DOI: 10.1542/peds.2009-2860 published online Sep 6, 2010; Pediatrics
Brierley, Nigel Cross, Sophie Skellett and Mark J. Peters Helen E. Rowlands, Allan P. Goldman, Karen Harrington, Ann Karimova, Joe
Young InfantsImpact of Rapid Leukodepletion on the Outcome of Severe Clinical Pertussis in
http://www.pediatrics.orglocated on the World Wide Web at:
The online version of this article, along with updated information and services, is
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275. Grove Village, Illinois, 60007. Copyright © 2010 by the American Academy of Pediatrics. All and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elkpublication, it has been published continuously since 1948. PEDIATRICS is owned, published, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly
. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
Impact of Rapid Leukodepletion on the Outcome ofSevere Clinical Pertussis in Young Infants
WHAT’S KNOWN ON THIS SUBJECT: Clinical B pertussis infectionin young infants can cause rapidly progressive multiple organfailure and severe pulmonary hypertension; there is a pooroutcome even with ECMO. Severe leukocytosis is typically seen inthese severe cases.
WHAT THIS STUDY ADDS: Treatments aimed at reducing veryhigh WBC counts rapidly in severe infantile pertussis seem toimprove survival compared with outcomes seen in historicalcontrols and ELSO registry data.
abstract +
OBJECTIVES: Bordetella pertussis is a common, underrecognized, andvaccine-preventable cause of critical illness with a high mortality ininfants worldwide. Patients with severe cases present with extremeleukocytosis and develop refractory hypoxemia and pulmonary hyper-tension that is unresponsive to maximal intensive care. This may re-flect a hyperviscosity syndrome from the raised white blood cell (WBC)count. Case reports suggest improved outcomes with exchange trans-fusion to reduce the WBC count. Our objective was to quantify possiblebenefits of aggressive leukodepletion.
METHODS: We, as a regional PICU and extracorporeal membrane oxy-genation referral center, adopted a strategy of aggressive leukodeple-tion in January 2005. The impact of this strategy on crude and casemix–adjusted survival of all infants who were critically ill with B per-tussis were compared with control subjects from January 2001 toDecember 2004 and Extracorporeal Life Support Organisation registrydata.
RESULTS: Nineteen infants (7 [37%] boys) received intensive care for Bpertussis from 2001 to 2009. Admission WBC counts were equivalent in2 time periods: 2001–2004 (mean: 52 000/�L) and 2005–2009 (mean:75 000/�L). In 2001–2004, 5 (55%) of 9 patients survived the ICU. Be-tween 2005 and 2009, 9 (90%) of 10 patients survived. When case-mixadjustment for age, WBC count, and extracorporeal membrane oxygen-ation referral were considered, the 2001–2004 predicted survival (4.4[49%] of 9.0) was equivalent to the observed mortality (4.0 [44%] of9.0). Between 2005 and 2009, observedmortality (1.0 [10%] of 10.0) wassignificantly better than predicted (4.7 [47%] of 10.0).
CONCLUSIONS: Leukodepletion should be considered in critically illinfants with B pertussis and leukocytosis. Pediatrics 2010;126:e816–e827
AUTHORS: Helen E. Rowlands, MBBS,a Allan P. Goldman,MBBChB,a Karen Harrington, MD,a Ann Karimova, MD,a
Joe Brierley, MBChB,b Nigel Cross, ACP,a Sophie Skellett,MBBChir,b and Mark J. Peters, PhDb,c
aCardiac Critical Care Unit and bPaediatric and NeonatalIntensive Care Units, Great Ormond Street Hospital for ChildrenNHS Trust, London, United Kingdom; and cCritical Care Group,Portex Unit, Institute of Child Health, University College London,London, United Kingdom
KEY WORDSpertussis, critical care, pulmonary hypertension, ECMO, whiteblood cell count
ABBREVIATIONSECMO—extracorporeal membrane oxygenationELSO—Extracorporeal Life Support OrganisationWBC—white blood cellPIM—Pediatric Index of MortalityPICANet—United Kingdom Paediatric Intensive Care AuditNetworkCI—confidence interval
Drs Rowlands, Goldman, and Karimova contributed to studydesign, data collection, interpretation, and writing; DrHarrington and Mr Cross contributed to study design and datacollection; Drs Brierley and Skellett contributed to datacollection, analysis, and writing; and Dr Peters contributed tostudy design, data collection, analysis, interpretation figures,and writing.
www.pediatrics.org/cgi/doi/10.1542/peds.2009-2860
doi:10.1542/peds.2009-2860
Accepted for publication Jun 15, 2010
Address correspondence to Mark J. Peters, PhD, Institute ofChild Health, Critical Care Group, Portex Unit, London, WC1N 1EH,England. E-mail: [email protected]
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
Copyright © 2010 by the American Academy of Pediatrics
FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.
e816 ROWLANDS et al. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
Bordetella pertussis is the fifth-leading cause of vaccine-preventabledeaths in children around the world1
and is a common2 and underrecog-nized cause of critical illness for in-fants.3,4 Mortality in young infants ishigh despite critical care, and manysurvivors experience long-termsequelae.5,6
Infants may present with apnea orbronchopneumonia, although a subsetof younger infants develop a fulminantcourse with refractory hypoxemia andpulmonary hypertension.6,7 Supportivetherapies, including extracorporealmembrane oxygenation (ECMO), areless effective for these cases than forother forms of cardiorespiratory fail-ure in infancy.8–10 The ExtracorporealLife Support Organisation (ELSO) regis-try from 1992 to 2009 contains 169cases of B pertussis pneumonia thatwere treated with ECMO. The case fa-tality rate of 69.8% is higher than forother respiratory indications, includ-ing respiratory syncytial virus pneu-monia (20%). Infants who are youngerthan 6 weeks and have pertussis havethe worst outcomes (83.6% case fatal-ity rate).
