 6185
a pneumology department in Africa]. Rev Pneumol Clin. 1994;50(2):59–62. 6186
205. Fahal AH, Azziz KA, Saliman SH, Galib H V, Mahgoub S. Dual infection with 6187
mycetoma and tuberculosis: report of two cases. East Afr Med J. 6188
1995;72(11):749–50. 6189
206. Aderaye G, Jajaw A. Bilateral pulmonary aspergilloma: case report. East Afr Med J. 6190
1996;73(7):487–8. 6191
207. Cox J, Lukande R. Needle autopsy to establish the cause of death in HIV-‐infected 6192
hospitalized adults in Uganda: A comparison to complete autopsy. J Acquir 6193
Immune Defic Syndr. 2014;67(2):169–76. doi:10.1097/QAI.0000000000000290. 6194
208. Auregan G, Chakib S, Etienne B, Ibrahim M, Simitzis AM. [Aspergilloma of the lung 6195
in Djibouti]. Ann Soc Belg Med Trop (1920). 1988;68(4):357–64. 6196
209. Lesbordes JL, Meunier DM, Georges-‐Courbot MC. [Mycoses in AIDS in the Central 6197
African Republic]. Médecine Trop. 1986;46(3):257–60. 6198
210. Pohl C, Jugheli L, Haraka F, Mfinanga E, Said K, Reither K. Pulmonary 6199
aspergilloma: a treatment challenge in sub-‐Saharan Africa. PLoS Negl Trop Dis. 6200
2013;7(10):e2352. doi:10.1371/journal.pntd.0002352. 6201
236
211. Adebonojo SA, Osinowo O, Adebo O. Lung abscess: a review of three-‐years’ 6202
experience at the University College Hospital, Ibadan. J Natl Med Assoc. 6203
1979;71(1):39–43. 6204
212. Ade SS, Touré NO, Ndiaye A, et al. [Epidemiological, clinical, therapeutic and 6205
evolutionary aspects of pulmonary aspergilloma in Dakar]. Rev Mal Respir. 6206
2011;28(3):322–7. doi:10.1016/j.rmr.2010.09.032. 6207
213. Meawed TE, M AF, Gamry RM El, Shaker A, Gazzar A El, Ragheb AS. Frequency of 6208
pulmonary fungal infection in Egyptian patients with re-‐treatment pulmonary 6209
tuberculosis and its clinical and radiological significance. J Am Sci. 2012;8(5):683–6210
691. 6211
214. Zendah I, Ayadi-‐Kaddour A, Gharsalli H, Khattab A, Ghedira H. Coexistence of 6212
active tuberculosis, cancer and aspergilloma of the lung. Tunis Med. 6213
2011;89(4):407. 6214
215. WHO | Global tuberculosis report 2014. 6215
http://www.who.int/tb/publications/global_report/en/. Accessed on 19/3/15. 6216
216. Loddenkemper R, Lipman M, Zumla A. Clinical aspects of adult tuberculosis. Cold 6217
Spring Harb Perspect Med. 2015. doi:10.1101/cshperspect.a017848. 6218
217. Andreu J, Cáceres J, Pallisa E, Martinez-‐Rodriguez M. Radiological manifestations 6219
of pulmonary tuberculosis. Eur J Radiol. 2004;51(2):139–49. 6220
doi:10.1016/j.ejrad.2004.03.009. 6221
218. Greene R. The radiological spectrum of pulmonary aspergillosis. Med Mycol. 6222
2005;43 Suppl 1:S147–54. 6223
219. Roberts CM, Citron KM, Strickland B. Intrathoracic aspergilloma: role of CT in 6224
diagnosis and treatment. Radiology. 1987;165(1):123–8. 6225
doi:10.1148/radiology.165.1.3628758. 6226
220. GAFFI Fact Sheet: TB and its sequela chronic pulmonary aspergillosis (CPA). 6227
Global Action Fund for Fungal Infections. 2013. http://www.gaffi.org/wp-‐6228
content/uploads/Chronic-‐pulmonary-‐aspergillosis-‐Fact-‐Sheet.pdf. Accessed on 6229
19/3/15. 6230
221. Iwata H, Miwa T, Takagi K. [Pulmonary tuberculosis and mycotic infection-‐-‐6231
clinical and serological diagnosis]. Kekkaku. 1989;64(1):7–13. 6232
237
222. Fullerton DG, Bruce N, Gordon SB. Indoor air pollution from biomass fuel smoke 6233
is a major health concern in the developing world. Trans R Soc Trop Med Hyg. 6234
2008;102(9):843–51. doi:10.1016/j.trstmh.2008.05.028. 6235
223. Greenberg AK, Knapp J, Rom WN, Addrizzo-‐Harris DJ. Clinical presentation of 6236
pulmonary mycetoma in HIV-‐infected patients. Chest. 2002;122(3):886–92. 6237
224. Addrizzo-‐Harris DJ, Harkin TJ, McGuinness G, Naidich DP, Rom WN. Pulmonary 6238
aspergilloma and AIDS. A comparison of HIV-‐infected and HIV-‐negative 6239
individuals. Chest. 1997;111(3):612–8. 6240
225. Leung AN, Brauner MW, Gamsu G, et al. Pulmonary tuberculosis: comparison of 6241
CT findings in HIV-‐seropositive and HIV-‐seronegative patients. Radiology. 6242
1996;198(3):687–91. doi:10.1148/radiology.198.3.8628855. 6243
226. Perlman DC, El-‐Sadr WM, Nelson ET, et al. Variation of chest radiographic 6244
patterns in pulmonary tuberculosis by degree of human immunodeficiency 6245
virus-‐related immunosuppression. Clin Infect Dis. 1997;25(2):242–246. 6246
doi:10.1086/514546. 6247
227. Picon PD, Caramori MLA, Bassanesi SL, et al. Diferenças na apresentação clínico-‐6248
radiológica da tuberculose intratorácica segundo a presença ou não de infecção 6249
por HIV. J Bras Pneumol. 2007;33(4):429–436. doi:10.1590/S1806-‐6250
37132007000400012. 6251
228. Vazquez JA. Therapeutic options for the management of oropharyngeal and 6252
esophageal candidiasis in HIV/AIDS patients. HIV Clin Trials. 2000;1(1):47–59. 6253
229. Batungwanayo J, Taelman H, Bogaerts J, et al. Pulmonary cryptococcosis 6254
associated with HIV-‐1 infection in Rwanda: a retrospective study of 37 cases. 6255
AIDS. 1994;8(9):1271–6. 6256
230. Boulware DR, Meya DB, Muzoora C, et al. Timing of antiretroviral therapy after 6257
diagnosis of cryptococcal meningitis. N Engl J Med. 2014;370(26):2487–98. 6258
doi:10.1056/NEJMoa1312884. 6259
231. van Oosterhout JJG, Laufer MK, Perez MA, et al. Pneumocystis pneumonia in HIV-‐6260
positive adults, Malawi. Emerg Infect Dis. 2007;13(2):325–8. 6261
doi:10.3201/eid1302.060462. 6262
232. Baily GG, Robertson VJ, Neill P, Garrido P, Levy LF. Blastomycosis in Africa: 6263
clinical features, diagnosis, and treatment. Rev Infect Dis. 1991;13(5):1005–8. 6264
238
233. Pappagianis D. Epidemiology of coccidioidomycosis. Curr Top Med Mycol. 6265
1988;2:199–238. 6266
234. McKinsey DS, McKinsey JP. Pulmonary histoplasmosis. Semin Respir Crit Care Med. 6267
2011;32(6):735–44. doi:10.1055/s-‐0031-‐1295721. 6268
235. Miller WT, Sais GJ, Frank I, et al. Pulmonary aspergillosis in patients with AIDS. 6269
Clinical and radiographic correlations. Chest. 1994;105(1):37–34. 6270
236. Khoo SH, Denning DW. Invasive aspergillosis in patients with AIDS. Clin Infect Dis. 6271
1994;19 Suppl 1(1):S41–8. 6272
237. Ellis M, Gupta S, Galant S, et al. Impaired neutrophil function in patients with AIDS 6273
or AIDS-‐related complex: a comprehensive evaluation. J Infect Dis. 6274
1988;158(6):1268–76. 6275
238. Libanore M, Prini E, Mazzetti M, et al. Invasive aspergillosis in Italian AIDS 6276
patients. Infection. 2002;30(6):341–5. doi:10.1007/s15010-‐002-‐2033-‐1. 6277
239. Lanjewar DN, Duggal R. Pulmonary pathology in patients with AIDS: an autopsy 6278
study from Mumbai. HIV Med. 2001;2(4):266–71. 6279
240. Katano H, Hishima T, Mochizuki M, et al. The prevalence of opportunistic 6280
infections and malignancies in autopsied patients with human immunodeficiency 6281
virus infection in Japan. BMC Infect Dis. 2014;14(1):229. doi:10.1186/1471-‐2334-‐6282
14-‐229. 6283
241. World Health Organization. Improving the diagnosis and treatment of smear-‐6284
negative pulmonary and extrapulmonary tuberculosis among adults and 6285
adolescents; Recommendations for HIV-‐prevalent and resource-‐constrained 6286
settings. 2007. Available at: 6287
http://whqlibdoc.who.int/hq/2007/WHO_HTM_TB_2007.379_eng.pdf?ua=1. 6288
242. Getahun H, Harrington M, O’Brien R, Nunn P. Diagnosis of smear-‐negative 6289
pulmonary tuberculosis in people with HIV infection or AIDS in resource-‐6290
constrained settings: informing urgent policy changes. Lancet. 6291
2007;369(9578):2042–9. doi:10.1016/S0140-‐6736(07)60284-‐0. 6292
243. Nakiyingi L, Bwanika JM, Kirenga B, et al. Clinical predictors and accuracy of 6293
empiric tuberculosis treatment among sputum smear-‐negative HIV-‐infected adult 6294
TB suspects in Uganda. PLoS One. 2013;8(9):e74023. 6295
doi:10.1371/journal.pone.0074023. 6296
239
244. Davis JL, Worodria W, Kisembo H, et al. Clinical and radiographic factors do not 6297
accurately diagnose smear-‐negative tuberculosis in HIV-‐infected inpatients in 6298
Uganda: a cross-‐sectional study. PLoS One. 2010;5(3):e9859. 6299
doi:10.1371/journal.pone.0009859. 6300
245. Henegar C, Behets F, van den Driessche K, et al. Mortality among tuberculosis 6301
patients in the Democratic Republic of Congo. Int J Tuberc Lung Dis. 6302
2012;16(9):1199–204. doi:10.5588/ijtld.11.0613. 6303
246. Kang’ombe CT, Harries AD, Ito K, et al. Long-‐term outcome in patients registered 6304
with tuberculosis in Zomba, Malawi: mortality at 7 years according to initial HIV 6305
status and type of TB. Int J Tuberc Lung Dis. 2004;8(7):829–36. 6306
247. Denning DW, Ribaud P, Milpied N, et al. Efficacy and safety of voriconazole in the 6307
treatment of acute invasive aspergillosis. Clin Infect Dis. 2002;34(5):563–71. 6308
doi:10.1086/324620. 6309
248. Lin SJ, Schranz J, Teutsch SM. Aspergillosis case-‐fatality rate: systematic review of 6310
the literature. Clin Infect Dis. 2001;32(3):358–66. doi:10.1086/318483. 6311
249. Denning DW. Therapeutic outcome in invasive aspergillosis. Clin Infect Dis. 6312
1996;23(3):608–15. 6313
250. Walsh TJ, Anaissie EJ, Denning DW, et al. Treatment of aspergillosis: clinical 6314
practice guidelines of the Infectious Diseases Society of America. Clin Infect Dis. 6315
2008;46(3):327–60. doi:10.1086/525258. 6316
251. Denning D, Cadranel J, Beigelman-‐Aubry C, et al. Clinical guidelines for the 6317
diagnosis and management of chronic pulmonary aspergillosis. Manuscr Submitt. 6318
2015. 6319
252. Daly RC, Pairolero PC, Piehler JM, Trastek VF, Payne WS, Bernatz PE. Pulmonary 6320
aspergilloma. Results of surgical treatment. J Thorac Cardiovasc Surg. 6321
1986;92(6):981–8. 6322
253. Kohno S, Izumikawa K, Kakeya H, et al. Clinical efficacy and safety of micafungin 6323
in Japanese patients with chronic pulmonary aspergillosis: a prospective 6324
observational study. Med Mycol. 2011;49(7):688–93. 6325
doi:10.3109/13693786.2011.561369. 6326
254. Sambatakou H, Dupont B, Lode H, Denning DW. Voriconazole treatment for 6327
subacute invasive and chronic pulmonary aspergillosis. Am J Med. 6328
2006;119(6):527.e17–24. doi:10.1016/j.amjmed.2005.11.028. 6329
240
255. Camara B, Reymond E, Saint-‐Raymond C, et al. Characteristics and outcomes of 6330
chronic pulmonary aspergillosis: a retrospective analysis of a tertiary hospital 6331
registry. Clin Respir J. 2015;9(1):65–73. doi:10.1111/crj.12105. 6332
256. Baxter CG, Marshall A, Roberts M, Felton TW, Denning DW. Peripheral neuropathy 6333
in patients on long-‐term triazole antifungal therapy. J Antimicrob Chemother. 6334
2011;66(9):2136–9. doi:10.1093/jac/dkr233. 6335
257. Lowes D. Baseline parameters of survival in chronic pulmonary aspergillosis. In: 6336
6th Advances Against Aspergillosis.; 2014:poster 23. 6337
258. Kosmidis C, Denning DW. The clinical spectrum of pulmonary aspergillosis. 6338
