Mandalakas_IJTLD_2011

Embed Size (px)

Citation preview

  • 8/4/2019 Mandalakas_IJTLD_2011

    1/16

    INT J TUBERC LUNG DIS e-publication ahead of print 8 June 2011

    2011 The Uniondoi:10.5588/ijtld.10.0631

    REVIEW ARTICLE

    Interferon-gamma release assays and childhood tuberculosis:systematic review and meta-analysis

    A. M. Mandalakas,* A. K. Detjen, A. C. Hesseling, A. Benedetti,# D. Menzies

    *Department of Pediatrics, Case Western Reserve University, Cleveland, Ohio, USA; Desmond Tutu TB Centre,Department of Paediatrics and Child Health, Stellenbosch University, Cape Town, South Africa; International UnionAgainst Tuberculosis and Lung Disease, New York, New York, USA; Respiratory and Epidemiology Clinical ResearchUnit, Montreal Chest Institute, Montreal Department of Medicine, and #Department of Epidemiology, Biostatistics &Occupational Health, McGill University, Montreal, Quebec, Canada

    Correspondence to: Anna Mandalakas, Global Health & Diseases, and Epidemiology & Biostatistics, Case Western ReserveUniversity, Cleveland, Ohio 44106, USA. Tel: (+1) 216 844 6246. Fax: (+1) 216 844 6265. e-mail: [email protected]

    Article submitted 14 October 2010. Final version accepted 21 December 2010.

    B AC KGR O UND: Children inected with Mycobacte-

    rium tuberculosishave signifcant risk o developing tu-

    berculosis (TB) and can thereore beneft rom preven-

    tive therapy.O B JE C T I VE : To assess the value o intereron-gamma

    release assays (IGRAs) and the tuberculin skin test (TST)

    in the diagnosis o TB inection and disease in children.

    ME T HO DS : Thirty-three studies were included, assess-

    ing commercial IGRAs (QuantiFERON-TB [QFT] and

    T-SPOT.TB) and TST. Reerence standards or inec-

    tion were incident TB or TB exposure. Test perormance

    or disease diagnosis was evaluated in studies assessing

    children with confrmed and/or clinically diagnosed TB,

    compared to children where TB was excluded.

    RESULTS: Two small studies measured incident TB in

    children tested with QFT and ound weak positive pre-

    dictive value. Association o test response with exposure

    categorized dichotomously or as a gradientwas sim-

    ilar or all tests. The sensitivity and specifcity o all tests

    were similar in diagnosing the disease. Stratifed analysissuggested lower sensitivity or all tests in young or hu-

    man immunodefciency virus inected children.

    C O NC L US I O NS : Available data suggest that TST and

    IGRAs have similar accuracy or the detection o TB in-

    ection or the diagnosis o disease in children. Heteroge-

    neous methodology limited the comparability o studies

    and the interpretation o results. A rigorous, standard-

    ized approach to evaluate TB diagnostic tests in children

    is needed.

    KE Y WO R DS : tuberculosis; pediatrics; TB inection;

    IGRAs; tuberculin skin test

    THE WORLD HEALTH ORGANIZATION (WHO)estimates that 2 billion people are inected with Myco-bacterium tuberculosis,1 creating a reservoir that leadsto 9 million new tuberculosis (TB) cases and 2 mil-lion deaths annually. Children carry approximately15% o the disease burden, 75% o which occurs inthe 22 high-TB-burden countries.2 TB exposure inyoung children requently occurs in the household,and the risk o non-household exposure increaseswith age. Young children not identifed or preventivetherapy have a disproportionately high risk o earlyprogression to disease and severe orms o TB.3 Theglobal TB epidemic is exacerbated by human immuno-defciency virus-1 (HIV) inection, especially in sub-Saharan Arica. HIV-inected children have an up to24-old higher risk o developing TB compared toHIV-negative children.4

    There is no gold standard or the detection o

    S U M M A R Y

    M. tuberculosis inection. The tuberculin skin test(TST) has known limitations in terms o sensitivityand specifcity,514 which may place children at risko under- or over-treatment or latent inection. Thiscould unnecessarily burden patients, amilies andresource-limited health systems, while potentiallyailing to identiy those children at highest risk o TBdisease progression.

    The diagnosis o active TB in children is also dif-cult due to the paucibacillary nature o childhoodTB, which makes bacteriological confrmation chal-lenging. As a result, the diagnosis o active disease inthis population oten relies on a composite o contacthistory, clinical symptoms and radiological fndings,as well as consideration o the TST reaction.15

    The TST and the recently developed intereron-gamma release assays (IGRAs) are immune-based diag-nostic tests. IGRAs measure ex-vivo intereron-gamma(IFN-) production by circulating T-lymphocyteswhen incubated in the presence o highly specifcAMM and AKD contributed equally to this work.