High white blood cell (WBC) counts aretypically seen in infants with the mostsevere disease,11 and WBC counts of�100 000/�L have been associatedwith a uniformly fatal course with con-ventional treatment.5,10,12 The hypothe-sis is that in the vulnerable youngestinfants with reactive pulmonary arte-rioles and immature coagulation andfibrinolytic systems, severe leukocyto-sis, contributes directly to severity viaa hyperviscosity syndrome and pulmo-nary arteriolar thrombosis.6,8,13 Thismicrovascular thrombotic obstructionto pulmonary blood flow is resistant toconventional management of pulmo-nary hypertension with pulmonary va-sodilators; indeed, inhaled nitric oxidemay worsen the effects of the pertus-sis tracheal cytotoxin.1,14 Postmortem
examination of the lungs reveals wide-spread tissue necrosis and pulmonaryarteriolar thromboses that containhigh numbers of leukocytes.6,15
We have used empirical strategiesaimed at reducing the impact of severeleukocytosis, including exchangetransfusions and mechanical leuko-cyte depletion via an extracorporealfilter. Similar techniques have beenreported in a total of 5 cases in 3series.12,13,16
We report the impact of this strategyon outcomes for infants who wereyounger than 3 months, had severe Bpertussis, and were referred for inten-sive care or ECMO at our institution.Comparison with historical controlsand case mix–adjusted mortality aremade.
METHODS
Hypothesis
Adoption of a strategy of rapid leu-kodepletion in critically ill infants whoare younger than 90 days and have Bpertussis infection and leukocytosis isassociated with a more rapid fall in to-tal WBC count than historical controlpatients and improved case mix–adjusted ICU survival in comparisonwith historical controls.
Study Design/Ethical Review
This case series describes all consec-utive cases of infants who wereyounger than 90 days, had proven Bpertussis infection, and were admittedfor intensive care in our institution be-fore (January 2001 to December 2004)and after (January 2005 to August2009) leukodepletion was adopted as achange in practice. In view of our pub-lished high case fatality rate at ourinstitution for infant pertussis withextreme leukocytosis10 and the world-wide case fatality rates of 68% to 84%with ECMO for pertussis, a randomizedstudy was not considered appropriate.Discussion with the local ethics com-
mittee provided support for leu-kodepletion as a pathophysiologicallybased intervention in patients at veryhigh-risk that did not require addi-tional formal ethical review. The un-proven nature and rationale for thisintervention was discussed in detailwith parents of affected infants.
Setting and Patients
Great Ormond Street Hospital ICUs aretertiary regional centers, and the Car-diac Unit is 1 of 4 designated UnitedKingdom ECMO centers. Referrals of in-fants with clinical B pertussis infectionaremade either to the PICU with apneaor respiratory failure or direct to theECMO service with established cardio-respiratory failure that is unrespon-sive to conventional management.
All infants who were younger than 3months’ postnatal age, had a positivepolymerase chain reaction or culture-proven B pertussis, and were admittedto the ICUs between January 2001 andAugust 2009 were identified from clin-ical databases by discharge diagnosis.Cases from 1994 to 1999 have beenpublished previously.10
Age, gender, total WBC count and dif-ferential counts, and details of historyand previous therapy were recordedfor all patients. Survival to ICU dis-charge and details of techniques to re-duce WBC counts are presented.
Management on ICU
All cases of respiratory failure aremanaged on a low-tidal volume (4–7mL/kg) strategy with permissive hy-percarbia (pH 7.25–7.35) tolerated inthe absence of clinical or echocardio-graphic evidence of pulmonary hyper-tension. Patients who require a meanairway pressure of �16 cm H2O toachieve adequate oxygenation aretransferred to high-frequency oscilla-tory ventilation (Sensormedics 3100A[Yorba Linda, CA]). Inhaled nitric oxy-gen therapy is not used routinely be-
ARTICLE
PEDIATRICS Volume 126, Number 4, October 2010 e817. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
yond the neonatal period, although itmay be considered in cases of severeright ventricular failure. ECMO is con-sidered when there is evidence of fail-ure of conventional management.
Leukodepletion
Exchange transfusion was performedover 4 hours via arterial and centralvenous lines with replacement fluid of200 mL/kg given as a combination of4.5% human albumin solution andpacked red blood cells to target a finalhematocrit level of 0.40 to 0.45 and aWBC count of�50 000/�L. Leukofiltra-tion was performed with a WBC filtersited in the bridge of the ECMO circuitat the time of circuit priming, which ispublished as Supplemental Informa-tion. After commencing ECMO, thebridge was opened (flow �100 mL/min) to the WBC filter. The WBC filterwas removed from the circuit once theWBC count was�15 000/�L (Fig 1).
Statistical Analysis
Statistical analyses were undertakenby using MedCalc 9.5.2.0 for Windows(Mariakerke, Belgium). WBC counts ap-proximated to normal distributions,and the maximum values recorded atGreat Ormond Street Hospital admis-sion values in the 2 cohorts were com-pared by t tests, whereas the percent-age fall in WBC count in the first 10hours were compared by t tests.