Thorax. 2014;70(3):270–277. doi:10.1136/thoraxjnl-‐2014-‐206291. 6339
259. Muldoon EG, Page ID, Bishop P, Denning DW. Aspergillus pulmonary nodules; 6340
presentation, radiology and histology features. In: 6th Advances Against 6341
Aspergillosis.; 2014:poster 100. 6342
260. Fraczek MG, Kirwan MB, Moore CB, Morris J, Denning DW, Richardson MD. 6343
Volume dependency for culture of fungi from respiratory secretions and 6344
increased sensitivity of Aspergillus quantitative PCR. Mycoses. 2013;57(2):69–78. 6345
doi:10.1111/myc.12103. 6346
261. Johnson GL, Sarker S-‐J, Hill K, et al. Significant decline in galactomannan signal 6347
during storage of clinical serum samples. Int J Mol Sci. 2013;14(7):12970–7. 6348
doi:10.3390/ijms140712970. 6349
262. Hendriks J, Stals C, Versteilen A, et al. Stability studies of binding and functional 6350
anti-‐vaccine antibodies. Bioanalysis. 2014;6(10):1385–93. doi:10.4155/bio.14.96. 6351
263. Michaut L, Laurent N, Kentsch K, Spindeldreher S, Deckert-‐Salva F. Stability of 6352
anti-‐immunotherapeutic antibodies in frozen human serum samples. Bioanalysis. 6353
2014;6(10):1395–407. doi:10.4155/bio.14.97. 6354
264. Nakamoto K, Takayanagi N, Kanauchi T, Ishiguro T, Yanagisawa T, Sugita Y. 6355
Prognostic factors in 194 patients with chronic necrotizing pulmonary 6356
aspergillosis. Intern Med. 2013;52(7):727–34. 6357
265. Stout JE, Kosinski AS, Hamilton CD, et al. Effect of improving the quality of 6358
radiographic interpretation on the ability to predict pulmonary tuberculosis 6359
relapse. Acad Radiol. 2010;17(13):157–162. doi:10.1016/j.acra.2009.08.013. 6360
266. Kisembo HN, Boon S Den, Davis JL, et al. Chest radiographic findings of pulmonary 6361
tuberculosis in severely immunocompromised patients with the human 6362
241
immunodeficiency virus. Br J Radiol. 2012;85(1014):e130–9. 6363
doi:10.1259/bjr/70704099. 6364
267. Chakrabarti A, Slavin MA. Endemic fungal infections in the Asia-‐Pacific region. 6365
Med Mycol. 2011;49(4):337–44. doi:10.3109/13693786.2010.551426. 6366
268. Cottle LE, Gkrania-‐Klotsas E, Williams HJ, et al. A multinational outbreak of 6367
histoplasmosis following a biology field trip in the ugandan rainforest. J Travel 6368
Med. 2013;20(2):83–7. doi:10.1111/jtm.12012. 6369
269. Brouwer J. Cross-‐reactivity between Aspergillus fumigatus and Penicillium. Int 6370
Arch Allergy Immunol. 1996;110(2):166–73. 6371
270. Froudist JH, Harnett GB, McAleer R. Comparison of immunodiffusion and enzyme 6372
linked immunosorbent assay for antibodies to four Aspergillus species. J Clin 6373
Pathol. 1989;42(11):1215–21. 6374
271. Pai NP, Tulsky JP, Cohan D, Colford JM, Reingold AL. Rapid point-‐of-‐care HIV 6375
testing in pregnant women: a systematic review and meta-‐analysis. Trop Med Int 6376
Health. 2007;12(2):162–73. doi:10.1111/j.1365-‐3156.2006.01812.x. 6377
272. Mugyenyi P, Walker AS, Hakim J, et al. Routine versus clinically driven laboratory 6378
monitoring of HIV antiretroviral therapy in Africa (DART): a randomised non-‐6379
inferiority trial. Lancet. 2010;375(9709):123–31. doi:10.1016/S0140-‐6380
6736(09)62067-‐5. 6381
273. Cribier B, Rey D, Schmitt C, Lang JM, Kirn A, Stoll-‐Keller F. High hepatitis C 6382
viraemia and impaired antibody response in patients coinfected with HIV. AIDS. 6383
1995;9(10):1131–6. 6384
274. Ballet JJ, Sulcebe G, Couderc LJ, et al. Impaired anti-‐pneumococcal antibody 6385
response in patients with AIDS-‐related persistent generalized lymphadenopathy. 6386
Clin Exp Immunol. 1987;68(3):479–87. 6387
275. Rey D, Krantz V, Partisani M, et al. Increasing the number of hepatitis B vaccine 6388
injections augments anti-‐HBs response rate in HIV-‐infected patients. Effects on 6389
HIV-‐1 viral load. Vaccine. 2000;18(13):1161–1165. doi:10.1016/S0264-‐6390
410X(99)00389-‐8. 6391
276. Yoon C, Cattamanchi A, Davis JL, et al. Impact of Xpert MTB/RIF testing on 6392
tuberculosis management and outcomes in hospitalized patients in Uganda. PLoS 6393
One. 2012;7(11):e48599. doi:10.1371/journal.pone.0048599. 6394
242
277. Yoo SD, Cattamanchi A, Den Boon S, et al. Clinical significance of normal chest 6395
radiographs among HIV-‐seropositive patients with suspected tuberculosis in 6396
Uganda. Respirology. 2011;16(5):836–41. doi:10.1111/j.1440-‐6397
1843.2011.01981.x. 6398
278. Iwai S, Huang D, Fong S, et al. The lung microbiome of Ugandan HIV-‐infected 6399
pneumonia patients is compositionally and functionally distinct from that of San 6400
Franciscan patients. PLoS One. 2014;9(4):e95726. 6401
doi:10.1371/journal.pone.0095726. 6402
279. Lawn SD, Mwaba P, Bates M, et al. Advances in tuberculosis diagnostics: the Xpert 6403
MTB/RIF assay and future prospects for a point-‐of-‐care test. Lancet Infect Dis. 6404
2013;13(4):349–61. doi:10.1016/S1473-‐3099(13)70008-‐2. 6405
280. Greene RE, Schlamm HT, Oestmann J-‐W, et al. Imaging findings in acute invasive 6406
pulmonary aspergillosis: clinical significance of the halo sign. Clin Infect Dis. 6407
2007;44(3):373–9. doi:10.1086/509917. 6408
281. Boonsarngsuk V, Niyompattama A, Teosirimongkol C, Sriwanichrak K. False-‐6409
positive serum and bronchoalveolar lavage Aspergillus galactomannan assays 6410
caused by different antibiotics. Scand J Infect Dis. 2010;42(6-‐7):461–8. 6411
doi:10.3109/00365541003602064. 6412
282. Hmama Z, Peña-‐Díaz S, Joseph S, Av-‐Gay Y. Immunoevasion and 6413
immunosuppression of the macrophage by Mycobacterium tuberculosis. Immunol 6414
Rev. 2015;264(1):220–32. doi:10.1111/imr.12268. 6415
283. Camuset J, Nunes H, Dombret M-‐C, et al. Treatment of chronic pulmonary 6416
aspergillosis by voriconazole in nonimmunocompromised patients. Chest. 6417
2007;131(5):1435–41. doi:10.1378/chest.06-‐2441. 6418
284. Agarwal R, Aggarwal AN, Gupta D, Jindal SK. Aspergillus hypersensitivity and 6419
allergic bronchopulmonary aspergillosis in patients with bronchial asthma: 6420
systematic review and meta-‐analysis. Int J Tuberc lung Dis. 2009;13(8):936–44. 6421
285. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating 6422
characteristic (ROC) curve. Radiology. 1982;143(1):29–36. 6423
doi:10.1148/radiology.143.1.7063747. 6424
286. Metz CE. Basic principles of ROC analysis. Semin Nucl Med. 1978;8(4):283–98. 6425
243
287. Rosenberg IL, Greenberger PA. Allergic bronchopulmonary aspergillosis and 6426
aspergilloma. Long-‐term follow-‐up without enlargement of a large multiloculated 6427
cavity. Chest. 1984;85(1):123–5. 6428
288. Wong SJ, Seligman SJ. Long-‐term stability of West Nile virus IgM and IgG 6429
antibodies in diluted sera stored at 4 degrees C. Ann N Y Acad Sci. 2001;951:369–6430
72. 6431
289. Richardson MD, Page ID, Richardson RR. Detection of Aspergillus antibodies by a 6432
new indirect haemagglutinaiton assay. In: 6th Advances Against Aspergillosis. 6433
Madrid; 2014:abstract 97. 6434
290. Knox KS, Hage C. Histoplasmosis. Proc Am Thorac Soc. 2010;7(3):169–72. 6435
doi:10.1513/pats.200907-‐069AL. 6436
291. Smith J a, Kauffman CA. Blastomycosis. Proc Am Thorac Soc. 2010;7(3):173–80. 6437
doi:10.1513/pats.200906-‐040AL. 6438
292. Ramos-‐e-‐Silva M, Lima CMO, Schechtman RC, Trope BM, Carneiro S. Systemic 6439
mycoses in immunodepressed patients (AIDS). Clin Dermatol. 30(6):616–27. 6440
doi:10.1016/j.clindermatol.2012.01.008. 6441
293. Kamali A, Wagner H, Nakiyingi J, Sabiiti I, Kengeya-‐kayondo JF, Mulder DW. 6442
Verbal autopsy as a tool for diagnosing HIV-‐related adult deaths in rural Uganda. 6443
Int J Epidemiol. 1996;25(3):679–684. 6444
294. Pasqualotto AC, Denning DW. An aspergilloma caused by Aspergillus flavus. Med 6445
Mycol. 2008;46(3):275–8. doi:10.1080/13693780701624639. 6446
295. Mabey D, Peeling RW, Ustianowski A, Perkins MD. Diagnostics for the developing 6447
world. Nat Rev Microbiol. 2004;2(3):231–40. doi:10.1038/nrmicro841. 6448
296. Banda H, Kang’ombe C, Harries AD, et al. Mortality rates and recurrent rates of 6449
tuberculosis in patients with smear-‐negative pulmonary tuberculosis and 6450
tuberculous pleural effusion who have completed treatment. Int J Tuberc Lung 6451
Dis. 2000;4(10):968–74. 6452
297. Saraceno JL, Phelps DT, Ferro TJ, Futerfas R, Schwartz DB. Chronic necrotizing 6453
pulmonary aspergillosis: approach to management. Chest. 1997;112(2):541–8. 6454
298. Wheat LJ, Connolly-‐Stringfield PA, Baker RL, et al. Disseminated histoplasmosis in 6455
the acquired immune deficiency syndrome: clinical findings, diagnosis and 6456
treatment, and review of the literature. Medicine (Baltimore). 1990;69(6):361–74. 6457
244
299. Singh UP, Aneja P, Aditi, Patel K. Co-‐existence of HIV, active tuberculosis and 6458
aspergilloma in a single individual-‐-‐a case report. Indian J Tuberc. 2013;60(1):55–6459
8. 6460
300. Kumar AA, Shantha GPS, Jeyachandran V, et al. Multidrug resistant tuberculosis 6461
co-‐existing with aspergilloma and invasive aspergillosis in a 50 year old diabetic 6462
woman: a case report. Cases J. 2008;1(1):303. doi:10.1186/1757-‐1626-‐1-‐303. 6463
301. Agarwal R, Singh N, Aggarwal AN. An unusual association between 6464
Mycobacterium tuberculosis and Aspergillus fumigatus. Monaldi Arch chest Dis. 6465
2008;69(1):32–4. 6466 6467
6468
6469
6470
245
APPENDICES 6471
Appendix 1 -‐ Aspergillus IgG ELISA comparison sheet 6472 DYNAMIKER GENESIS SERION
DILUTION Add 1 µL sample to 1ml
diluent Add 5 µL sample to 1ml
diluent 2 STEP
1st add 10 µL sample to 1ml diluent THEN
add 50 µL from above to 200 µL diluent
STANDARD SERA FIVE FIVE PLUS POSITIVE
CONTROL
TWO PLUS NEGATIVE CONTROL
VOLUME OF SERA / STANDARDS TO ADD 100 µL 100 µL
100 µL
INCUBATION LOCATION (for all steps) 37oC incubator with seal on plate
Bench at room temperature
37oC incubator in moist chamber
1ST INCUBATION DURATION 60 mins 30 mins
60 mins
FIRST WASH STEP 3 washes 3 washes
4 washes
VOLUME OF CONJUGATE 100 µL 100 µL
100 µL
2ND INCUBATION DURATION (with conjugate) 30 mins 30 mins
30 mins
SECOND WASH STEP 3 washes 4 washes
4 washes
VOLUME OF SUBSTRATE 100 µL 100 µL
100 µL
3rd INCUBATION DURATION 15 mins 10 mins
30 mins
VOLUME OF STOPPING SOLUTION 50 µL 100 µL
100 µL
QUALITY CONTROL CRITERIA SB od <0.1 Sa od 0.1-‐0.5 Se od 1.6-‐2.0
PC 32-‐60 STANDARD od 0.42 – 1.43
INDICATION FOR DILUTION AND RETESTING Result > Se Result > 100 U/ml Software reports as HIGH or
>1000 units HIGH IAV RANGE
132 UNITS OR 11% 41 UNITS or 12% 62 UNITS or 23%
246
Appendix 2 – Patient Information Sheet for Paper 2 -‐ Prevalence of chronic pulmonary 6473 aspergillosis secondary to tuberculosis: a cross-‐sectional survey in an area of high 6474 tuberculosis prevalence. 6475
6476 ASPERGILLOSIS STUDY 6477
6478 PATIENT INFORMATION SHEET 6479