  • 8/4/2019 Mandalakas_IJTLD_2011

    2/16

    2 The International Journal of Tuberculosis and Lung DiseaseM. tuberculosis antigens (early secreted antigenic tar-get 6 [ESAT-6], culture fltrate protein 10 [CFP-10]TB7.7). There are two commercially available IGRAs:QuantiFERON-TB (QFT; QuantiFERON-TB Gold[QFT-G] and QuantiFERON-TB Gold In-Tube

    [QFT-GIT], Cellestis, Carnegie, VIC, Australia) andT-SPOT.TB (Oxord Immunotec, Oxord, UK). TheQFT test incubates whole blood and measures IFN-production with an enzyme-linked immunosorbentassay (ELISA), while T-SPOT.TB measures the num-ber o IFN- producing peripheral mononuclear cells(PBMCs). Like the TST, IGRAs cannot dierentiatebetween M. tuberculosis inection and active TB.

    A growing number o studies have compared theTST and IGRAs in the detection oM. tuberculosisinection and active TB in children. In the absence oa gold standard or inection, some studies have mea-

    sured sensitivity in populations with active TB as asurrogate or M. tuberculosis-inected persons, whileothers have used M. tuberculosis exposure as a surro-gate or inection.16,17 The value o IGRAs or thediagnosis o active TB and M. tuberculosis inectionin children remains unclear.

    We systematically reviewed the existing evidenceon the accuracy o IGRAs or the detection oM. tu-berculosis inection and diagnosis o active TB inchildren in settings with varying incidence o TB. Thisreview was presented at the July 2010 WHO expertmeeting where recommendations were drated orthe use o IGRAs in high-burden settings.

    METHODS

    In collaboration with an experienced librarian atCase Western Reserve University, we systematicallysearched Medline and Web o Science or articlespublished in English, French or Spanish rom 1998until January 2010. Search terms included tubercu-losis inection, or tuberculosis disease, AND Pedi-atrics or child*, AND QuantiFERON, or ELISpot,or intereron-gamma assays, or intereron-gammarelease assays, or T-cell assays, AND ESAT-6, or

    CFP10, or RD1 antigens. We searched reerencelists o all articles selected, existing systematic reviewsand an existing IGRA database (courtesy o M Pai,McGill University, Montreal, QC, Canada). We alsocaptured new publications identifed until 1 June2010 (Figure 1).

    We considered published peer-reviewed studiesthat included at least 20 children aged

  • 8/4/2019 Mandalakas_IJTLD_2011

    3/16

    IGRAs and childhood TB: systematic review 3

    For the assessment o sensitivity or the diagnosiso active TB, two defnitions o disease were accepted:defnite (confrmed) and probable TB. Defnite TB wasdefned as the presence o at least one clinical speci-men positive or M. tuberculosis on culture, or posi-

    tive acid-ast bacilli smear microscopy, or one histol-ogy sample positive or necrotizing granulomas, ornucleic acid amplifcation test positive or M. tubercu-losis. Probable TB was defned as the presence o threeor more o the ollowing: 1) chest radiologic fndingsconsistent with active TB; 2) typical symptoms suchas cough and weight loss; 3) other radiological evi-dence o active TB, including extra-pulmonary TB(e.g., computed tomography/magnetic resonance im-aging fndings consistent with TB meningitis) in con-junction with symptoms; 4) exposure to a case withactive inectious TB; and 5) response to appropriate

    anti-tuberculosis therapy. TST and IGRA could notbe included in either case defnition. For assessmento specifcity in the diagnosis o active TB, we useddata rom groups o children in whom active TB wassystematically excluded: 1) TB suspects with symp-toms suggestive o active TB, or 2) children with ex-posure to a case with active TB. We did not includedata rom groups i they did not have clear risk ac-tors or active TB or i it was unclear whether activeTB was systematically excluded in all subjects.

    Data were extracted on the number o positive,negative and indeterminate results or each o the re-erence standards and tests assessed. Data were also

    extracted to support stratifed and subgroup analysis.We used 13 QUADAS (Quality Assessment o Diag-nostic Accuracy Studies) items or the assessment ostudy quality and added an additional item on indus-try involvement.18,19

    Analysis

    Data were analyzed using SAS version 9.2 (SAS Insti-tute Inc, Cary, NC, USA) and STATA version 11(Stata Corporation, College Station, TX, USA). Sev-eral analytic approaches were used to evaluate testperormance or the detection oM. tuberculosis in-

    ection. In studies that measured exposure dichoto-mously, we assessed the association between test re-sult and TB exposure as an odds ratio (OR). Theseresults were pooled using both a fxed- and a random-eects approach.20 In studies that expressed exposurevia a gradient, we measured the correlation betweenexposure gradients and prevalence o positive tests.We estimated the Spearman correlation between thecategorical test result and the outcome and calculateda pooled correlation coefcient or each test withboth a fxed- and a random-eects approach.21 Next,we used the OR as the measure o eect to assessperormance by calculating the OR or each level oexposure relative to the reerence group, and thenan overall OR or increasing exposure category.We again used both fxed- and random-eects ap-

    proaches to estimate a pooled exposure eect acrossstudies.22,23 We estimated inter-study heterogeneity viaI2 statistics.20