Other continuous variables were com-pared with the Mann-Whitney U test.The crude case fatality rates werecompared with Fisher’s exact test. Inaddition, because of the known highmortality risk in the extreme leuko-cytosis and ECMO subsets of pa-tients, comparisons were made be-tween predicted and observed ICUmortality rates with the 1-sampletest of proportion.
Consideration was given to using stan-dard severity of illness physiologicscoring systems for predicted mortal-
FIGURE 1Proposed algorithm for leukodepletion in infants who are younger than 90 days and have clinical Bpertussis that requires intensive care.
e818 ROWLANDS et al. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
ity estimates (eg, Pediatric Index ofMortality [PIM]17) but our publisheddata10 and a review of the United King-dom Paediatric Intensive Care AuditNetwork (PICANet) data 2004–2006(www.picanet.org.uk) indicated thatthese underestimate very significantlythe risk for death from pertussis in in-fants. For example, for all recordedcauses of respiratory failure at �90days of age (n� 2101), PIM predicted157 deaths (7.5% mortality), whereas62 deaths were observed (3% mortal-ity), giving a standardizedmortality ra-tio of 0.39. In contrast, among the 31cases of pertussis at�90 days of ageidentified, the predicted mortalityfrom PIM was 3 (9.7% mortality),whereas 9 deaths (29% mortality)were observed (standardized mortal-ity ratio: 3). (R.C. Tasker, PhD, and R.Parslow, PhD, personal written com-munication, 2008); therefore, we as-signed values to expected ICU mortal-ity on the basis of the most specificavailable published data sources: theELSO registry 2000–2009 and our pre-vious experience of pertussis caseswith extreme leukocytosis in our insti-tution and relevant PICANet data.10
All patients who had pertussis at�90days of age were assigned to 1 of thefollowing 4 groups:
1. younger than 6 weeks’ correctedpostnatal age and receiving ECMO(risk for mortality: 0.78);
2. older than 6 weeks’ corrected post-natal age and receiving ECMO (riskfor mortality: 0.63);
3. any proven B pertussis with admis-sion WBC count of �100 000/�L(risk for mortality: 1.00); or
4. any proven B pertussis with admis-sion WBC count of�100 000/�L andreferral to ICU but not for ECMO(risk for mortality was low and ap-proximated the observed PICANet�90-day infant respiratory failurevalue: 0.03).
Patients who were referred for ECMOand also had a WBC count of�100 000/�L were treated as for astandard ECMO risk because these val-ues were generated from a muchlarger sample (n� 169 vs 4).
RESULTS
Cases
Between 2001 and 2004, 9 patients withclinical B pertussis that required in-tensive care were confirmed by cul-ture or polymerase chain reaction. Sixwere referred for and received ECMO,and an additional 3 required PICUalone. Between 2005 and 2009, 10 pa-tients with proven B pertussis wereidentified. Five were referred for andreceived ECMO, and 5 required PICUalone. Comparisons between the 2periods (Table 1) reveals that patientswere referred for ECMO after signifi-cantly shorter periods of ventilation inthe referring ICUs in 2005–2009 (me-dian 1 day) compared with 2001–2004(median: 5.5 days; P � .04). Details ofindividual patients are shown inTable 2.
Leukocytosis
In the 2001–2004 cohort, mean WBCcount on ICU admission was 52 000/�L(95% confidence interval [CI]: 36 000–68 000). Between 2005 and 2009, meanWBC count on ICU admission was75 000/�L (95% CI: 40 000–110 000;P� .2 vs 2001–2004 cohort, 2-sample ttest). In that period, 5 infants had pre-sentation WBC counts of �90 000/�L,including 2 patients with WBC counts�100 000/�L. In all cases the leukocy-tosis was a combination of lympho-cytes and left-shifted neutrophils(30%–60%).
The time course of WBC count duringICU admissions in each of the 2 cohortsare shown in Fig 2. Even without spe-cific leukofiltration in the 2001–2004cohort, ECMO was associated with arapid fall in WBC count. The total WBC
count on admission to our institutionwas reduced by a mean of 55% of theadmission value by 10 hours (95% CI:40%–70%; P � .001, 1-sample t test).Patients who did not receive ECMO ex-hibited a slight rise in WBC count(14%–22%) at 10 hours (data notshown). In the 2005–2009 cohort, theaddition of the leukofiltration strategyto ECMO meant that admission WBCvalues fell by a mean of 83% at 10hours. This represents a significantlygreater fall in total WBC count (P �.017, 2-sample t test) in the latter pe-riod. The 3 patients who received adouble volume exchange transfusionwithout ECMO exhibited a similar fall inWBC count (67%–74% fall by 10 hours).
Outcome
Between 2001 and 2004, 5 patients sur-vived ICU (5 [55%] of 9), including 2 ofthe 6 patients who received ECMO. Be-tween 2005 and 2009, 9 patients sur-vived (9 [90%] of 10), including 4 of the5 patients who received ECMO. Thesevalues are not significantly different (5of 9 vs 9 of 10; P � .14 Fisher’s exacttest; Table 1).
When case mix of the 2001–2004 wasconsidered with risk estimates as de-scribed already, the sum of expectedICU mortality of 4.4 (49%) of 9.0 wasequivalent to the observed mortality of4.0 (44%) of 9.0 (95% CI: 19%–73%;1-sample test for proportion P � .87against expected mortality). In con-trast, the 2005–2009 cohort had an ex-pected mortality of 4.7 (47%) of 10.0,but the observed mortality was 1.0(10%) of 10.0 (95% CI: 1.8%–40%; P �.02 against expected mortality; Table1). There were no late deaths.