6480 We would like to invite you to join our study. You have been selected because you are being 6481 treated for TB, or because you have been treated for TB in the past. Research in England has 6482 shown that patients who have been treated for TB sometimes develop a second illness called 6483 Chronic Pulmonary Aspergillosis or CPA. 6484 6485 This illness is caused by a fungus growing in the lungs. This fungus is very similar to mould on 6486 bread. It can be breathed into the lungs as an invisible dust. If you are healthy this is normally 6487 harmless, but if your lungs have been damaged by TB or other illnesses then it can grow in your 6488 lungs and make you ill. This illness can make you very tired or very short of breath. It can kill you 6489 by causing bleeding inside your lungs. There is treatment available for this illness. Some people 6490 can have an operation to cure them. Others can be made less ill by taking a drug called 6491 Itraconazole. This drug can be bought in pharmacies in Gulu. 6492 6493 In England this illness was found in around 1 person in 20 after they had TB. No one has ever 6494 tested to see how many African patients get the illness after TB treatment. The number might be 6495 bigger or smaller than it is in England. Also, no one has ever tested for this illness in patients who 6496 have had both TB and HIV. It might be that people with HIV are more likely to get this illness 6497 because their immune systems are weaker. 6498 6499 We hope this study will tell us how many patients get this illness in Gulu. We will tell other doctors 6500 the results of our study. If we find that this disease is a big problem we hope that this study will 6501 convince doctors and government ministers to find all the people with the illness and give them 6502 treatment. This could help many people all over the world! We also hope to develop a better blood 6503 test for this illness, which will be cheap and can be used at any African clinic – even ones that don’t 6504 have electricity. 6505 6506 It is your choice whether you want to join this study or not. If you choose to enter the study we 6507 will listen to your chest for signs of the illness and ask you some questions about your health and 6508 where you live and work. This will help us work out if some people are more likely to get this 6509 illness than others. 6510 6511 We will take some blood. This will be used to test for the illness. We will take some blood back to 6512 Manchester University in England. We will use it to help make the new test for Africa. We will also 6513 test it for other types of fungal disease. This will help us decide what diseases to look for in our 6514 next study. Lastly we will use the blood to find genetic markers of the illness. This will not give us 6515 an immediately useful result, but in many years we hope it will let us design a much better test for 6516 people who are at risk of this sickness. If you are coughing we will take a sample of your cough 6517 sputum and test it for fungus infection. After this you will go to Lacor hospital by car where you 6518 will have a chest X ray to see if there is any sign of disease in your lungs. You will then come back 6519 to Gulu by car. We expect the whole process to take a few hours. You will be provided with some 6520 cold drink for refreshment. 6521 6522 We will get results from the tests in a few months. We will pass this result to your doctor if you are 6523 coming back to clinic. If you do not come to clinic we will give the result to the District Health 6524
247
Officer. He will phone your village health worker if your test is positive and ask you should come 6525 to the Infectious Diseases Clinic at Gulu Hospital. They will advise you if you need treatment. If you 6526 do you can buy it from the pharmacy in town. You are welcome to ask any questions you like 6527 before you decide to join the study. If you change your mind and decide not to be in the study later 6528 on we can remove your details from the study and this will not have any impact on your health 6529 care now or in the future. 6530 6531 If you wish to complain about the conduct of the study please contact me in the first instance and if 6532 you are still unhappy you can contact my supervisor Professor Denning 6533 ([email protected]) or the Research Governance Office at the University of 6534 Manchester (research-‐[email protected]). 6535
6536 Finally we would like to contact you again to take part in further studies. This is optional and you 6537 can still take part in this study if you do not want to take part in the other studies. 6538 6539 This project was reviewed by the University of Manchester Research Ethics Committee 1. 6540 6541 6542 6543 6544 6545 6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561
6562
248
6563 Appendix 3 – Patient Consent form for Paper 2-‐ Prevalence of chronic pulmonary 6564 aspergillosis secondary to tuberculosis: a cross-‐sectional survey in an area of high 6565 tuberculosis prevalence. 6566
6567 ASPERGILLOSIS STUDY 6568 6569 CONSENT FORM 6570 6571 6572
If you agree to enter the Aspergillosis study please sign the consent form below. In doing so 6573 you agree to the following:-‐ 6574 6575 I have read the patient information sheet, have had the opportunity to consider it’s contents 6576 and ask questions and had these answered satisfactorily. 6577 6578 I will undergo a medical examination. The doctor will record the results of this and my 6579 answers to his questions. 6580 6581 I will give blood to be used in the study. I understand these samples will be taken to 6582 Manchester University in England and that they will be used for research including 6583 developing a new test and identifying genetic risk factors for Aspergillosis. 6584 6585 I agree to my blood being tested for Aspergillosis and other fungal lung diseases. If I am HIV 6586 positive I understand my CD4 count may be re-‐checked as part of this study. 6587 6588 I agree that the results of the Aspergillosis study will be sent to my doctor at Gulu Hospital 6589 or to the District Health and that they may contact me with the result. 6590 6591 I agree to travel to Lacor Hospital and have a chest X-‐ray when I arrive there. 6592 6593 I understand I may be contacted by study staff in future if I am suitable to enter further 6594 studies. 6595 6596 I understand that I may at any time withdraw my approval for tissue and information to be 6597 stored without giving any reason and without it affecting my treatment. If I do this my 6598 tissue samples will be destroyed and my information will not be used for future research. 6599 6600 ……………………. …………….. ………………........................X 6601 Name of participant Date Signature 6602 6603 I have explained the request for research purposes and answered such questions as the patient has asked. I am 6604 satisfied that the donor signing this form understands the content and the purpose and nature of this consent 6605 process 6606 6607 6608 …………………….. ……………… …………………………………. 6609 Name of person Date Signature 6610 taking consent 6611 6612 Study number -‐ 6613
6614
249
Appendix 4 – Medical Research Council Dyspnoea Scale (MRC Scale) 6615 6616 Grade Degree of breathlessness related to activities 1 Not troubled by breathlessness except on strenuous exercise 2 Short of breath when hurry on the level or walking up a slight hill 3 Walks slower than most people on the level, stops after a mile or so, or
stops after 15 minutes walking at own pace 4 Stops for breath after walking about 100 yards or after a few minutes on
level ground 5 Too breathless to leave the house, or breathless when undressing 6617
6618
250
Appendix 5 – Patient Information Sheet for patients eligible for CT scan 6619
6620 ASPERGILLOSIS STUDY 6621 RE-SURVEY 2014 6622
6623 6624
PATIENT INFORMATION SHEET 6625 FOR PATIENTS WITH POSITIVE SEROLOGY 6626
6627 6628 Thank you for joining our study in 2012/2013. Your blood test results from the first survey showed that 6629 you might be suffering from pulmonary aspergillosis or CPA. Blood tests alone are not enough to 6630 diagnose this disease as people can have positive tests even when they don’t have the disease. We 6631 performed chest X-ray as well as blood tests to provide a picture of the lungs to help us see if 6632 aspergillosis was really present in the lungs. 6633 6634 While chest X-ray is the best test available for this purpose in Gulu it is better to do a test called a CT 6635 scan of the chest. This gives a much better picture of the inside of the chest and so is much better for 6636 identifying aspergillosis in the chest. By having this test we will be able to give you a clear answer as to 6637 whether you are suffering from CPA or not. The test also tells us exactly where in the lungs the CPA is. 6638 This is important as some patients can be cured of CPA by an operation, but it depends on where in the 6639 lung the disease is. A chest X-ray does not give a good enough picture to decide if an operation is 6640 possible. 6641 6642 We therefore plan to transport you to Kampala to undergo a CT scan of the chest at the Kampala 6643 Imaging Centre. We will arrange transport and provide accommodation for you. You will be provided 6644 with an allowance to spend on food while you are away. 6645 6646 A CT scan has a bigger dose of radiation than a chest X-ray. There is a very small risk this might cause 6647 cancer, but this risk is much smaller than the risk that you will become sick because of aspergillosis if it 6648 is not confirmed and treated. We therefore recommend you have this test. 6649 6650 As well as being used to decide if you have aspergillosis or not as part of our study, your scans will also 6651 be available for you to show to a surgeon at Mulago hospital to decide if you can be cured by surgery or 6652 not. We hope to arrange a trial of surgery and if this goes ahead you will be offered the chance to join it. 6653 Any treatment as part of a trial would be free. If there is no trial you would still be able to access 6654 surgical treatment at Mulago Hospital (if the scan shows this is possible) as part of Mulago hospitals 6655 standard provision of care. 6656 6657 If you wish to complain about the conduct of the study please contact me in the first instance and if you 6658 are still unhappy you can contact my supervisor Professor Denning ([email protected]) 6659 or the Research Governance Office at the University of Manchester (research-6660 [email protected]). 6661
6662 Finally we would like to contact you again to take part in further studies. This is optional and you can 6663 still take part in this study if you do not want to take part in the other studies. 6664 6665 This project was reviewed by the University of Manchester Research Ethics Committee 1. 6666 6667 Version 1- 6.8.13 6668