    To evaluate test perormance or the diagnosis oactive TB, we used data rom studies that measured

    IGRAs in children with active TB to estimate testsensitivity and data rom studies that included anappropriate group in whom active TB was excludedto estimate test specifcity. We used a random eectsmeta-analysis to estimate the overall pooled estimateso sensitivity, specifcity and 95% confdence interval(95%CI; Proc Nlmixed in SAS), and used the exactbinomial likelihood approach to approximate thedistribution o the outcome o interest.24 We assessedheterogeneity by estimating the I2 statistic and associ-ated 95%CIs.25 To calculate the I2, zero cells werecorrected by 0.5. To explore sources o heterogene-

    ity, we perormed sub-group analyses stratifed bypredefned covariates o interest. These included TBincidence, age, World Bank income, bacille Calmette-Gurin (BCG) vaccination rate, HIV prevalence, TSTcut-o point and QFT test type (QFT-G or QFT-GIT).

    To assess the impact o indeterminate IGRA re-sults, sensitivity estimates were calculated with inde-terminates considered as alse-negatives; the relation-ship between the requency o IGRA indeterminateresults and important covariates was described.

    RESULTS

    Selection and quality of studiesOur search strategy identifed 240 studies. Fromthese, 67 articles were selected or ull text review,o which 31 articles describing 32 studies (one arti-cle describing two studies conducted in two coun-tries was included; Figure 1, Table 1).2693 Thirty-six studies that completed ull text review wereexcluded or a number o reasons, including datanot available;27,42,47,48,61,62,64,80,81,83,91 no exposuregradient used;31,34,71,74,86,89,90 pre-commercial IGRAsused;29,37,40,66,76,85 ewer than 20 children in-cluded;28,60,65,70,77 IGRA testing completed in TST-

    positive children only, precluding a comparisongroup;46,56,78,93 study population overlapped with anincluded study;39,84 and children with positive TSTreerred preerentially to the study, creating incorpo-ration bias.63

    Studies were perormed in 18 countries; the inci-dence o smear-positive TB was >25 per 100 000population in 10. Nineteen studies (59%) were per-ormed in high-income countries; these included 11%to 100% immigrant children. The 32 studies de-scribed results in 5525 children; our analysis included4122 o these children, as some sub-groups did notmeet the criteria or any reerence standard. Themean age o the children was 7.6 years (range 1.914.6); BCG coverage ranged between 8% and 100%.Six studies included HIV-inected children (range

  • 8/4/2019 Mandalakas_IJTLD_2011

    4/16

    4 The International Journal of Tuberculosis and Lung DiseaseTable 1 Characteristics of included studies sorted by World Bank income index

    Author, year,reference Country

    WorldBank

    incomeindex

    TBincidence*

    Includedfor analysis/total study

    n/N

    Mean ormedian age,

    yearsBCG%

    Immigrants%

    HIV-infected

    %

    Immunesuppressed

    %

    Adetifa, 201026 The Gambia LIC 113 225/285 7.3 77 NR 1 NROkada, 200879 Cambodia LIC 219 195/217 2.5 88 NR NR NRPetrucci, 200882 Nepal LIC 77 146/145 8.5 94 NR NR NRDogra, 200744 India LMIC 75 105/105 6 92 NR 1 57Nakaoka, 200673 Nigeria LMIC 131 207/207 7.4 90 NR NR NRWarier, 200992 India LMIC 75 100/143 6.7 NR NR NR NRHansted, 200951 Lithuania UMIC 30 120/120 13.9 100 NR 0 NRHesseling, 200954 South Africa UMIC 358 29/29 2.9 100 NR 0 0Mandalakas, 200869 South Africa UMIC 358 23/23 4.4 91 NR 100 0Nicol, 200975 South Africa UMIC 358 214/243 2.6 100 NR 0.5 NRPetrucci, 200882 Brazil UMIC 26 113/113 8.4 100 NR NR NRStavri, 200987 Romania UMIC 51 36/36 15 100 NR 100 67Stefan, 201088 South Africa UMIC 358 34/34 7 NR NR 0 100Bamford, 200930 UK HIC 6.8 195/195 8.5 53 42 NR NRBergamini, 200932 Italy HIC 3.3 480/480 10.9 38 71 0 0Bianchi, 200933 Italy HIC 3.3 16/336 4.5 52 97 0 0Chun, 200835 Korea HIC 40 162/227 3.2 100 NR 0 NR