These results might simply representa nonspecific improvement in ECMO orICU care, but ELSO registry data pro-vide no evidence of a general improve-ment in outcome. The overall mortalityfor patients who had B pertussis and
ARTICLE
PEDIATRICS Volume 126, Number 4, October 2010 e819. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
received ECMO remained stable be-tween 2000–2004 (74.1%; 40 of 54 pa-tients) and 2005–2009 (75%; 51 of 68patients, excluding the 5 patients and 4survivors presented here).
DISCUSSION
The poor outcome for a subset of pa-tients with clinical B pertussis infec-tion in young infants has persisted de-spite modern critical care.7,18 Severalauthors have observed the associationwith leukocytosis,8,10 and autopsy dataare consistent with a hyperviscositysyndrome with widespread lung ne-crosis. Three case series to date sug-gest the value of techniques aimed atreducing the WBC counts rapidly.12,13,16
To our knowledge, this is the largestseries on leukodepletion, and data wepresent provide support that leukofil-tration or exchange transfusion is ef-
fective in rapidly dropping the num-bers of circulating WBCs. Although wefully acknowledge themany limitationsof this study, we have provided theclearest indication to date that a strat-egy of leukodepletion may be associ-ated with improved ICU survival.
Young infants such as those reportedhere have few options of immune-protection against pertussis beforecompleting their primary immuniza-tion schedule. No neonatal vaccinationagainst pertussis is yet approved;therefore, the ideal protective strategyfor infants who are younger than 2months would be universal vaccina-tion of adults and adolescents, be-cause they represent the main sourceof infection for these young infants. Inreal life, this seems almost impossibleto achieve for many reasons. The
global impact of pertussis in this vul-nerable population remains high, and,in contrast to other conditions, thepoor prognosis for pertussis has beenlittle improved by the many develop-ments in respiratory support for criti-cally ill children (high-frequency oscil-latory ventilation, inhaled nitric oxide,ECMO).18 In this context, our study mayrepresent a real advance for care ofthese infants that can be delivered inany intensive care setting without theneed for these expensive, advancedtechniques.
We are unable to exclude the possibil-ity that the observed improvement incase mix–adjusted outcome reflects aselection bias resulting from changesin practice of referral or commence-ment of ECMO. The shorter prerefer-ral for ECMO ventilation times in
TABLE 1 Characteristics and Outcome of Patients Who Were Admitted to the ICU for Severe Pertussis, 2001–2004 and 2005–2009
Parameter 2001–2004 2005–2009 P
ICU admissions with proven pertussisCorrected postnatal age�3 mo, n 9 10Corrected postnatal age�6 wk, n (%) 5 (55) 5 (50) 1.00a
Male gender, n (%) 3 (33) 4 (40) 1.00a
Total duration of symptoms before admission to Great OrmondStreet PICU, median (range), d
11 (2–26) 8 (4–21) .56b
Duration of ventilatory support in referring institution, median(range), d
4 (0–8) 1 (0–3) .16b
No. of patients who received ECMO in referring institution 6 5 .04b
Duration of ventilatory support of patients who received ECMOin referring institution, median (range), d
5.5 (1.0–8.0) 1.0 (1.0–3.0)
WBC count on ICU admission, mean (95% CI),/�L 52 000 (36 000–68 000) 75 000 (40 000–119 000) .20c
Maximum recorded WBC count, mean (95% CI),/�L 52 000 (36 000–68 000) 83 000 (49 000–118 000) .09c
Patients with maximum WBC count, n (%),/�L�100 000 0 (0) 2 (20) .47a
�90 000 0 (0) 5 (50) .03a
�70 000 2 (22) 7 (70) .07a
�50 000 5 (55) 8 (80) .34a
ICU mortalityn (%) 4 (44) 1 (10)95% CI 19–73 1.8–40 .14a
ECMO mortalityn (%) 4 (66) 1 (20)95% CI 30–90 4–62 .24a
Predicted mortality (ELSO and 1995–1999 data)ICUn (%) 4.4 (49) 4.7 (47)Pd .76 .02ECMOn (%) 4.3 (71) 3.5 (69)Pd .76 .02
Comparisons between groups were made with a Fisher’s exact test, b Mann-Whitney U test, and c unpaired t test.d The 1-way test of proportions was used to compare predicted and observed mortality.