6669
251
Appendix 6 – Consent form for CT scan 6670
6671 ASPERGILLOSIS STUDY 6672 6673 CONSENT FORM FOR CT SCAN 6674 6675 6676
If you agree to undergo CT scan as part of the aspergillosis study please sign the consent 6677 form below. In doing so you agree to the following:-‐ 6678 6679 I have read the patient information sheet, have had the opportunity to consider its contents 6680 and ask questions and had these answered satisfactorily. 6681 6682 I will travel to Kampala where I will undergo a CT scan of my chest. 6683 6684 I understand I have been selected for this test because my blood tests suggest I may have 6685 aspergillosis and that this test will decide if I have aspergillosis or not. 6686 6687 I am aware that there is a very small risk of developing cancer from the CT scan. 6688 6689 I understand that the results of my scan will be assessed to decide whether it is possible to 6690 cure my aspergillosis through an operation or not. This consent is limited to the scan and 6691 does not mean I have decided to undergo surgery. 6692 6693 If surgery is possible I agree to be contacted to discuss the option of having surgery. 6694 6695 I consent to the results of my scan being stored as part of the study. 6696 6697 I agree that images from my scan may be included in publications or presentations relating to this 6698 study. 6699 6700 I understand that I may at any time withdraw my approval for information to be stored or 6701 presented without giving any reason. If I do this my tissue samples will be destroyed and my 6702 information will not be used for future research. I realize that it will not be possible to assess my 6703 suitability for surgery if my images are destroyed. 6704 6705 6706 ……………………. …………….. 6707
………………........................X 6708 Name of participant Date Signature 6709 6710 I have explained the request for research purposes and answered such questions as the patient has asked. I am 6711 satisfied that the donor signing this form understands the content and the purpose and nature of this consent 6712 process 6713 6714 6715 6716 …………………….. ………………6717 …………………………………. 6718 Name of person Date Signature 6719 taking consent 6720
6721
252
Appendix 7 – MIND-‐IHOP study protocol 6722 6723 6724 Mulago Inpatient Noninvasive Diagnosis – International HIV Opportunistic 6725 Pneumonia (MIND-‐IHOP) Study 6726 6727 Principal Investigators: 6728 6729 William Worodria, M.B.Ch.B., M.Med. 6730 Mulago National Referral Hospital 6731 Kampala, Uganda 6732 [email protected] 6733 Telephone: +256-‐(0)772-‐424-‐601 6734 6735 Adithya Cattamanchi, M.D. 6736 University of California, San Francisco 6737 San Francisco, California, U.S.A. 6738 [email protected] 6739 Telephone: +1-‐415-‐206-‐5489 6740 6741 J. Lucian Davis, M.D., M.A.S. 6742 University of California, San Francisco 6743 San Francisco, California, U.S.A. 6744 [email protected] 6745 Telephone: +1-‐415-‐206-‐4694 6746 6747 Charles Everett, M.D. 6748 University of California, San Francisco 6749 San Francisco, California, U.S.A. 6750 [email protected] 6751 Telephone 0001-‐415-‐206-‐3779 6752 6753 Irene Ayakaka, MBChB, MIPH 6754 Makerere University-‐UCSF Research Collab’n 6755 P.O. Box 7475, Kampala, Uganda 6756 [email protected] 6757 Telephone: +256 (0) 772-‐868-‐838 6758 6759 Laurence Huang, M.D., M.A.S. 6760 University of California, San Francisco 6761 San Francisco, California, U.S.A. 6762 [email protected] 6763 Telephone: +1-‐415-‐476-‐4082 ext. 406 6764 6765 Samuel Yoo, M.D., M.Med. 6766 MU-‐UCSF Research Collaboration 6767 Kampala, Uganda 6768 [email protected] 6769 Telephone: +256-‐(0)772-‐461-‐101 6770
253
6771 Nicholas Walter, M.D., M.S. 6772 University of Colorado, Denver 6773 Aurora, Colorado 6774 [email protected] 6775 Telephone 0001-‐415-‐794-‐7527 6776 6777 Christina Yoon, MD, MPH 6778 University of California, San Francisco 6779 San Francisco, California, U.S.A. 6780 [email protected] 6781 Telephone 0001-‐415-‐206-‐831 6782 Co-‐Investigators: 6783 6784 Huyen Cao 6785 California Department of Public Health 6786 Richmond, California 6787 6788 Charles Chiu, M.D. 6789 University of California, San Francisco 6790 San Francisco, California 6791 6792 Saskia den Boon, MSc, PhD 6793 World Health Organisation, WHO, 6794 Geneva, Switzerland 6795 6796 Karen Dobos, PhD 6797 Dept of Microbiology 6798 Colorado State University 6799 6800 Greg Dolganov, Ph.D. 6801 Stanford University 6802 Palo Alto, California 6803 6804 Mark Geraci, M.D. 6805 University of Colorado, Denver 6806 Aurora, Colorado 6807 6808 Moses Joloba, M.B.Ch.B., M.A., Ph.D. 6809 Makerere University 6810 Kampala, Uganda 6811 6812 Harriet Kisembo, MBChB 6813 Makere Univeristy 6814 Kampala Uganda 6815 6816 Joseph Kovacs, M.D. 6817 National Institutes of Health 6818 Bethesda, Maryland 6819
254
6820 Susan Lynch, PhD 6821 University of California, San Francisco 6822 San Francisco, California 6823 6824 Henry Masur, M.D. 6825 National Institutes of Health 6826 Bethesda, Maryland 6827 6828 Steve Meshnick, M.D., Ph.D. 6829 University of North Carolina, Chapel Hill 6830 Chapel Hill, North Carolina 6831 6832 Alison Morris, M.D. 6833 University of Pittsburgh 6834 Pittsburgh, PA 6835 6836 Payam Nahid, MD, MPH 6837 University of California, San Francisco 6838 San Francisco, California 6839 6840 Gary Schoolnik, MD 6841 Stanford University 6842 Palo Alto, CA 6843 6844 Michael Strong, Ph.D. 6845 National Jewish Medical Center 6846 Denver, Colorado 6847 6848 Martin Voskuil, PhD 6849 University of Colorado, Denver 6850 Aurora, Colorado 6851 6852 Alan Wu, PhD. 6853 University of California, San Francisco 6854 San Francisco, California 6855 6856 Jeff Schorey, PhD 6857 Dept of Biological Sciences 6858 University of Notre Dame 6859 6860 STUDY DESIGN 6861 6862 Synopsis: 6863 6864 Respiratory infections are a leading cause of death in Africa, especially among Human 6865 Immunodeficiency Virus (HIV)-‐infected patients, and the lack of understanding of host 6866 and pathogen biology constitutes a major barrier to developing new management 6867 approaches for improving outcomes. Over the past 4 years, rapid, noninvasive tests and 6868
255
strategies have been validated for the diagnosis of tuberculosis (TB), yet substantial 6869 improvements in mortality have yet to be realized. Understanding the fundamental 6870 biological principles underlying human-‐microbial interactions in patients with 6871 respiratory illness offers the possibility for reshaping current approaches to care. 6872 Therefore, we propose minor modifications to our current platform for study of the 6873 diagnosis and epidemiology of HIV-‐associated pulmonary infections and the human 6874 responses to those infections that will combine earlier and more frequent patient 6875 assessment with the latest technologies for studying the biology of host-‐pathogen 6876 interaction. 6877 6878 Objectives: 6879 6880 Our specific aims are: 6881 1. To determine the frequency, quantity, and diversity of bacterial, mycobacterial, 6882
fungal, and viral organisms in respiratory specimens using microbiologic, 6883 serologic, and nucleic-‐acid amplification techniques to determine the 6884 relationship between presence of these organisms and clinical outcomes; 6885
2. To evaluate the performance and impact of novel independent and integrated 6886 approaches to TB diagnosis using both smear microscopy and automated nucleic 6887 acid amplification testing; 6888
3. To evaluate the operational and performance characteristics of novel approaches 6889 to treatment monitoring using intensified measures including clinical 6890 characteristics, microbiologic results, automated nucleic acid amplification, and 6891 cytokine profiling for prediction of clinical and microbiologic outcomes among 6892 patients with TB and other pneumonias; 6893
4. To describe mycobacterial and host gene expression profiles and cytokine 6894 responses in blood and respiratory specimens to gain insights into the 6895 pathophysiology of TB and to more accurately classify TB disease states. 6896
5. To describe the influence of airway pathogens on the gastrointestinal 6897 microbiome by comparing respiratory samples, stool samples, and clinical 6898 outcomes 6899
6900 Some of our associated hypotheses are: 6901 1. The frequency, quantity, and diversity of microbial species in oral and 6902
respiratory specimens will generate new hypotheses about the predictors and 6903 roles of microbial communities and provide insights about clinical outcomes. 6904
2. Novel approaches to microscopy will have equivalent sensitivity to existing 6905 approaches but integrated approaches to TB diagnosis and treatment monitoring 6906 will optimize diagnostic accuracy and maximize clinical impact. 6907
3. Two independent hypotheses: 6908 a. Disease response markers (e.g. quantitative nucleic acid 6909
amplification results, host or pathogen gene expression, quantitative 6910 microbiologic results, host cytokine and inflammatory marker responses) 6911 measured early during the course of anti-‐tuberculosis treatment will provide 6912 insights into the kinetics and biology of treatment response in HIV-‐infected and 6913 HIV-‐uninfected patients; 6914
b. Molecular or microscopy markers measured during treatment of 6915 PCP will have a high positive predictive value for treatment failure. 6916
256
4. Description of mycobacterial and host gene expression profiles and host 6917 cytokine profiles will provide insights into pathogenesis and correctly classify TB 6918 disease states. 6919
6920 6921 Background: 6922 6923 Overview. 6924 6925 Respiratory infections are a leading cause of death in Africa, especially among HIV-‐6926 seropositive adults and children (Lopez, 2006; Ansari, 2003; Lucas, 1993). Definitive 6927 diagnosis of respiratory infections in Africa is difficult because resources are often 6928 limited and because non-‐invasive techniques for diagnosing opportunistic infections 6929 lack adequate sensitivity and specificity. 6930 6931 Key Findings to Date. 6932 6933 Since 1998, members of the study team have been refining molecular methods in 6934 clinical studies of non-‐invasive diagnosis of PCP and tuberculosis (Huang, 2000; Fischer, 6935 2001; Zelazny, 2004). We have applied some of these tools to study respiratory 6936 infections at Mulago Hospital in Kampala, Uganda. Since March, 2007, we have safely 6937 enrolled almost 2000 patients and have facilitated a thorough diagnostic evaluation for 6938 the etiology of pneumonia in each of these patients including chest radiography, CD4 T-‐6939 lymphocyte count measurement in HIV-‐infected patients, sputum smear microscopy 6940 and culture, and bronchoscopy with BAL when requested by the treating physician. 6941 6942 Through this process, we have produced the following key findings: 6943 • HIV seroprevalence is over 80% among patients admitted to Mulago Hospital 6944