    Connell, 200636 Australia HIC 2.8 9/101 3.9 33 78 NR NRConnell, 200838 Australia HIC 2.8 9/100 8.2 23 89 NR NRDetjen, 200741 Germany HIC 2.7 50/73 3.3 8 52 0 0Diel, 200843 Germany HIC 2.7 168/168 13.3 41 24 0 0Dominguez, 200845 Spain HIC 13 134/134 9.6 64 66 0 0Girardi, 200749 Italy HIC 3.3 9/161 13 9 11 NR NRGrare, 201050 France HIC 6.2 7/51 6.5 43 100 NR NRHaustein, 200952 UK HIC 6.8 27/253 7.3 51 NR 0.9 24Herrmann, 200953 France HIC 6.2 62/129 7 92 27 0 0Higuchi, 200957 Japan HIC 9.5 313/313 9.6 99 NR NR NRHiguchi, 200855 Japan HIC 9.5 102/102 14.6 100 NR NR NRHiguchi, 200958 Japan HIC 9.5 10/61 8.6 80 NR NR NRKampmann, 200959 UK HIC 6.8 91/209 9.2 60 33 0 NRLighter, 200967 USA HIC 1.8 207/207 9 36 34 0 0Lucas, 201068 Australia HIC 2.8 524/524 7.4 69 100 NR NR

    *Incidence of smear-positive TB per 100 000 population reported by the WHO in 2007.

    These data indicate the proportion of BCG-vaccinated/immigrants/HIV-infected or immune-suppressed children within the sub-group used for analysis. If thesubgroup data were not available, data for the whole group were used.Author contacted to ensure that there was no data overlap with other publications or to provide clarification/re-analysis of data.Data provided in the published manuscript were analyzed as two independent studies.Median age used as mean age was not reported.TB = tuberculosis; BCG = bacille Calmette-Gurin; HIV = human immunodeficiency virus; LIC = low-income country; NR = not reported; LMIC = low- andmiddle-income countries; UMIC = upper-middle-income country; HIC = high-income country; WHO = World Health Organization.

    0.5100),26,44,52,69,75,87 and three studies included chil-dren with non-HIV-related immune suppression.52,87,88

    Two studies reported incident TB in cohorts, 18described the association o tests with exposure (Ta-ble 2), 21 assessed test sensitivity in children with ac-tive TB, and nine provided data that could be used to

    estimate test specifcity or active TB. Studies evalu-ated one or more index tests, including T-SPOT.TB(n= 15), QFT-G (n= 10) and QFT-GIT (n= 21).Thirty studies provided TST data that could be usedor analysis (Table 1).

    Assessment o the study quality using QUADASshowed that only a minority o studies clearly re-ported on sampling methods or included a spectrumo subjects representative o patients in whom thetests might be used clinically. Blinding o clinicians toIGRA results was reported in 29% o the studies as-sessing active TB. O 32 studies, 11 (33%) were sup-ported by either or both IGRA manuacturers, mainlythrough donation o test kits. In 43% o the studiesassessing active TB, it remained unclear whether thereerence standard was applied to all subjects in-

    cluded (i.e., whether active TB was excluded in allsubjects). Seventeen o 21 studies (81%) describedtheir defnition o the reerence standard in detail su-fcient to support replication. However, there was stillwide variation among studies regarding the criteriaused and data provided on the defnition o confrmed

    or probable TB.Test ailure was inrequently reported. IGRA test

    ailure was defned as technical errors, ailed phlebot-omy or insufcient peripheral blood mononuclearcells (T-SPOT.TB). For the TST, ailure was defned asunread tests. Failure rates ranged rom 0% to 7% orQFT-GIT (n= 19), 0% or QFT-G (n= 6), 0% to21% or T-SPOT.TB (n= 15) and 0% to 11% or TST(n= 20). Average rates o indeterminate results acrossall studies were respectively 6.5% or QFT-GIT, 6.4%or QFT-G and 3.5% or T-SPOT.TB. In several stud-ies, indeterminate rates >10% were associated withmultiple actors such as young age, helminth co-inection and immune suppression.32,50,54,68,36,52,87,88,92In our analysis, a signifcant correlation o indetermi-nate results with specifc risk actors could not be

  • 8/4/2019 Mandalakas_IJTLD_2011

    5/16

    IGRAs and childhood TB: systematic review 5

    Table 2 Characteristics of dichotomous and graded expression of M. tuberculosis exposure, sorted by World Bank income index

    Author, year,reference

    Dichotomous comparison groups

    Description exposed Description unexposed

    LMIC Hansted, 200951 Household or school contact No TB contact, no symptoms, chest radiography normal

    Hesseling, 200954

    Known TB contact No known TB contactMandalakas, 200869 Known household contact No known household contactStefan, 201088 Known TB contact No known TB contact

    HIC Bianchi, 200933 TB contacts, Italian and immigrant Immigrants without TB contactChun, 200835 Household contacts Contact outside householdDominguez, 200845 Children from contact investigations TST-positive children detected during routine screening*Higuchi, 200957 Same class as index case (contact 90 h) Same school, different classes from index case (contact

  • 8/4/2019 Mandalakas_IJTLD_2011

    6/16

    6 The International Journal of Tuberculosis and Lung Diseaseshown; there was no dierence in the requency oIGRA indeterminate results in stratifed analysis.