e820 ROWLANDS et al. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
TABLE2ClinicalDetailsofPatients
AgeatICU
Admission,wk,
Gender,
Gestation
Weight,kg
History
TreatmentBefore
AdmissiontoICU
Microbiological
Findings
AdmissionWBCCount
(�103 /
�L)/Platelet
Count(
�106 /
�L)
PeakWBCCount
(�103 /
�L)/
PlateletCount
(�106 /
�L)
Leukodepletion
ICUAdmissionSeverityofOrganFailuresandTherapy
Outcome
Cardiovascular
Respiratory
Neurologic
Other
2001–2004
8, F, term,
4.0
14dofcough,
increasing
difficulty
feeding,and
apnea
Observationfor
apnea;referral
forICU
BpertussisPCR
positive;no
co-infection
detected
29/257
29/290
NoNone
Intermittentapnea;
notventilated
None
None
Alive;ICUuntil
day3
9, M,term,
5.8
4dofcoughand
respiratory
distress
Intubatedand
ventilatedfor
respiratory
failureat
regionalICU
for7dbefore
ECMOreferral
BpertussisPCR
positive;no
co-infection
detected
88/694
88/674
NoPoorbiventricular
function;VA-ECMO
onadmissionfor
epinephrine-and
norepinephrine-
resistantsystemic
hypotension
OI38
Seizures
Acuterenalfailure;
hemofiltration
onECMO
days1–7
Died;lungbiopsy
widespread
infarction;
failedECMO
weanafter
31d
6, F, 35wk,
3.5
2wkofcoughand
increasing
respiratory
distress
Intubatedand
ventilatedfor
respiratory
failureat
regionalICU
for4dbefore
ECMOreferral
BpertussisPCR
positive;no
co-infection
detected
35/64
35/149
NoPoorrightventricular
function;severe
pulmonary
hypertension;VA-
ECMOwithin3hof
admissionfor
epinephrine-and
norepinephrine-
resistantsystemic
hypotension
OIdeteriorated28
to40despite
high-frequency
ventilation,
inhalednitric
oxide,and
surfactant;
VA-ECMO
Seizuresand
encephalopathic
EEG
Acuterenalfailure;
hemofiltration
days2–5;ECMO
Alive;ECMOuntil
day9;
dischargedto
referringICU
day10
4, F, term,
3.6
2dofcoughand
respiratory
distress
Intubatedfor
poorgas
exchangeand
apnea;referral
forICU
BpertussisPCR
positive;no
co-infection
detected
38/572
39/572
NoNone
Moderate
respiratory
failure;OI15
None
None
Alive;ventilation
untilday6;ICU
untilday7
6, F, term,
3.5
1wkofcoughand
respiratory
distress
Intubatedfor
poorgas
exchange;
referralforICU
BpertussisPCR
positive;no
co-infection
detected
54/672
54/672
NoNone
Moderate
respiratory
failure;OI12
None
None
Alive;ventilation
untilday6;ICU
untilday7
9, M,term(twin),
4.0
10dofcoughand
respiratory
distress
8dward
inpatientthen
intubatedand
ventilatedat
regionalICU
for8dbefore
ECMOreferral
BpertussisPCR
positive;RSV
co-infection
detectedfor
15d
10/263
63/263
NoNormal
echocardiogram;
low-dosedopamine
andnoradrenaline
AdmissionOI24;
initial
stabilization
withhigh-
frequency
ventilationand
inhalednitric
oxidetherapy;
VV-ECMOday16
fordeterioration;
OI52
Seizures
None
Alive;21dof
VV-ECMO;
dischargedto
referringICU
day24
ARTICLE
PEDIATRICS Volume 126, Number 4, October 2010 e821. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
TABLE2Continued
AgeatICU
Admission,wk,
Gender,
Gestation
Weight,kg
History
TreatmentBefore
AdmissiontoICU
Microbiological
Findings
AdmissionWBCCount
(�103 /
�L)/Platelet
Count(
�106 /
�L)
PeakWBCCount
(�103 /
�L)/
PlateletCount
(�106 /
�L)
Leukodepletion
ICUAdmissionSeverityofOrganFailuresandTherapy
Outcome
Cardiovascular
Respiratory
Neurologic
Other
3, M,36wk,
3.7
3dofcough
andrespiratory
distress
Intubatedand
ventilated
forrespiratory
failureat
regionalICU
for1d
before
ECMO
referral
Bpertussis
PCRpositive;
noco-
infection
detected
34/350
34/350
NoSeverepulmonary
hypertension;
dopamineand
epinephrine
AdmissionOI
18;severe
respiratory
acidosis;
high-
frequency
ventilation
andinhaled
nitricoxide
therapy;VV-
ECMOwithin
2hof
admission
Seizures
None
Died;VA-ECMO
untilday37;
weanedbut
rapid
recurrenceof
respiratory
andcardiovascular
failure;on
autopsy,
widespread
pulmonaryand
cardiac
infarction
4, F, 30wk,
2.7
2dofrespiratory
distress
Intubatedand
ventilated
forrespiratory
failureat
regionalICU
for6d
before
ECMO
referral
Bpertussis
PCRpositive;
adenovirus
andenteroviral
PCRpositive
78/123
78No
Systemic-level
pulmonary
hypertension;
leftventricular
hypertrophy
Profound
hypoxia
despitehigh-
frequency
ventilation
andinhaled
nitricoxide
therapy;
immediate
VV-ECMO
convertedto
VAat12h
Seizures
Acuterenal
failure;
hemofiltration
onECMO
Died;ECMO
withdrawnat
day25
7, F, term,
2.7
12dofcough
andrespiratory
distress
Intubatedand
ventilated
forrespiratory
failureat
regionalICU
for5d
before
ECMO
referral
Bpertussis
PCRpositive;
RSVco-
infection
detected
for14d
52/130
52/249
NoVA-ECMOon
admissionfor
suprasystemic
pulmonary
arterial
pressureand
catecholamine-,
epoprostenol-,
andinhaled
nitricoxide–
resistantshock
OI70;
immediate
VA-ECMO
Seizures
Acuterenal
failure;
hemofiltration
onECMO
Died;weaned
fromVA-ECMO
onday26but
rapid
recurrenceof
respiratory
andcardiovascular
failure
e822 ROWLANDS et al. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
TABLE2Continued
AgeatICU
Admission,wk,
Gender,
Gestation
Weight,kg
History
TreatmentBefore
AdmissiontoICU
Microbiological
Findings
AdmissionWBCCount
(�103 /
�L)/Platelet
Count(
�106 /
�L)
PeakWBCCount
(�103 /
�L)/
PlateletCount
(�106 /
�L)
Leukodepletion