with pneumonia 6945 • Tuberculosis is the most common cause of pneumonia, accounting for over 50% 6946
of cases. 6947 • Pneumocystis pneumonia is a rare cause of respiratory infections. 6948 • Clinical symptoms and chest radiography have poor positive and negative 6949
predictive values for TB diagnosis. 6950 • Nucleic acid tests have moderate sensitivity and substantial clinical impact for 6951
the diagnosis of smear-‐negative TB. 6952 • T-‐cell interferon-‐gamma release assays perform poorly for diagnosis of TB. 6953 • Same-‐day microscopy has equivalent sensitivity to conventional two-‐day 6954
microscopy for TB diagnosis. 6955 • LED fluorescence microscopy increases the sensitivity of smear-‐examination for 6956
TB. 6957 6958 Aim 1 – Lung microbial diversity. Increasing evidence for other lung diseases such as 6959 cystic fibrosis suggests that alterations in host bacterial communities contribute to the 6960 pathogenesis of lower respiratory tract infections. No studies have been conducted to 6961 determine the composition of host communities present in the HIV-‐infected lung. We 6962 will evaluate sputum, tongue scrapings, oro-‐pharnygeal washes, and BAL using 6963 bacterial, mycobacterial, fungal, and viral nucleic acid tests (including microarrays), 6964 complemented by conventional microbiologic, serologic, and other biochemical assays 6965
257
to detect with the presence of these organisms, and correlate results with clinical 6966 outcomes. We will describe the types and variation of microbial populations resident in 6967 a variety of respiratory specimens. 6968 6969 Aim 2 – Diagnosis of active TB. TB is the leading cause of mortality in HIV-‐infected 6970 patients in sub-‐Saharan Africa. Failure to promptly diagnose TB has adverse 6971 consequences including disease progression, acceleration of HIV-‐related 6972 immunodeficiency in dually infected persons, and increased TB transmission in the 6973 community (WHO, 2004; Steen, 1998). Despite these consequences, failure to rapidly 6974 diagnose TB is common, in part due to inadequate diagnostic tests. We will evaluate the 6975 clinical impact of integrated algorithms employing combinations of traditional 6976 microbiologic (sputum smear microscopy and culture) and novel nucleic acid testing 6977 (Xpert MTB/Rif) for TB diagnosis on patient-‐ and health-‐system important outcomes. In 6978 addition, we will collect biological specimens (sputum, blood, urine) for evaluation of 6979 novel diagnostic biomarkers for development into new TB diagnostic assays. 6980 6981 Aim 3 – Surrogate markers of response to anti-‐TB and pneumonia chemotherapy. 6982 Surrogate markers of treatment response are needed to decrease the cost and duration 6983 of clinical trials of new anti-‐tuberculosis medications. Documenting clinical cure and 6984 absence of relapse currently requires following patients for up to two years after 6985 treatment completion. Some studies have used two-‐month culture conversion as a 6986 surrogate endpoint, but recent data suggests this approach has limited sensitivity and 6987 specificity. Some studies have explored the role that pathogen and host specific markers 6988 may play in predicting treatment outcomes. However, new measurement tools (e.g. 6989 whole genome gene expression studies, quantitative nucleic acid amplification testing, 6990 multiplex cytokine assays) are now available to improve the precision of our measures. 6991 We will also explore novel approaches to monitoring with smear microscopy, the 6992 standard method for monitoring treatment response in TB patients, by measuring serial 6993 levels of inflammatory markers. For other pneumonias, we will explore other surrogate 6994 markers such soluble TREM-‐1 (bacterial pneumonia), and serum S-‐adenosylmethionine 6995 and co-‐trimoxazole drug levels (PCP) (Gibot, 2004; Skelly 2003). 6996 6997 Aim 4 – TB pathophysiology. There is an urgent need to distinguish between people 6998 who are not infected with MTB, infected but without active disease, and infected with 6999 active disease. Novel techniques can provide insights into host and pathogen 7000 characteristics in different disease states, potentially leading to novel diagnostic 7001 interventions. We will assess (1) Mycobacterial gene expression in respiratory 7002 specimens and (2) Host gene expression and cytokine profiles in respiratory and blood 7003 specimens. We will correlate these results among patients with different MTB disease 7004 states. 7005 7006 Aim 5 – Gastrointestinal Microbiome. We would like to compare the lung microbiome to 7007 the gastrointestinal microbiome in order to better understand the influence of 7008 gastrointestinal microflora on opportunistic pulmonary conditions. We will do this by 7009 comprehensively comparing the bacterial populations in respiratory specimens (tongue 7010 scraping, oral wash, bronchoalveolar lavage specimens) to those in gastrointestinal 7011 specimens (stool specimens). 7012 7013 7014
258
Design: 7015 7016 This is a prospective cohort study of patients with pneumonia admitted to Mulago 7017 Hospital. We will enroll consenting patients with cough, and collect respiratory 7018 specimens (including bronchoalveolar lavage fluid) and blood as indicated to obtain as 7019 definitively as possible a diagnosis for the pulmonary complaints. A subset of patients 7020 will be followed for 2 months as part of a treatment monitoring sub-‐study. An overview 7021 of the protocol follows. 7022 7023 After identifying eligible patients with the assistance of the medical and nursing staff, 7024 the study medical officers will screen and enroll patients. This will take place on the 7025 casualty ward (emergency department, 3BE) on weekdays. All alert, English-‐ or 7026 Luganda-‐speaking adults with respiratory complaints (cough) will be invited to join the 7027 study on the day of hospital admission, through a verbal and/or written invitation in 7028 English or Luganda. Interested patients will subsequently be enrolled at the bedside. 7029 The study team (which includes physicians, medical officers, nurses, and laboratory 7030 technicians) will administer a brief questionnaire to enrolled patients, in English or 7031 Luganda, and collect several biological samples. 7032 7033 These will include expectorated or induced (if the patient is unable to expectorate 7034 spontaneously) sputum. (Please see description below in Part 3: Procedures). Sputum 7035 will be processed, and will undergo staining, interpretation, mycobacterial culture, and 7036 other clinical tests as necessary for care of the patient on site at Mulago Hospital and at 7037 the National Tuberculosis and Reference Laboratory (NTRL). When clinically indicated, 7038 sputum will also undergo testing for M. tuberculosis and rifampicin drug resistance 7039 using the GeneXpertTM MTB automated nucleic acid amplification assay as well as 7040 smear microscopy at Mulago Hospital. The results of sputum acid-‐fast bacilli (AFB) 7041 smears and GeneXpert testing will be available to the treating clinicians within 24 7042 hours. 7043 7044 During enrollment, samples of blood (totaling up to 42 mL) will be drawn from all 7045 subjects and an additional 22mls will be drawn for patients undergoing bronchoscopy. 7046 This blood will be used for CD4 count in the majority of patients who are HIV-‐infected, 7047 and research assays in all patients. Clinical testing will be performed at the Makerere 7048 University–Johns Hopkins University (MU-‐JHU) Core Lab, the Makerere University 7049 College of Health Sciences Clinical Lab, or the Mulago Hospital Clinical Lab, unless 7050 services become unavailable, in which case alternative local labs will be used. Unused 7051 blood will be separated into its constituents (erythrocytes, mononuclear cells, plasma, 7052 serum, etc) and stored for research studies in patients who specifically provide 7053 informed consent. 7054 7055 As soon as testing is complete, sputum AFB results will be collected from the laboratory 7056 and delivered to the ward. Additional sputum will be collected on Day 2. A portion of 7057 this sample will be delivered to the NTLP for smear microscopy and culture. These AFB 7058 results will again be collected and delivered to the ward on the following working day. 7059 Patients with evidence of rifampicin drug resistance on the GeneXpert assay will have 7060 drug susceptibility testing performed, with the results provided to patients. Patients 7061 with drug-‐resistant TB will be registered for treatment with second line TB drugs at the 7062 National TB and Leprosy Programme once a drug-‐resistant-‐TB treatment program has 7063
259
been introduced. The Uganda NTLP has received approval from the WHO Green Light 7064 Committee to acquire second line drugs for treatment of drug-‐resistant TB. The 7065 remainder of all sputum, as well as BAL specimens, and all culture isolates will be 7066 stored for future studies. 7067 7068 HIV results will be received from the hospital-‐run HIV-‐testing service as soon as they 7069 are available after admission. Ward physicians will be encouraged to refer any patient 7070 who is HIV-‐infected, with persistent respiratory symptoms and negative-‐AFB smears, to 7071 the pulmonary service for bronchoscopy as soon as two sputum samples have been 7072 examined for acid-‐fast bacilli. Bronchoscopy is routinely performed in such patients in 7073 high-‐income countries and increases the yield for diagnosis of PCP, pulmonary Kaposi’s 7074 sarcoma, fungal pneumonias, and possibly TB. This study will pay all costs associated 7075 with bronchoscopy and testing of BAL fluid such that all consenting patients can 7076 undergo bronchoscopy unless the clinicians deem it unsafe. 