    Results of review: latent TB infection

    Two longitudinal studies assessed incident active

    TB.43,57 A school outbreak investigation in Japan as-sessed 313 children with TST and QFT-G tests.57QFT-G-positive children and QFT-G-indeterminate/TST-positive children received preventive chemother-apy. One year ater the index case was reported, allchildren underwent chest radiography; no child de-veloped active TB during the 3-year ollow-up (posi-tive predictive value [PPV] 0%, 95%CI 035, nega-tive predictive value [NPV] 100%, 95%CI 01.5).German contact investigations assessed 168 childrenwith QFT-GIT and completed approximately 2 yearso ollow-up.43 Three o seven QFT-GIT-positive chil-

    dren developed probable TB (PPV 43%, 95%CI 1675), whereas none o the 161 QFT-GIT-negative chil-dren developed active TB (NPV 100%, 95%CI 03).

    As seen in Table 3, the pooled ORs (random e-ects model) or the association o positive TST usingrespectively 5, 10 or 15 mm cut-os, with dichoto-mous exposure, was 1.34 (95%CI 0.662.72), 1.93(95%CI 0.983.77) and 1.83 (95%CI 0.675.02).For QFT-G/QFT-GIT, the pooled OR was 3.51 (95%CI1.856.66) and or T-SPOT.TB it was 1.31 (95%CI0.762.27; Table 3). When the analysis was restricted

    to low- and middle-income countries (LMICs), QFTand T-SPOT.TB results were positively correlatedwith exposure (OR 1.30 and 2.24, respectively), butTST was not (OR 1.04 [95%CI 0.462.36], 0.81[95%CI 0.381.74] and 0.48 [only one study] or

    cut-os o 5, 10 and 15 mm, respectively).The pooled correlation coefcient between QFT-G/

    QFT-GIT results and TB exposure expressed as agradient (random-eects model) was 0.19 (95%CI0.010.54). The pooled correlation coefcient orT-SPOT.TB was 0.13 (95%CI 0.010.35) and orTST respectively 0.17 (95% CI 0.050.39), 0.18(95% CI 0.080.31) and 0.09 (95%CI 0.110.38)when 5, 10 and 15 mm were used as the cut-o (Ta-ble 4). Findings were similar when the analysis wasrestricted to studies rom LMIC.

    We also estimated slopes o the eect o TB expo-

    sure by regressing the log-OR between each succes-sive higher exposure category. Using either fxed- orrandom-eects models, all slope estimates were simi-lar and associated with wide CIs, such that no testcould be declared superior (Figure 2). Regressionslopes or studies rom LMICs showed a similar trend(Figure 2).

    Sensitivity and specificity in active TB

    The sensitivity and specifcity among all children withconfrmed and/or probable active TB was slightly

    Table 4 Correlation of tests for M. tuberculosis infection with graded exposure to tuberculosis

    TestAuthor, year,

    reference

    Exposure category,n positive/total tested Pooled correlation (95%CI)

    I20least 1 2 3most* Fixed Random

    TST 5 mm Bergamini, 200932 6/99 10/93 16/69 0.18 (0.110.32) 0.17 (0.050.39) 0.67Diel, 200843 7/55 10/35 3/62 13/62Nakaoka, 200673 12/41 23/80 49/78 Petrucci, 200882 3/9 24/36 21/40 35/60Petrucci, 200882 3/10 9/40 17/49 8/14

    TST 10 mm Adetifa, 201026 10/72 34/126 16/27 0.19 (0.130.25) 0.18 (0.080.31) 0.71Bergamini, 200932 2/99 2/93 9/69 Diel, 200843 1/55 3/35 0/16 1/62Nakaoka, 200673 6/41 13/80 38/72

    Okada, 200779

    8/54 14/64 13/49 12/28Petrucci, 200882 3/9 22/36 17/40 33/60Petrucci, 200882 3/10 8/40 16/49 8/14

    TST 15 mm Bergamini, 200932 0/99 1/93 5/69 0.10 (0.010.24) 0.09 (0.110.38) 0.75Diel, 200843 0/55 1/35 0/16 0/62

    QFT-G/QFT-GIT Adetifa, 201026 19/70 42/26 15/27 0.21 (0.160.36) 0.19 (0.010.54) 0.88Bergamini, 200932 0/71 5/47 1/36 Diel, 200843 2/55 3/35 0/16 2/62Girardi, 200749 1/1 4/7 0/1 Nakaoka, 200673 4/39 8/81 53/72 Okada, 200779 3/54 12/64 9/49 9/38Petrucci, 200882 3/9 17/35 19/38 30/59Petrucci, 200882 4/10 10/39 23/49 10/14

    T-SPOT.TB Adetifa, 201026 18/72 40/126 14/27 0.13 (0.030.31) 0.13 (0.010.35) 0.20Bergamini, 200932 1/14 0/56 4/40 Girardi, 200749 1/1 4/7 0/1

    *In four studies there were only three exposure categories. In Bergamini, one un-exposed group had significant prior risk of exposure, and was excluded fromanalysis. Indicates country classified as low-, low and middle- or upper middle-income according to the World Bank.Petrucci 2008 study conducted in two countries. Results shown separately for each country.CI = confidence interval; TST = tuberculin skin test; QFT-G = QuantiFERON-TB Gold; QFT-GIT = QuantiFERON-TB Gold In-Tube.