ICUAdmissionSeverityofOrganFailuresandTherapy
Outcome
Cardiovascular
Respiratory
Neurologic
Other
2005–2009
12,F, term,
6.0
11dofcough
andrespiratory
distress
Intubatedand
ventilated
forrespiratory
failureat
regionalICU
for3d
before
ECMO
referral
Bpertussis
PCRpositive;
noco-
infection
detected
58/973
58/973
Leukofiltration
onECMO;
no complications
Epinephrine-and
dopamine-
responsive
shock
Widespreadair
leak;OI37
despitehigh-
frequency
ventilation
andinhaled
nitricoxide;
immediate
VA-ECMO
Seizures
None
Alive;VA-ECMO
untilday20,
ventilation
untilday25;
ICUdischarge
day26
6, F, term,
3.6
3wkofcough,
increasing
respiratory
distress,
andcyanotic
episodes
Intubatedand
ventilated
forapneaat
local
hospital;
referralfor
ICU
Bpertussis
PCRpositive;
noco-
infection
detected
23/433
23/433
None
Stable;normal
echocardiogram
OI9;moderate-/
low-pressure
ventilation
None
None
Alive; ventilation
untilday4;
ICUdischarge
day5
4, F, 35wk,
3.1
Coughand
respiratory
distressfor
1wk
Intubatedand
ventilated
forrespiratory
failureat
regionalICU
for1d
before
ECMO
referral
Bpertussis
PCRpositive;
noco-
infection
detected
77/315
77/315
Leukofiltration
onECMO
for48h;no
complications
Catecholamine-
andphosphodiesterase
inhibitor–resistant
shock;severe
pulmonary
hypertension
Profound
hypoxia
despitehigh-
frequency
ventilation
andinhaled
nitricoxide;
VA-ECMO10h
after
admission
Seizures
Hemofiltration
forrenal
failure
Died;VA-ECMO
for27d,
failureto
wean;
asystole
during
circuit
maintenance
8, F, term,
4.8
3dofcough
andrespiratory
distress
Intubationand
high-
frequency
oscillatory
ventilation
atregional
ICUfor1d
before
ECMO
referral
Bpertussis
PCRpositive;
noco-
infection
detected
179/538
179/538
Leukofiltration
onECMO
for4d;no
complications
Intermittent
supraventricular
tachycardia;
structurally
normalheart;
severe
pulmonary
hypertension
andright
ventricular
dysfunction
OI22;high-
frequency
ventilation;
VV-ECMO3h
after
admission
convertedto
VAonday3
Seizures
48hof
hemofiltration
forrenal
failure
Alive;ECMO
untilday29,
ventilation
untilday31;
ICUdischarge
day32
ARTICLE
PEDIATRICS Volume 126, Number 4, October 2010 e823. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
TABLE2Continued
AgeatICU
Admission,wk,
Gender,
Gestation
Weight,kg
History
TreatmentBefore
AdmissiontoICU
Microbiological
Findings
AdmissionWBCCount
(�103 /
�L)/Platelet
Count(
�106 /
�L)
PeakWBCCount
(�103 /
�L)/
PlateletCount
(�106 /
�L)
Leukodepletion
ICUAdmissionSeverityofOrganFailuresandTherapy
Outcome
Cardiovascular
Respiratory
Neurologic
Other
2, M,term,
4.4
4dofcough
andrespiratory
distress
Intubatedand
ventilated
forrespiratory
failureat
regionalICU
for3d
before
ECMO
referral
Bpertussis
PCRpositive;
noco-
infection
detected
99/72
99/161
Leukofiltration
onECMO
for20h;no
complications
Catecholamine-
andphosphodiesterase
inhibitor–resistant
shock;
structurally
normalheart;
suprasystemic
pulmonary
hypertension,
mitraland
tricuspid
regurgitation,
anddynamic
leftventricular
outflowtract
obstruction
OI5;high-
frequency
ventilation;
VA-ECMO3h
after
admission
Seizures
Hemofiltration
for48h
during
secondary
sepsis
Alive;VA-ECMO
untilday28
becauseof
persistent
right
ventricular
failure;
ventilation
untilday46;
ICUuntilday
50
8, F(twin),
term,
4.5
1wkofcough
andfever
Intubationand
ventilation
forrespiratory
failureand
shockat
regionalICU
for1d
before
ECMO
referral
Bpertussis
PCRpositive;
noco-
infection
detected
99/638
99/638
Exchange
transfusion
atregional
ICUdidnot
reduce
WBCcount;
leukofiltration
onECMO
for24h;no
complications
Norepinephrine-
anddopamine-
responsive
shock;
structurally
normalheart;
pulmonary
hypertension
OI3;pressure
control
ventilation
tidalvolume
5mL/kg;VA-
ECMO4h
after
admission
Seizures
None
Alive;VAECMO
untilday21,
ventilation
untilday58
11,M,32wk,
2.7
2wkof
increasing
coughand
respiratory
distress
Intubatedand
ventilatedfor
respiratory
failureat
local
hospital;
referralfor
ICU
Bpertussis
PCRpositive;
Haemophilus
influenza
cultured1
wkearlier
92.5/495
118/495
Double
volume
exchange
transfusion
day1ICU;
no complications
Fluid-responsive
shock
OI6;pressure
control
ventilation
tidalvolume
5–7mL/kg
Seizures
None
Alive; ventilation
untilday6;
ICUuntilday
8
10,F, term,
3.6
5dofcough
andrespiratory
distress
Intubatedand
ventilated,
seizuresand
respiratory
failureat
local
hospital;
referralfor
ICU
Bpertussis
PCRpositive;
noco-
infection
detected
81/727
92/727
Double
volume
exchange
transfusion
day1ICU;
no complications
Dopamine-
responsive
shock
OI23;pressure
control
ventilation
tidalvolume
Seizures;focal
cerebritison
CTbrainscan
None
Alive; ventilation
untilday9;
ICUuntilday
12
e824 ROWLANDS et al. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
2005–2009 indicates that this islikely to be a factor; however, ex-treme leukocytosis has been a con-sistently adverse prognostic markerto date, so the trend toward higherWBC counts in the 2005–2009 cohortargues against a reduction in initialseverity of illness underlying thesechanges. In particular, 5 patients inthe latter period had presentationWBC counts of �90 000/�L.