7077 7078 On the morning of bronchoscopy, the bronchoscopist and a bronchoscopy nurse will 7079 consent the patient for the procedure. All patients will be monitored with continuous 7080 pulse oximetry and receive continuous oxygen supplementation, if required. An 7081 oropharyngeal wash (OPW) specimen will be collected at this time by having the patient 7082 gargle 10 mL of sterile normal saline for 60 seconds and expectorate it into a cup. In 7083 addition, 22mL of blood will be collected as well as gentle tongue scrapings , an oral 7084 rinse specimen, and sputum. Baseline vital signs will be recorded. In preparation for the 7085 procedure, the patient may receive intramuscular midazolam for anxiolysis, at the 7086 clinician’s judgment. The nurse will then anesthetize the upper airway with 10 mL of 7087 2% lignocaine, to be administered by nebulizer. Additional aliquots of lignocaine, not to 7088 exceed a total dose of 5 mg/kg of body weight, may be delivered topically to diminish 7089 coughing. Multiple 25 mL aliquots of sterile normal saline will be lavaged through the 7090 bronchoscope channel into a bronchus occluded by the bronchoscope and suction 7091 applied to return a target of at least 50 mL of lavage fluid. After the procedure, the 7092 patient will be monitored by nursing staff to see that vital signs and clinical status have 7093 stabilized before returning the patient to the ward. Bronchoalveolar lavage specimens 7094 will be delivered to the study’s microbiology technician, who will stain the specimens 7095 for PCP, and send them for mycobacterial and fungal stains and cultures. A portion of 7096 the BAL will be saved for research studies. Patients with Pneumocystis pneumonia will 7097 have an additional 5 mL of blood taken from them for sulfa steady-‐state drug levels 7098 after the 5th dose of treatment with any sulfa antibiotic. 7099 7100 Patients undergoing bronchoscopy for clinical indications will be asked to participate in 7101 a sub-‐study in which we will ask them to provide a stool sample near to or on the day of 7102 bronchoscopy. This sample will be used to compare the microbiome of the 7103 gastrointestinal tract to the microbiome of the lung. 7104 7105 A subset of approximately one hundred smear-‐positive TB patients will be asked to 7106 submit serial sputum specimens and provide additional blood during the initial days of 7107 treatment to evaluate treatment response. These patients will undergo serial sputum 7108 sampling prior to and following initiation of standard 4-‐drug TB therapy. Smear, 7109 culture, and automated nucleic acid amplification testing on sputum will be done at 7110 baseline (pre-‐treatment) and after 2 months (60 days) of therapy; automated nucleic 7111 acid amplification testing alone will be done on sputum around days 2, 4, 7, 14, and 30 7112
260
of therapy; and up to 30 mL of blood will be collected to assess gene expression and 7113 cytokine responses at baseline and around days 7, 14, 30, and 60 of therapy. A finger 7114 prick will also be performed for point-‐of-‐care C-‐reactive protein (POC CRP) testing at 7115 the time of enrollment and at each follow-‐up visit. The data for all 100 patients will be 7116 analyzed to identify the 1-‐2 measurement time points during treatment (minimum 7117 sampling frame) that most accurately represent the slope of decline in quantitative 7118 sputum MTB DNA and CRP concentrations described by the full set of time points 7119 (maximum sampling frame). QPCR and cytokine assays will be performed at additional 7120 time points following treatment initiation based on the minimum sampling frame. We 7121 will also collect a small amount of clinical data from those enrolled on treatments taken 7122 after discharge. 7123 7124 Sputum samples provided by patients at 5 month follow-‐up for AFB treatment may be 7125 used for GeneXpert testing including staining/culture for acid-‐fast bacilli and other 7126 pathogens as clinically indicated. 7127 7128 All respiratory specimens will subsequently be processed, de-‐identified, divided into 7129 triplicate sets, and stored frozen in the MIND study freezers located in Mulago Hospital. 7130 At least one set of specimens will remain at Mulago Hospital/Makerere University and 7131 offered to local investigators for research studies. One or more sets (depending on the 7132 yield of each specimen after processing) of non-‐personally-‐identifiable specimens of 7133 sputum, blood, oral specimens, BAL fluid, and culture isolates will be shipped to the U.S. 7134 for testing by laboratory collaborators with different areas of expertise. The 7135 investigators will analyze them according to previously validated protocols and return 7136 the results to the clinical investigators. 7137 7138 Statistical Methods and Sample Size Calculation: 7139 Our sample-‐size generating hypothesis relates to Aim 2, in which we expect to show the 7140 equivalent sensitivity of portable fluorescence microscopy to conventional LED 7141 fluorescence microscopy (FM). To calculate sample size, the following equation is 7142 necessary: 7143 7144 Equation for proportions: N = C * [(P1) (1-‐ P1) + (P2) (1-‐ P2)] * [1/d2] + [2/d] + 2 7145 7146 Sample size N for a study whose endpoint is a difference in proportions can be 7147 estimated using this equation where P1 is the expected proportion in group 1 and P2 is 7148 the expected proportion in group 2, and d is the difference between P1 and P2, 7149 expressed as a positive quantity, and where C is a constant that depends on the values 7150 chosen for alpha and beta. (Fleiss, 1981) 7151 7152 We used PASS 11.0 (NCSS, Kaysville, USA), a comprehensive and validated software 7153 program for Power and Sample Size calculations, to determine the number of patients 7154 needed to demonstrate that portable LED FM is no less sensitive than conventional LED 7155 FM, as defined by a non-‐inferiority margin of 10%. To demonstrate non-‐inferiority with 7156 80% power and a 5% significance level using a one-‐sided equivalence test of correlated 7157 proportions (Liu JP, Stat Med 2002), 370 tuberculosis patients will be needed. This 7158 calculation is based on the 60% sensitivity of conventional LED FM using culture as the 7159 gold standard in our cohort to date and assumes the actual difference in sensitivity 7160 between conventional and portable LED FM will be 0%. Given the 50% prevalence of 7161
261
culture-‐confirmed TB to date, we will need to enroll at least 740 patients. To account for 7162 patients with incomplete work-‐up and contaminated culture results, we plan to enroll 7163 800 patients for this aim of the study. 7164 7165 Data Analysis: 7166 7167 Clinical data will be reviewed, interpreted, cleaned, and analyzed by the clinical 7168 investigators. Chest radiographs will be interpreted according to standardized 7169 categories by a clinical investigator who will be blinded to the diagnoses of the patients. 7170 All data will be entered using study numbers for identification. Data will be entered in 7171 duplicate and compared using SAS Corporation Statistical Software. Data will be 7172 analyzed using Microsoft Access, Microsoft Excel, and STATA Corporation Statistical 7173 Software. The data will be managed by members of the MIND team working in the MU-‐7174 UCSF Research Collaboration Data Centre. The data will be accessible through a secured, 7175 password protected web server stored in the MU-‐UCSF Data Centre, or on secured 7176 servers at UCSF. Bivariate and multivariate analyses of associations between clinical 7177 data and the outcome of disease will be performed. Receiver operating curves will be 7178 generated using measures of test accuracy at various thresholds of results. Other 7179 statistical comparisons between the data points may be performed to test other 7180 hypotheses that arise. 7181 7182 STUDY PROCEDURES 7183 7184 Study Instruments, Procedures, and Location: 7185 7186 • Questionnaire: The medical officer will interview the patient at the bedside to 7187
gather demographic information and obtain a clinical history (Please see 7188 Appendix). 7189
7190 • Sputum: Subjects will submit sputum on Day 0 to the laboratory technician. 7191
Sputum will be delivered to the microbiology lab for smear examination and 7192 culture. If negative, a portion of the sputum samples will be used for GeneXpert 7193 testing including staining/culture for acid-‐fast bacilli and other pathogens as 7194 clinically indicated, while the remainder of the samples will be used for research. 7195 Also, additional sputum will be sent for culture on Day 2 if the GeneXpert test is 7196 negative (Please see Appendix, Flow Diagram). 7197
7198 In addition, for the monitoring sub-‐study, sputum will be collected daily during 7199 inpatient hospitalization and up-‐to-‐weekly during the follow-‐up period. If patients are 7200 unable to expectorate sputum spontaneously, sputum induction may be performed. 7201 7202 • At 5 month follow-‐up, sputum will be collected and used for GeneXpert testing. 7203 7204 • Oropharyngeal Wash/Oral Rinse: The subject will pour 10 milliliters of normal 7205
saline into the mouth, and then “gargle” for sixty seconds. The timing and quality 7206 of the procedure will be recorded. These will be collected at the time of 7207 bronchoscopy. Patients will also be asked to rinse the mouth without gargling 7208 and expectorate. 7209
7210
262
• Tongue Scraping. A sample of oral microbiologic flora will be obtained by a 7211 trained lab technician applying a wooden stick with a smooth edge to the tongue 7212 immediate prior oropharyngeal wash collection. 7213
7214 • Urine. A urine sample will be collected in a 30 mL specimen cup on the day of 7215
admission. 7216 7217 • Finger prick: The laboratory technician will obtain 1-‐2 drops of whole blood via 7218
finger prick. The skin will be prepped with an alcohol prep pad and dried with 7219 cotton. A lancet will be applied to a fingerpad and blood expressed. A capillary 7220 will draw up 1-‐2 drops of blood which will then be mixed with a reagent for POC 7221 CRP measurement. POC CRP will be performed at baseline (Day 0) and at each 7222 follow-‐up visit (Days 2, 4, 7, 14, 28, and 56). This will allow evaluation of the 7223 accuracy and acceptability of this assay. 7224
7225 • Blood: The laboratory technician will collect one ~1 mL sample EDTA-‐containing 7226
tube to measure CD4+ T-‐cell count, one ~10 mL tube for serum/plasma studies, 7227 and three 8 mL tubes for measurement of T-‐cell telomerase enzymes. For 7228 patients undergoing bronchoscopy, an additional 22 mL of blood will be 7229 collected for gene expression and telomerase assays. For patients with TB 7230 enrolled in the treatment monitoring sub-‐study, up to 30 mL of additional blood 7231 will be collected for gene-‐expression and cytokine profiling studies in a subset of 7232 patients at baseline and around days 7, 14, 30, & 60. For patients without TB in 7233 the gene expression profiling study, 5 mL will be collected around day 60. 5 mL 7234 more will be gathered in patients with PCP after the 5th dose of trimethoprim 7235 sulfamethoxazole. Clinical blood tests will be performed in the MU-‐JHU lab, 7236 which is certified by the Clinical Laboratory Improvements Amendments (CLIA) 7237 Advisory Committee. 7238
7239 • Clinical data: The medical officer will measure the patient’s vital signs, lung 7240
physical exam results, and clinician diagnosis from the bedside chart at the time 7241 of admission. 7242
7243 • Chest radiographs: Chest radiographs will be taken routinely at enrollment, and 7244
for the IM patients, at the 8-‐month follow-‐up visit. If a chest x-‐ray has not been 7245 performed just prior to admission, one will be obtained. The medical officer will 7246 photograph x-‐rays with patient-‐identifying text obscured. If additional x-‐rays are 7247 requested for clinical purposes while the patient is being followed in the study, 7248 the study will provide those as well. Chest radiographs will be interpreted using 7249 a standardized research form. 7250
7251 Bronchoscopy: For HIV-‐infected patients who are not shown to have tuberculosis after 7252 sputum analysis, bronchoscopy will be performed in the bronchoscopy suite upon 7253 request of the treating physician according to the local protocol. Bronchoscopy will be 7254 deferred in patients with unstable respiratory status. 7255 7256 • Stool Collection: We will ask bronchoscopy subjects for a stool sample. We will 7257
ask that patients place a paper collection device onto the toileting area just prior 7258 to having a bowel movement. After depositing the stool on the device, it will be 7259
263
divided between 3 plastic cups using a scoop. Once the stool has been placed in 7260 the plastic cups, they will be sealed tightly. 7261
• Follow-‐up: All patients will have vital status assessed by phone or by a home 7262 visit if not contactable by phone at 2 months from the time of enrollment. A 7263 subset of patients will receive more intensive follow-‐up described below after 7264 enrollment into the treatment monitoring sub-‐study. In addition, for non-‐TB 7265 patients only, 30 mL blood for gene-‐expression profiling will be collected at the 7266 visits that occur around 60 days. 7267
7268 Study procedures will take place on the wards or clinics of Mulago Hospital, Makerere 7269 University, in Kampala, Uganda. Bronchoscopy will be performed in the designated 7270 bronchoscopy area. 7271 7272 The procedures will be administered by the study’s Ugandan coordinators. Estimates of 7273 patient time required to participate in the study are as follows: 7274 7275 Inpatient period (2 hours and 45 minutes): 7276 • Day 1: 30-‐minute visit from medical officer for informed consent and 7277
questionnaire. 30-‐minute visit from laboratory technician for collection of 7278 sputum, blood, urine, and stool. 7279
• Day 2 (or next working day): 15-‐minute visit from laboratory technician for 7280 collection of sputum sample and finger prick. 7281
• Day 3 (or as soon as can be scheduled) (this is only for patients undergoing 7282 bronchoscopy): 7283 o 15-‐minute visit from laboratory technician for collection of oral wash 7284 sample and blood draw. 7285 o 60-‐minute visit with attending pulmonologist for bronchoscopy. 7286 o 15-‐minute visit for collection of sputum, mouthwash, oral rinse, and 7287 tongue scraping, and for administration of a short questionnaire (Please see 7288 Form 4B) 7289
7290 Follow-‐up visits for all patients (10 minutes): 7291 • Telephone visit to assess vital status. If not available by phone, home visit from 7292
individual designated to track patients to home. 7293 7294 Follow-‐up visits for about 100 smear-‐positive TB patients enrolled in an intensive 7295 monitoring sub-‐study (7 inpatient visits of no more than 30 minutes each = 3 hours 30 7296 minutes). Patients will be asked to stay in hospital in an area where WHO-‐standard 7297 infection control measures have been implemented during this initial period, and then 7298 will be followed up during 3 30-‐minute home visits around Days 14, 30, and 60: 7299 • Collection of 5 mL sputum volume, by expectoration or induction 7300 • Collection of 1-‐2 drops of blood via finger prick 7301 • Collection of 30 mL blood via venipuncture 7302 7303 Follow-‐up for up to 50 non-‐TB patients for collection of 5 mL blood for gene expression 7304 profiling. (60 minutes) 7305 7306
264
Total patient time for patients for the standard inpatient procedures and follow-‐up is 7307 estimated as 2 hours, 55 minutes. Total additional patient time for patients enrolled in 7308 the intensive monitoring sub-‐study is estimated as 5 hours. 7309 7310 All study procedures will be reviewed by the Makerere University School of Medicine 7311 Research Ethics Committee, the University of California, San Francisco, Committee on 7312 Human Research, and the Mulago Hospital Institutional Review Board. They will also be 7313 reviewed annually by the Uganda National Council for Sciences and Technology. 7314 7315 Specimen Testing and Data Review: 7316 7317 Specimen analysis and data review will take place in the laboratories/offices of co-‐7318 investigators, listed above, according to well-‐defined protocols that include the testing 7319 of positive and negative controls as indicated. Any positive result will be communicated 7320 to the primary team. DNA and RNA will be processed and analyzed in laboratories of the 7321 investigators in accordance with the above proposed protocols and the banking 7322 permissions granted. All specimens will be coded and de-‐identified, and non-‐clinical 7323 investigators will not have access to the key. 7324 7325 One specimen, blood will be subjected to tests of host gene expression at the laboratory 7326 of Mark Geraci, M.D., and collaborating core lab facilities under the scientific direction of 7327 Dr. Geraci and Dr. Walter at the University of Colorado. These specimens will be used to 7328 study host responses to pulmonary infections. We will not collect DNA or analyze 7329 individual-‐specific genetic characteristics. Instead we will analyze patterns of RNA 7330 expression (gene expression). This analysis therefore does not meet the definition of 7331 genetic research. Once laboratory testing is complete, the results will be linked to the 7332 clinical outcome data by the clinical investigators, who will perform the primary 7333 analysis. Only researchers listed as investigators for this protocol will have access to the 7334 specimens and clinical data. 7335 7336 Tissue Banking Procedures: 7337 7338 De-‐identified specimen material remaining after the completion of study assays will be 7339 stored in secured freezers at three sites: 1) Mulago Hospital; 2) San Francisco General 7340 Hospital, and 3) the National Institutes of Health. At Mulago Hospital, the specimens will 7341 be stored in a locked freezer in the Department of Microbiology on the 2nd floor. At San 7342 Francisco General Hospital, the specimens will be stored in a locked laboratory located 7343 in Building 100, Room 109, San Francisco General Hospital, 1001 Potrero Avenue, San 7344 Francisco, CA 94110, USA. At the National Institutes of Health (NIH), the specimens will 7345 be stored in the Lung HIV Specimen Bank at the National Heart Lung and Blood 7346 Institute. The code and all identifiable clinical information will be stored separately on 7347 password-‐protected computer servers located at the MU-‐UCSF Research Collaboration, 7348 Mulago Hospital, and at the University of California, San Francisco. If outside 7349 investigators request use of these specimens for research suited to the scientific aims 7350 listed in the consent form (i.e. research to learn about, prevent, or treat other 7351 respiratory infections or diseases and diseases related to HIV, as we deem appropriate), 7352 we would release the specimens under our control after our scientific merit review of 7353 the proposed research and after receipt of a copy of the IRB-‐approval letter for the new 7354
265
protocol. Procedures for release of de-‐identified specimens from the NIH Specimen 7355 Bank will be subject to procedures overseen by the NIH Institutional Review Board. 7356 7357 If the participant decides that he or she does not wish for his or her oral wash specimen 7358 or clinical information to be used for future research, he or she may tell us, and we will 7359 destroy any remaining identifiable sample and information, and ask our collaborators to 7360 do so also. 7361 7362 A material transfer agreement governing this study has been approved by legal and 7363 governing authorities at Makerere University, UCSF, the University of Colorado and the 7364 National Institutes of Health. The Foundation for Innovative New Diagnostics, a non-‐7365 governmental organization supporting some of this work and carrying out laboratory 7366 assays on some of these specimens, is also party to a material transfer agreement 7367 governing this study. 7368 7369 7370 RISKS AND BENEFITS 7371 7372 Alternatives to Participation: 7373 7374 Patients who choose not to enroll in the study may receive whatever care they would 7375 have ordinarily received had they not been approached to participate in the study. This 7376 might include bronchoscopy with bronchoalveolar lavage, which we will provide for 7377 HIV-‐infected inpatients who require a procedure to diagnose a pneumonia of 7378 undetermined etiology. Otherwise, only noninvasive tests (AFB-‐sputum-‐smears) are 7379 available for diagnosing pneumonia at Mulago Hospital. 7380 7381 Risks and Discomforts: 7382 7383 Subjects performing oropharyngeal wash may gag, experience nausea, or feel short of 7384 breath as a result of gargling. These symptoms are likely caused by the act of gargling 7385 rather than by the solution gargled. These side effects are rare and usually self-‐limited. 7386 7387 Tongue scraping with a smooth wooden spatula should be painless in subjects with 7388 normal oral mucosa and should have no lasting consequences. If the mucosa is damaged 7389 such that the procedure could induce pain, the test will be omitted. 7390 7391 Patients having blood drawn via finger prick or venipuncture may experience local 7392 discomfort at the sit of the needle puncture, where the skin may become dark or tender. 7393 Patients with anemia may be unable to tolerate having large amounts (>25 mL) of blood 7394 drawn. To guard against this possibility, patients with evidence of conjunctival pallor or 7395 other clinical signs of anemia will be screened with a hemocue hemoglobin 7396 measurement. Anyone with a hemoglobin less than 7 mg/dL will not have blood drawn. 7397 7398 Patients performing sputum induction may gag, experience nausea, or feel short of 7399 breath as a result of the nebulization procedure. These symptoms are likely caused by 7400 the act of saline inhalation rather than by the solution gargled. These side effects are 7401 rare and usually self-‐limited. Severely ill patients may develop oxygen desaturation 7402 during the procedure. To guard against the possibility of oxygen desaturation, patients 7403