  • 8/4/2019 Mandalakas_IJTLD_2011

    7/16

    IGRAs and childhood TB: systematic review 7

    higher or both IGRAs than or the TST, but the CIswere overlapping. Forest plots show study-specifcand pooled estimates o sensitivity and specifcity byWorld Bank income index (Figures 3 and 4). Therewas signifcant heterogeneity among studies, suggest-

    ing that pooled results be interpreted cautiously (Ta-ble 5). Stratifed analysis suggested reduced sensitivityo TST, QFT-G/QFT-GIT and T-SPOT.TB in studiesconducted in LMICs or when the average age was50%were BCG-vaccinated. The inclusion o indeterminateresults as alse-negatives had a modest eect on thesensitivity o both IGRAs (Table 6).

    No studies systematically assessed operational as-pects o IGRAs such as cost, reproducibility, trans-port time, time to results or impact on treatmentor easibility. Several studies briey addressed opera-

    tional challenges such as phlebotomy in young chil-dren or the relatively high costs o IGRAs or low-income settings.44,50,73,79,87

    DISCUSSION

    Commercial IGRAs are increasingly used and recom-mended or the diagnosis o M. tuberculosis inec-tion, mainly in high-income countries with a low in-cidence o active TB.9496 The value o IGRAs in TBintermediate and high-burden settings is less clearand depends not only on test perormance but alsoon the availability o preventive TB therapy, healthsystems capacity, laboratory inrastructure and asso-ciated health system costs.

    Figure 2 Regression slopes for exposure gradients. The slopesare estimated from the regression of the logs of ORs of eachsuccessive higher exposure compared to the least exposedgroup. Hence, a steeper slope represents a greater change inthe log ORs as exposure increases. A greater change in the logOR in response to increasing exposure (i.e., steeper slope) sug-gests that a test is better able to detect infection. A randomeffects model was used to calculate regression slopes for TST10 mm and QFT. A fixed effects model was used to calculatethe slopes for TST 5 mm, 15 mm and T-SPOT.TB. HIC = high-

    income country; LMIC = low- and middle-income country;TST = tuberculin skin test; QFT-G = QuantiFERON Gold;QFT-GIT= QuantiFERON Gold In-Tube; OR = odds ratio.

    Figure 3 Pooled sensitivity in active TB, stratified by World Bank income index.

    (continued)

  • 8/4/2019 Mandalakas_IJTLD_2011

    8/16

    8 The International Journal of Tuberculosis and Lung Disease

    Figure 3 (Continued)

    Various systematic reviews on IGRAs have beenpublished, mainly ocusing on adults.97,98 Unique con-siderations in children include their developing im-mune system, high risk o disease progression ollow-ing recent exposure and substantial benefts rompreventive therapy. Current recommendations in high-burden settings are to provide preventive therapyto children based on risk actors i the TST is notavailable.99 An accurate test or the identifcation oM. tuberculosis-inected children could support more

    ocused programmatic delivery o preventive therapyto at-risk children.

    As there is no gold standard or the diagnosis oM. tuberculosis inection, surrogate measures or in-ection are considered reerence standards in diagnos-tic studies. Existing systematic reviews on the value oIGRAs or M. tuberculosis inection have mainly usedactive TB as a surrogate reerence standard.97,98 This ap-proach assumes that estimates o IGRA perormancein populations with active TB reect perormance in

  • 8/4/2019 Mandalakas_IJTLD_2011

    9/16

    IGRAs and childhood TB: systematic review 9

    Figure 4 Pooled specificity in active TB, stratified by World Bank income index.

  • 8/4/2019 Mandalakas_IJTLD_2011

    10/16

    10 The International Journal of Tuberculosis and Lung Disease

    populations with inection. Not only does the im-mune status o diseased persons dier rom that oinected persons, but populations evaluated or active

    TB dier greatly rom populations evaluated or in-ection. For M. tuberculosis inection, this reviewthereore used two reerence standards, incident TBand TB exposure, which were considered superiorsurrogate measures o inection.

    A test identiying individuals at highest risk ordisease progression would greatly aid in the selectiono individuals who would most beneft rom preven-tive therapy. Longitudinal studies using incident TBas the reerence standard are thereore ideal. We iden-tifed two studies completed in high-income countriesthat employed this reerence standard using commer-cial IGRAs. These studies, which included very small

    sample sizes o QFT-positive children, showed a weakassociation between positive QFT assays and sub-sequent active TB, suggesting low PPVs.43,57 Non-commercial ELISpot assays (not included in this re-view) have also shown perormance similar to the TSTin longitudinal studies assessing incident TB.29,100

    Exposure to M. tuberculosis increases the likeli-hood o inection. A strong correlation o the indextest with exposure thereore suggests that the indextest can identiy people most likely to beneft rompreventive therapy. We compared the perormance othe IGRAs and the TST in exposed vs. unexposed in-dividuals as well as across a gradient o exposure andound comparable perormance between all tests us-ing both methodological approaches.