This study has a number of majorweaknesses: it was performed in a sin-gle center, albeit 1 that will receive themajority of referrals for ECMO inthe south of England. Furthermore, thenumbers involved are very small,meaning that a selection bias is inher-ent in publishing the series now ratherthan in an additional 5 years, when anydifference could be assessed with agreater degree of confidence. Perhapsthe greatest weakness is the attemptsmade to predict case mix–adjustedmortality, including cases from 10 to15 years ago. We attempted to derivevalues from the largest series andcarefully maintained databases andsuggest that these are the best avail-able. The stability of the poor out-comes on the ELSO registry is striking.We would welcome any data that allowadditional refinement of these values.
Our failure to undertake a randomizedstudy deserves specific comment. Leu-kodepletion was introduced in an at-tempt to alter what the clinical teamrecognized as a very severe and typi-cally fatal natural history. A random-ized studywould have had to run amin-imum of 8 years assuming 100%recruitment and stable control groupoutcomes. Clearly, both are rare find-ings. Furthermore, the early apparentsuccess of this approach means thatequipoise has been lost. Prof Sir Mi-chael Rawlins’ in his Harveian Oration2008 reminded us of the value of thistype of historical-controlled trial, par-TA
BLE2Continued
AgeatICU
Admission,wk,
Gender,
Gestation
Weight,kg
History
TreatmentBefore
AdmissiontoICU
Microbiological
Findings
AdmissionWBCCount
(�103 /
�L)/Platelet
Count(
�106 /
�L)
PeakWBCCount
(�103 /
�L)/
PlateletCount
(�106 /
�L)
Leukodepletion
ICUAdmissionSeverityofOrganFailuresandTherapy
Outcome
Cardiovascular
Respiratory
Neurologic
Other
5, M,35wk,
3.2
5dofcough
andmild
respiratory
distress
Intubatedand
ventilated
forapneaat
local
hospital;
referralfor
ICU
Bpertussis
PCRpositive;
RSVco-
infection
detected
3.1/369
11/505
None
Stable
Low-pressure
ventilation
for24h
None
None
Alive; ventilation
for2d;ICU
untilday4
3, M,term,
4.4
7dofcough
andrespiratory
distress
CPAPfor
apnea1dat
local
hospital
before
intubation
andventilation
forrespiratory
failure;
referralfor
ICU
Bpertussis
PCRpositive;
noco-
infection
detected
39/333
78/333
50mL/kg
exchange
transfusion
day2ICU,
WBCcount
78–50;
double
volume
exchange
transfusion
day3ICU,
WBCcount
65–21;no
complications
Day2systemic-
level
pulmonary
hypertension
andright
ventricular
dilation;day3
(after
exchange)no
evidenceof
pulmonary
hypertension,
rightventricle
lessdilated
Highfrequency
oscillatory
ventilation
None
None
Alive; ventilation
for20d;ICU
untilday22
PCRindicatespolymerasechainreaction;VA-ECMO,veno-arterialECMO;VV-ECMO,venovenousECMO;OI,oxygenationindex(meanairwaypressure
�fractionofinspiredoxygen
�arterialpartialpressureofoxygen[mmHg])20;EEG,electroenceph-
alogram;RSV,respiratorysyncytialvirus;CT,computedtomography;CAP,continuouspositiveairwaypressure.
ARTICLE
PEDIATRICS Volume 126, Number 4, October 2010 e825. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
ticularly in scenarios in which exist-ing treatments were ineffective—including defibrillation for ventricularfibrillation.19
A subtler weakness is that in focus-ing on WBC count specifically, wemay be addressing a paraphenom-enon of another process in pertussispathophysiology—for example, thedose of the initial inoculum that mayalso be modified by exchange trans-fusion and leukofiltration. This re-quires additional investigation.
A review article asked, “Rapidly fatalinvasive pertussis in young infants—how can we change the outcome?”18
Acknowledging the many limitationsof this nonrandomized study, we sug-gest that urgent leukodepletionshould be considered alongsideearly ECMO referral in cases of se-vere infant pertussis with extremeleukocytosis. We offer a simplealgorithm for clinicians suggest-ing indications for leukodepletion(Fig 1).
ACKNOWLEDGMENTSThis work was undertaken at Great Or-mond Street Hospital/University CollegeLondon Institute of Child Health, whichreceived a proportion of funding fromtheDepartmentofHealth’sNational Insti-tute for Health Research Biomedical Re-search Centres funding scheme.
We thank Prof James D. Cherry for in-valuable comments and advice on themanuscript and Roger Parslow andRobert Tasker for providing PICANetdata.
FIGURE 2Serial total WBC counts after admission to the ICU. A, 2001–2004: only patients who received ECMO (n� 6). B, 2005–2009: patients who received ECMO andwere treated with immediate leukofiltration after starting ECMO (n � 5). C, 2005–2009: patients who received exchange transfusions (n � 3). Time 0indicates approximate time of admission to our ICU;�, ICU nonsurvivors;Œ, ICU survivors. Note the difference in y-axis scale.
e826 ROWLANDS et al. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
REFERENCES
1. Mattoo S, Cherry JD. Molecular pathogene-sis, epidemiology, and clinical manifesta-tions of respiratory infections due to Borde-tella pertussis and other Bordetellasubspecies. Clin Microbiol Rev. 2005;18(2):326–382
2. Crowcroft NS, Booy R, Harrison T, et al. Se-vere and unrecognised: pertussis in UK in-fants. Arch Dis Child. 2003;88(9):802–806
3. Centers for Disease Control and Prevention(CDC) . Per tuss is—Uni ted Sta tes ,2001–2003. MMWR Morb Mortal Wkly Rep.2005;54(50):1283–1286
4. Vitek CR, Pascual FB, Baughman AL, MurphyTV. Increase in deaths from pertussisamong young infants in the United States inthe 1990s. Pediatr Infect Dis J. 2003;22(7):628–634
5. Surridge J, Segedin ER, Grant CC. Pertussisrequiring intensive care. Arch Dis Child.2007;92(11):970–975
6. Paddock CD, Sanden GN, Cherry JD, et al.Pathology and pathogenesis of fatal Borde-tella pertussis infection in infants. Clin In-fect Dis. 2008;47(3):328–338
7. Goulin GD, Kaya KM, Bradley JS. Severe pul-monary hypertension associated withshock and death in infants infected withBordetella pertussis. Crit Care Med. 1993;21(11):1791–1794
8. Williams GD, Numa A, Sokol J, Tobias V, DuffyBJ. ECLS in pertussis: does it have a role?Intensive Care Med. 1998;24(10):1089–1092
9. Halasa NB, Barr FE, Johnson JE, EdwardsKM. Fatal pulmonary hypertension associ-ated with pertussis in infants: does extra-corporeal membrane oxygenation have arole? Pediatrics . 2003;112(6 pt 1) :1274–1278
10. Pierce C, Klein N, Peters M. Is leukocytosis apredictor of mortality in severe pertussisinfection? Intensive Care Med. 2000;26(10):1512–1514
11. Mikelova LK, Halperin SA, Scheifele D, et al.Predictors of death in infants hospitalizedwith pertussis: a case-control study of 16pertussis deaths in Canada. J Pediatr. 2003;143(5):576–581
12. Grzeszczak MJ, Churchwell KB, Edwards KM,Pietsch J. Leukopheresis therapy for severeinfantile pertussis with myocardial and pul-monary failure. Pediatr Crit CareMed. 2006;7(6):580–582
13. Romano MJ, Weber MD, Weisse ME, Siu BL.Pertussis pneumonia, hypoxemia, hyperleu-kocytosis, and pulmonary hypertension: im-provement in oxygenation after a doublevolume exchange transfusion. Pediatrics.2004;114(2) Available at: www.pediatrics.org/cgi/content/full/114/2/e264
14. Heiss LN, Lancaster Jr Jr, Corbett JA, Gold-man WE. Epithelial autotoxicity of nitricoxide: role in the respiratory cytopathologyof pertussis. Proc Natl Acad Sci USA. 1994;91(1):267–270
15. Pooboni S, Roberts N, Westrope C, et al. Ex-tracorporeal life support in pertussis. Pedi-atr Pulmonol. 2003;36(4):310–315
16. Donoso AF, Cruces PI, Camacho JF, Leon JA,Kong JA. Exchange transfusion to reversesevere pertussis-induced cardiogenicshock. Pediatr Infect Dis J. 2006;25(9):846–848
17. Pearson GA, Stickley J, Shann F. Calibrationof the paediatric index of mortality in UKpaediatric intensive care units. Arch DisChild. 2001;84(2):125–128
18. Theilen U, Johnston ED, Robinson PA. Rap-idly fatal invasive pertussis in younginfants: how can we change the outcome?BMJ. 2008;337:a343
19. Rawlins M. De testimonio: on the evidencefor decisions about the use of therapeuticinterventions. Lancet. 2008;372(9656):2152–2161
20. PetersMJ, Tasker RC, Kiff KM, Yates R, HatchDJ. Acute hypoxemic respiratory failure inchildren: case mix and the utility of respira-tory severity indices. Intensive Care Med.1998;24(7):699–705
ARTICLE
PEDIATRICS Volume 126, Number 4, October 2010 e827. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from
DOI: 10.1542/peds.2009-2860 published online Sep 6, 2010; Pediatrics
Brierley, Nigel Cross, Sophie Skellett and Mark J. Peters Helen E. Rowlands, Allan P. Goldman, Karen Harrington, Ann Karimova, Joe
Young InfantsImpact of Rapid Leukodepletion on the Outcome of Severe Clinical Pertussis in
& ServicesUpdated Information
http://www.pediatrics.orgincluding high-resolution figures, can be found at:
Supplementary Material http://www.pediatrics.org/cgi/content/full/peds.2009-2860/DC1
Supplementary material can be found at:
Permissions & Licensing
http://www.pediatrics.org/misc/Permissions.shtmltables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,
Reprints http://www.pediatrics.org/misc/reprints.shtml
Information about ordering reprints can be found online:
. Provided by UCL Library Services on September 22, 2010 www.pediatrics.orgDownloaded from