266
with respiratory rates greater than 30 or requiring oxygen supplementation will be 7404 excluded from sputum induction for research purposes. Patients may still be referred 7405 for sputum induction for clinical purposes, but only at the request of the primary 7406 clinical team caring for the patient. Such referrals will need approval of the attending 7407 physician on the 4C Pulmonology ward. 7408 7409 Bronchoscopy is performed regularly at Mulago for patients who can afford it. Study 7410 participants undergoing bronchoscopy assume the same risks as any patient 7411 undergoing bronchoscopy at Mulago. These risks include coughing, gagging, aspiration, 7412 minor bleeding, pneumothorax, respiratory failure, and death. The risks of 7413 bronchoscopy will be disclosed in the consent form. Of the risks described above, 7414 coughing and gagging are common but self-‐limited. To prevent them, all patients receive 7415 a pre-‐procedure treatment with nebulized lignocaine, and are treated with additional 7416 topical lignocaine to control coughing or discomfort at the clinician’s discretion during 7417 the procedure. Aspiration is rare, but to prevent its dangerous consequences (acute 7418 pneumonitis or pneumonia) patients take nothing by mouth for 8 hours prior to the 7419 procedure. Any patients not in compliance with this requirement will have 7420 bronchoscopy postponed until they comply. Bleeding and pneumothorax are extremely 7421 uncommon adverse effects of bronchoscopy, especially when biopsies are not planned. 7422 Respiratory failure may occur in patients undergoing bronchoscopy for evaluation of 7423 pneumonia because of worsening of the underlying disease process with lavage of the 7424 lung. To guard against this possibility, all patients referred for bronchoscopy are 7425 routinely screened by a Pulmonary physician before the procedure. If bronchoscopy is 7426 deemed unsafe, the procedure will not be performed and the patient will be returned to 7427 the ward. Death from bronchoscopy is extremely rare. When fatal complications occur, 7428 they are usually the result of bleeding, pneumothorax, or respiratory failure. To screen 7429 for early signs of such adverse events, all patients will be monitored with continuous 7430 pulse oximetry and receive continuous oxygen supplementation, if required. 7431 7432 Finally, testing for TB using sputum induction or bronchoscopy produces aerosols that 7433 may be infectious, and pose a risk to individuals who are subsequently exposed to these 7434 aerosols. To reduce this risk of nosocomial TB transmission, sputum induction will be 7435 performed in a well-‐ventilated room on the Pulmonology ward. After sputum induction, 7436 fan ventilation out of the open window will be performed for at least 15 minutes for 7437 another patient enters the room. Similarly, in the bronchoscopy suite, fan ventilation 7438 through an open window will be used to remove infectious aerosols. N95 respirators 7439 will be supplied to all staff working with patients in these settings. 7440 7441 Stool should be handled with gloved hands at all time. Direct contact between stool and 7442 ungloved broken skin may lead to infection. Some people, particularly those sensitive to 7443 odor, develop nausea and occasionally vomiting. 7444 7445 Benefits: 7446 7447 All patients will have sputums stained for AFB on the first hospital day. In addition, 7448 sputum samples will be tested for TB using rapid nucleic acid testing (PCR), and 7449 cultured for mycobacteria. Patients will be notified of the results of these tests as soon 7450 as they are available. Both patients and providers have described this as a major benefit 7451 of the study. In our previous study, a large proportion of smear-‐negative patients were 7452
267
confirmed to have TB through nucleic acid testing or mycobacterial culture results. 7453 Follow-‐up smears and cultures for clinical purposes will no longer be collected at two 7454 months to monitor response to therapy, because of poor uptake of these services 7455 previously. Instead, patients will be referred to TB dispensaries in their local area. 7456 7457 Accurate diagnosis in HIV-‐infected patients with pneumonia, as noted previously, is 7458 extremely difficult without an integrated approach that includes a range of diagnostic 7459 tests including bronchoscopy. Appropriate treatment depends on accurate diagnosis. 7460 Without appropriate treatment, these patients face certain mortality, many in the first 7461 few days of hospitalization. The benefits of this study are thus enormous, in terms of 7462 providing for free testing that otherwise would not occur. 7463 7464 Confidentiality and Privacy: 7465 7466 Most patients enrolled in the study will be HIV-‐infected. If individuals outside the study 7467 learn that an individual enrolled in our study is HIV-‐infected, that person may 7468 experience stigma, such as trouble obtaining employment or problems being accepted 7469 by family or community. We have proposed many safeguards to prevent disclosure of 7470 personal health information during the course of the study. We will be collecting 7471 personal health information, including the unique identifiers name and date of birth, but 7472 we do not plan to share this information outside the research team and names will not 7473 be recorded in our databases. 7474 7475 Risk/Benefit Analysis: 7476 7477 The individual benefits of early and definitive sputum diagnosis of tuberculosis are 7478 tremendous in this clinical setting. Even more sensitive and specific bronchoscopic 7479 diagnosis of pneumonia for HIV-‐infected patients provides these patients with the 7480 international standard-‐of-‐care test for pneumonia. The research aspects of the protocol 7481 do not enhance the ordinary clinical risks to the patient. In addition, the potential 7482 benefits of better diagnostic modalities for respiratory infections for many of the 33 7483 million patients infected with HIV worldwide are inestimable. The results of the 7484 previous study have been widely disseminated to invested personnel at Mulago 7485 Hospital, Makerere University, and the NTLP, and we will continue this in the future. 7486 Where possible educational intervention and policy recommendations will be 7487 developed and introduced in cooperation with local stakeholders. On balance, benefits 7488 outweigh risks. 7489 7490 7491 SUBJECTS 7492 7493 General Description of Study Subjects: 7494 7495 Our target population consists of all HIV-‐infected adults undergoing evaluation on the 7496 grounds of Mulago Hospital or associated clinics with a clinical suspicion of pneumonia. 7497 We will screen over 10,000 patients, and enroll at least 3300 patients. 7498 7499 Inclusion Criteria: 7500 7501
268
Adults undergoing evaluation on the grounds of Mulago Hospital or associated clinics 7502 with cough may screened and invited to enroll. 7503 7504 Exclusion Criteria: 7505 7506 Patients who are under the age of 18, unable to provide consent, or unable to 7507 communicate in English or Luganda will be excluded. We will also exclude patients with 7508 heart failure. 7509 7510 Inclusion criteria for monitoring sub-‐study: 7511 Sputum AFB or GeneXpert automated nucleic acid test positive for TB and willing to 7512 participate in intensive follow-‐up program 7513 7514 Exclusion criteria for monitoring sub-‐study: 7515 Patients residing >30 km from Mulago Hospital and patients otherwise unable to adhere 7516 to intensive follow-‐up plans will be excluded. 7517 7518 Screening Procedures: 7519 7520 The medical officer/nurse will screen all patients as they undergo clinical evaluation. 7521 Patients meeting inclusion criteria will be invited to join the study, through a process 7522 described in more detail below in the “Recruitment” section. Patients enrolled in the 7523 sub-‐study will be identified after TB diagnosis according to inclusion and exclusion 7524 criteria above. 7525 7526 7527 RECRUITMENT 7528 7529 The medical officer will approach patients identified as they undergo evaluation in the 7530 clinic or after hospital admission in their beds on the open ward. Patients will be asked 7531 if they would like to participate in a study to evaluate the etiology of respiratory 7532 infections, without any reference to HIV-‐status. HIV-‐status will not be mentioned to 7533 protect study subjects from disclosure of HIV status. If an individual expresses interest 7534 in the study, his/her name and bed number will be recorded, with enrollment deferred 7535 until after screening. If the patient needs more time to decide (e.g. because he/she 7536 needs to consult his/her attendant or family member), the coordinator will attempt to 7537 return later. After the screening described above, patients enrolled in the sub-‐study will 7538 be randomly enrolled after TB diagnosis. 7539 7540 INFORMED CONSENT PROCEDURES 7541 7542 At the time of enrollment, a study officer will introduce himself or herself, and explain 7543 the study by reading the standardized consent form to the subject. One of these officers, 7544 who are all bilingual, will read the consent in English or Luganda, according to the 7545 subject’s preference. The subject will be provided with a copy of the consent form to 7546 read, but literacy will not be required for consent (For patients unable to read, a witness 7547 will be required to co-‐sign the consent form). (Please see Appendix). After the 7548 coordinator has read through the document, which is written in a question and answer 7549 format, the subject will be asked if he or she has any questions. Then, the subject will be 7550
269
asked whether he or she wishes to grant, refuse, or defer a decision on participation in 7551 the study. If the subject is unable to decide before the coordinator leaves the ward for 7552 the day, he or she will not be enrolled. If the subject agrees to participate, he or she will 7553 be asked to sign the consent form. A separate consent form will be used for specimen 7554 banking. 7555 7556 Patients will be asked to enroll in the stool sub-‐study if they are going to undergo 7557 bronchoscopy. Patients enrolling in the intensive follow-‐up for treatment monitoring 7558 study will be consented according to a separate consent process discussing fully the 7559 issues related to longitudinal follow-‐up. We will draw on our experience in previously 7560 approved SOM-‐REC MIND protocols which involved longitudinal follow-‐up. 7561 7562