    Our analysis compared dierences in exposure

    gradients (our gold standard comparison) rather thanabsolute exposure. Although studies used dierentselection methods and measure o exposure, we ana-

    lyzed studies collectively by defning study-specifcgradients that progressed rom the least to highestexposure. In theory, a steeper slope in the regressionanalysis is associated with a greater change in oddsacross exposure categories, indicating increased abil-ity o the test to distinguish inection across expo-sure categories. O note, dierences between expo-sure grades were sometimes subtle (e.g., exposure tosmear+ vs. smear++TB); the expected magnitude othe test eect is thereore unclear. Despite the meth-odological heterogeneity o study groups, we ound apositive and increasing correlation between each testand each progressive grade o exposure. This correla-

    tion would be stronger i studies used a standard-ized tool to measure TB exposure. However, despitethe heterogeneous nature o the studies, our analysisclearly illustrates that both the IGRAs and the TSThave the ability to detect M. tuberculosis inection.

    In studies reviewed or dichotomous exposure,the overall odds o a positive QFT-G or QFT-GITwas higher in exposed than unexposed children (OR3.51, 95%CI 1.856.66). QFT may dierentiate be-tween these two groups more accurately than theTST (10 mm cut-o) and T-SPOT.TB (OR 1.93,95%CI 0.983.77 and OR 1.31, 95%CI 0.762.27,respectively). Nevertheless, the wide and overlappingCIs preclude identifcation o a superior test andhighlight the signifcant heterogeneity among studies.Studies were perormed in countries and populations

    Table 5 Characteristic of comparison groups for active TB, sorted by World Bank income index

    Author, year, reference Active TB category* No TB category

    LMIC Dogra, 200744 Definite/probable TB combined Hospitalized children with clinical suspicion of TB or TBcontact, TB disease was ruled out

    Warier, 200992 Definite TB, probable TB Hospital ized chi ldren with other diagnosis, no TB contact

    Hansted, 200951 Definite TB Reported group not usedNicol, 200975 Definite/probable TB combined Children admitted for either clinically suspected TB or TB

    contact, active TB ruled out by CXR and cultureStavri, 200987 Definite TB No group reported

    HIC Bamford, 200930 Definite TB, probable TB No group reportedBergamini, 200932 Definite/probable TB combined Reported group not usedBianchi, 200933 Definite/probable TB combined Reported group not usedChun, 200835 Probable TB TB ruled out, other diagnosisConnell, 200636 Definite/probable TB combined Reported group not usedConnell, 200838 Probable TB Reported group not usedDetjen, 200741 Definite TB Children with other respiratory illness, low risk for TBDominguez, 200845 Definite/probable TB combined Reported group not usedGrare, 201050 Definite/probable TB combined Reported group not usedHaustein, 200952 Definite TB, probable TB Reported group not usedHerrmann, 200953 Definite/probable TB combined Children hospitalized for any other disease, no TB contactHiguchi, 200855 Definite/probable TB combined

    (detected in school outbreak)

    School outbreak investigation, active TB excluded by CXR

    Higuchi, 200958 Definite/probable TB combined Reported group not usedKampmann, 200959 Definite TB, probable TB Children with risk factors for TB but disease ruled out,

    other diagnosis madeLighter, 200967 Definite/probable TB combined Reported group not used

    *Definite TB = culture-confirmed disease, probable TB = diagnosis made on the basis of symptoms and radiologic findings, no culture result. Definite/probablecombined = some cases confirmed but others diagnosed on clinical and radiological criteria only, and results not stratified by method of diagnosis.No TB = active TB systematically excluded in either TB suspects with symptoms suggestive of active TB, or TB contacts. No group reported = study assessedactive TB group only. Reported group not used = reported control group did not meet review criteria for an appropriate control group.TB = tuberculosis; LMIC = low- and middle-income countries; CXR = chest radiography; HIC = high-income countries.

  • 8/4/2019 Mandalakas_IJTLD_2011

    11/16

    Table 6 Diagnostic accuracy of TST, QFT and T-SPOT.TB for definite and/or probable active TB stratified by key variables

    Sensitivity (true-positives) Specificity (false-positives)

    n

    Positive/tested

    n/NSensitivity

    % (95%CI) n

    Positive/testedn/N

    Specificity% (95%CI)

    TSTOverall* 18 373/534 80 (7090) 6 105/362 85 (63100)

    TST, mm

    5 13 220/265 91 (8498) 4 104/217 70 (17100)10 15 248/321 84 (7593) 5 86/276 88 (62100)15 11 246/389 67 (5083) 3 44/131 92 (71100)

    World Bank income indexHIC 14 299/407 81 (7390) 4 94/217 79 (39100)LMIC 4 74/127 70 (4594) 2 11/145 93 (77100)

    TB incidence

  • 8/4/2019 Mandalakas_IJTLD_2011

    12/16

    12 The International Journal of Tuberculosis and Lung Diseasewith dierent annual risk o TB inection and otheractors modiying the risk o inection, such as diag-nostic delay and strain type. The categories unex-posed and exposed thereore describe very dierentpopulations in each study. Sub-group analysis, in-

    cluding World Bank income status, did not suggestpossible explanations or dierences in estimates otest perormance. By comparing studies with similarmethodological approaches and comparable studypopulations, the correlation o either test with expo-sure would be stronger and dierences between testsmight become apparent.

    For the diagnosis o active TB, the overall sensi-tivity o both IGRAs and the TST was similar whenassessed in children with all categories o active TBcombined. Our results suggest that TST and QFTsensitivity may be higher in children with defnite TB

    compared to children with defnite and probable TB.Although estimates o test sensitivity are most certainin the defnite TB group, study-specifc case defni-tions o probable TB varied considerably and mighthave introduced dierential bias among studies. Theassessment o test perormance in children with allcategories o active TB is thus most representativeo clinical practice and provides useul insight uponwhich to base clinical guidelines. Overall, the abilityo either TST or IGRAs was suboptimal to rule in orrule out active TB, reinorcing the appropriate useo these tests as adjuncts in the clinical diagnosis oactive TB.

    Young children have an increased risk o develop-ing active TB ater inection, which may indicate di-erences in the pediatric immune system compared toadults. Individual studies have suggested poorer sen-sitivity and higher rates o indeterminate results inchildren aged 50% BCG coverage.Although BCG vaccination can explain dierences inspecifcity between TST and IGRAs, BCG is not ex-

    pected to inuence sensitivity. We note that BCG cov-erage is high in countries with higher TB burden oramong immigrants in low-burden settings. Popula-tions with higher BCG coverage rates may also havehigher rates o underlying conditions that may impair

    test accuracy, such as co-inections with helminthsand malnutrition.68 Although small subgroup samplesizes limited our ability to demonstrate statisticallysignifcant dierences, trends in sub-group analysishighlight important actors to consider when choos-ing a preerred testing strategy in dierent settings orpatient groups or targeting research areas.

    Operational aspects and easibility are importantconsiderations when contemplating the programmaticimplementation o a diagnostic test, particularly inresource-constrained settings. We aimed to assess someimportant aspects such as cost, transport times, re-

    producibility and easibility in this systematic review,but studies rarely addressed operational aspects ostudy implementation or test perormance. Futurestudies should assess operational aspects and easibil-ity o IGRAs to comprehensively inorm guidelinesregarding their use in children.

    Our review had limitations. In addition to thosepreviously discussed, the studies included assessedvery dierent populations in diverse settings. Samplesizes were less than 10 in some subgroup analyses,limiting our ability to generalize these subgroup re-sults. The tremendous variation in methodologicalapproaches and the use o non-uniorm reerence stan-

    dards led to signifcant analytic heterogeneity. Never-theless, we used a variety o analytic approaches toimprove the robustness o our fndings. A limited num-ber o studies employed an acceptable defnition othe non-diseased group, limiting our power to esti-mate specifcity.

    Although additional data in children exist on in-house assays and pre-commercial versions o currentIGRAs, this review ocused on two standardized com-mercial IGRAs, which are the most widely used andhave recently been considered or policy guidanceand programmatic implementation.

    This review ound that TST, QFT and T-SPOT.TBhad similar accuracy or the diagnosis oM. tubercu-losis inection and active TB in children. Diversestudy methodologies limited the comparability ostudies and interpretation o results, emphasizing theimportance o rigorous, standardized approaches orthe evaluation o TB diagnostic tests in children. Well-defned exposure gradients serve as a good surrogatereerence standard or M. tuberculosis inection andpatient-important outcomes. Based on the evidencegenerated by this review, the WHOs Strategic andTechnical Advisory Group or TB (STAG-TB) has dis-couraged the use o IGRAs or the diagnosis o activeTB and M. tuberculosis inection in children livingin LMICs.101

    There is a clear need or more coordinated re-search on the value o IGRAs in dierent pediatric

  • 8/4/2019 Mandalakas_IJTLD_2011

    13/16

    IGRAs and childhood TB: systematic review 13

    populations in settings with high burden o TB andHIV. Although we were not able to determine a supe-rior test, our analysis confrms that the TST and bothIGRAs are able to identiy children with M. tubercu-losis inection. In settings with high burdens o TB,

    accurate identifcation o children with M. tuberculo-sis inection may help guide the targeted delivery oisoniazid preventive therapy (IPT), and might im-prove the cost-eectiveness o IPT. Children o allages, but particularly very young and HIV-inectedchildren with the highest risk o disease progression,stand to beneft most rom TB preventive therapy. Al-though the limited available data did not support ameaningul comparison o test perormance or theTST and the commercial IGRAs in these subgroupso vulnerable children, more high-quality diagnosticstudies and cost-eectiveness analyses are needed to

    inorm recommendations regarding IGRA use in thesehigh-risk groups.There was sufcient data to assess test peror-

    mance in the diagnosis o active TB. Our analysisdemonstrated that neither the TST nor the IGRAsperorm sufciently to rule in or rule out active TBas a single test. