20/03/2012

1

Global value chain in action: Xpert MTB/RIF

Mark NicolDivision of Medical Microbiology and 

Institute for Infectious Diseases and Molecular Medicine, 

University of Cape Town and National Health Laboratory Service

Mark.Nicol@uct.ac.za

Disclosure

The University of Cape Town (PI Mark Nicol) has received funding from FIND to support evaluation and demonstration studies for GeneXpert.

I have no other conflicts of interest to declare. 

Acknowledgement

I gratefully acknowledge Prof. Wendy Stevens, NHLS Priority Programmesand University of the Witwatersrand for information and slides on the national roll‐out of Xpert MTB/RIF

20/03/2012

2

Xpert MTB/RIF for the diagnosis of TB

Lawn, Nicol. Future Microbiol 2001; 6(9): 1067‐82

Lawn, Nicol. Future Microbiol 2001; 6(9): 1067‐82

20/03/2012

3

Sensitivity of Xpert for diagnosis of TB in adults

Lawn, Nicol. Future Microbiol 2001; 6(9): 1067‐82

TB diagnostic services in SA: before Xpert

Microscopy Sites and Smear volumes (2010)4.7 million smears

1 million cultures (22% positive)

20/03/2012

4

WHO recommendation process

• WHO Expert group meeting (September 2010)

– Reviewed 4 published papers and unpublished data (GRADE)

– Confirmed solid evidence base to support widespread use

• STAG‐TB endorsed Expert group recommendations

– Advised implementation within the context of national strategic plans

– Recommendations:

• Xpert should be used as the initial diagnostic test in individuals suspected of having MDR‐TB or HIV‐associated TB. (Strong recommendation)

• Xpert MTB/RIF may be considered as a follow‐on test to microscopy in settings where MDR‐TB or HIV is of lesser concern, especially in further testing of smear‐negative specimens. (Conditional recommendation acknowledging major resource implications)

WHO recommendation process

• Global consultation called by WHO (December 2010)– Discuss implementation considerations

– Interim diagnostic algorithms and case definitions

– Positioning of Xpert• District and sub‐district level

• Prioritize high MDR‐TB and HIV‐TB sites

• Sufficient workload

• Adequate infrastructure

• Dedicated personnel

• Capacity for treatment

– Implementation considerations to optimize use and benefits

20/03/2012

5

The decision to roll out Xpert in SA

• December 2010: WHO recommendation

• Early 2011: Minister of Health requests implementation

• NDoH proposed pilot study (while feasibility being assessed)

– >1 instrument per province in high burden districts (selected by TB cluster)

– 25 sites selected and 30 instruments placed (funded by NDoH, FIND, PEPFAR)

– All instruments placed by world TB day (March 24th) 

Phase I placements

20/03/2012

6

Phase I results

• By October 2011:

– 135,601 specimens tested

– 16% positive

– Average rifampicin resistance 7%

– Error rate <3% (except May – 5% due to batch issue)

– Invalid rate <1%

Phase I sites

20/03/2012

7

Instrument utilization rates

Phased roll out of Xpert in SA

• Phase I: completed • Phase IIa: full capacitation of existing labs (17 additional 

instruments, completed October 2011) • Phase IIb: full capacitation of high burden districts 

(Global Fund, roll‐out in April 2012) • Phase IIIa and b: BMGF funded study (Gauteng, EC and 

Free State) • Phase IIIc: ensuring all districts have a minimum of 1 

instrument per district • Phase IIId: completion of all current microscopy and 

clinic sites 

20/03/2012

8

Phase I and IIa results

• By January 5, 2012:

– 209,099 specimens tested

– 17% positive (35,737 new cases)

– Average rifampicin resistance 7%

– Error rate <3% (except May – 5% due to batch issue)

– Invalid rate <1%

Related activities

• Staff training: >500 staff trained (including laboratory and clinical staff)

– Technical, GCLP, EQA, data entry, safety

– Algorithms 

• Quality assurance programme

20/03/2012

9

Challenges in implementation

• Timelines

• Identification of priorities (political vs. rational)

• Costing and funding– Alignment of donors expectations

• High error rates

• Interface with LIS

• Algorithms– How to deal with RIF resistance

– Request forms

– Clinical training and adherence

• From the sublime….

…to the ridiculous: the Infinity

20/03/2012

10

Current algorithm for MDR detection

Specimen 1Smear microscopy

(bleach treated)

Previous TB or high risk suspect (MDR contact, health care personnel or prisoner)

Specimen 2Smear microscopyCulture and DST

Smear-positive Smear-negative

MGIT CultureDirect line probe assay

Indirect line probe assay

If Rif R – culture and 2nd

line DSTIf Rif R 2nd line DST

Current algorithm for MDR detection

Specimen 1Smear microscopy

(bleach treated)

Previous TB or high risk suspect (MDR contact, health care personnel or prisoner)

Specimen 2Smear microscopyCulture and DST

Smear-positive Smear-negative

MGIT CultureDirect line probe assay

Indirect line probe assay

If Rif R – culture and 2nd

line DSTIf Rif R 2nd line DST

Variable implementation

20/03/2012

11

Current algorithm for MDR detection

Specimen 1Smear microscopy

(bleach treated)

Previous TB or high risk suspect (MDR contact, health care personnel or prisoner)

Specimen 2Smear microscopyCulture and DST

Smear-positive Smear-negative

MGIT CultureDirect line probe assay

Indirect line probe assay

If Rif R – culture and 2nd

line DSTIf Rif R 2nd line DST

LPA replaced MGIT DST for INH and Rif

Performance of Xpert for detection of Rif resistance

• Evaluation study

– Comparison with phenotypic DST:• sensitivity 98% specificity 98%

– After sequencing discordant isolates• sensitivity 99% specificity 100%

• Early demonstration study

– 2 cases identified in Khayelitsha• Rif R on Xpert, Rif S on LPA (confirmed  WT on sequencing)

– Subsequently 2 cases identified in Paarl, other sites

– Sensitivity 99 (96‐100) % specificity 96 (95‐97) %205/208                     (679/706)

20/03/2012

12

False rifampicin resistance calls

Perceptions and PPV

95 Perspective of the TB programme

54

5

91 True susceptible

False resistantTrue resistant

20/03/2012

13

Perceptions and PPV

Perspective of the MDR clinician

False resist7 cases

True resist

10 cases

Prevalence of MDR and PPV

Positive Predictive Value

Specificity96%

Specificity 98%

Prevalenceof RIF resistance

5% 56 71

10% 73 84

20% 86 92

50% 96 98

New G4 cartridge: redesigned probes, fluidics software changes

20/03/2012

14

Adherence to algorithm…RIF resistant cases

GeneXpert  Rif Resistance cases and comparison to Culture DST and/or LPA

LPA/DST confirmation Conc RIF cases either LPA/DST:  LPA Culture

Province Rif res cases number  % number %RIF conc only LPA 

% RIF mono RIF conc to culture %Rif mono

Eastern Cape 142 37 26% 20 54% 84% 10% 21% 5%

Free State 46 23 50% 16 70% 78% 26% LPA off culture

Gauteng 109 52 48% 46 88% 92% 50% 88% 40%

KZN 691 266 38% 222 83% 90% 19% 93% 12%

Limpopo 76 23 30% 23 100% 81% 9% 95% 5%

Mpumalanga 120 70 58% 62 89% 91% 19% 100%* 40%*

North West 43 11 26% 9 82% 90% 10% check algorithm

Northern cape  103 61 59% 49 80% 82% 25% LPA off culture

Western Cape 100 89 89% 84 94% 93% 12% LPA off culture

National 1430 632 47% 531 82% 87% 20% 74% 16%

20/03/2012

15

Instrument costs

Test costs

20/03/2012

16

Cost and cost‐effectiveness

• The scale‐up will lead to the placement of 65 GX4, 169 GX16, and 4 GX48 instruments, – total national test capacity of 11,428 tests per day. 

• Total capital cost (including instruments, additional space, security, and training) between 2011/12 and 2016/17 will be 149 million ZAR;

• Total incremental recurrent cost (including cartridges, staff, transport, and quality assurance) – 2.03 billion ZAR (Gradual scale‐up) and 2.34 billion ZAR (Accelerated scale‐up). 

• Additional annual budget requirement – 217 and 539 million ZAR (Accelerated scale‐up) ‐200 and 462 million ZAR (Gradual 

scale‐up). 

• The NHLS (or laboratory) share of total diagnostic cost increases by – 298 million ZAR in 2013/14, 345 million ZAR in 2014/15, 397 million ZAR in 

2015/16, and 444 million ZAR in 2016/17 (Accelerated scale‐up only). 

• The incremental NHLS cost per suspect is between ZAR 129 and ZAR 141• The incremental cost per patient diagnosed is between ZAR 485 and ZAR 

650. 

A: inB: out

A: sequencedB: linked

A: inB: out

Patient with suspected TB(positive symptom screen)

Sputum microscopy(1st and 2nd) Empirical

antibiotics

Chest X-rayTB culture

Sputum microscopy(3rd)

TB diagnosis

if positive

if negative

if failedHIV-

Empirical antibiotics

+

if positive

if positive

SCENARIO 1(Old algorithm)

SCENARIO 2(New algorithm)

GeneXpert(1st)

Clinical exam

if negative

if failed

HIV+

GeneXpert(2nd)

c1

c1

c2

c2

c7

c3

c4

c6

c6

p2

p3

1-p3

p4

p5

p10

p16

p19

1-p9-p10

RIF susceptibleRIF resistant RIF inconclusive

Sputum microscopy

TB culture

DST +/-LPA

Sputum microscopy

if positiveif unsuccessful

Patient without TB history

DST +/-LPA

TB culture

Sputum microscopy (1st and 2nd)

Empirical antibiotics

if failed

Chest X-raySputum microscopy(3rd)

Patient with TB history

c3

c1

c4

if negative

c1

c5

c2

if positive

Empirical antibiotics

if still suggestive of TB

if positive

if positive p15

c2

p18

p9

1-p2

p6

1-p6

p7 p8

p17

c5

c4

c1 c1

p1

1-p1

if positive

p11*p12 p13 p14

20/03/2012

17

Xpert Arm

SmearMGIT

(regional lab)

TB suspect / MDR suspectWeekly randomization blocks

Control arm

Sputum 1

Xpert(local lab)

2 and 6 months follow upIn all TB treated and TB test positive patients

Speciation and DST for C+

Sputum 2 Sputum 1

Smear(local lab)

Sputum 2

SmearMGIT

(regional lab)

Impact assessment: Cape Town Demo Study

20/03/2012

18

Khayelitsha (1 clinic)

768 TB suspects

2 smear sensitivity: 46%

TB suspect / MDR suspectTwo sites

Paarl (6 clinics)

1399 TB suspects

Xpert 373

Xpert sensitivity: 88%

Control 395 Xpert 614 Control 785

2 smear sensitivity: 48%Xpert sensitivity: 88%

Cape Town Demo Study: Summary

MGIT pos 37%

Ref smear pos: 15%

MGIT pos 35%

Ref smear pos: 14%

MGIT pos 18%

Ref smear pos: 5%

MGIT pos 21%

Ref smear pos: 7%

Additional yield 42% Additional yield 40%

• Sensitivity of Xpert:

– HIV‐negative patients• All HIV neg: 95%

• Smear pos: 100% 

• Smear neg: 92%

– HIV‐infected patients• All HIV pos: 85%

• Smear pos: 100% 

• Smear neg: 79%

Impact of HIV on performance of smear microscopy and Xpert

S‐ X‐

S‐ X‐

S+ X+

S‐ X+S+ X+

S‐ X+

20/03/2012

19

Both laboratory sites:– Fluorescent AFB microscopy– NHLS QA programme

KY clinic laboratory – started operation 1 month prior to study– Single experienced technician– 40‐50 smears per day

vs. 

Paarl laboratory– Established district laboratory– 2 staff members performing/reading AFB smears– 200 smears per day (multiple clinics)

What is in a smear?

Variable performance of smear microscopy

Both smears +

Local smear +

Reference smear +

38

53

9

Both smears +

Local smear +

Reference smear +

30

60

10

Khayelitsha Paarl

Xpert sensitivity: 88.3% Xpert sensitivity: 87.5%

20/03/2012

20

Time to treatment

6

38

4

4

9

7

938

6

38

44

938

9

7

Time to treatment

P<0.001P<0.001

82737165

20/03/2012

21

Further impact assessment

• Phase II of Demo study

• XTEND study

– to evaluate the effectiveness and cost effectiveness of Xpert MTB/RIF in the investigation of TB and TB drug resistance, and its impact on patient and programmeoutcomes and transmission at a population level

– cluster randomised pragmatic trial (CRT) with the unit of randomisation as the laboratory

Acknowledgements

• FIND– Catharina Boehme– Pamela Nabeta– Christen Gray

• UCT– Neisha Mohess– Silindile Mbhele– Widaad Zemanay

• NHLS– Wendy Stevens– Andrew Whitelaw– Gerrit Coetzee– John Simpson– Marlein Bosman

• MSF– Helen Cox– Cheryl McDermid

• City of Cape Town– Virginia De Azevedo

• Provincial Government of the Western Cape

• Patients 

• Funders:– The Wellcome Trust– EDCTP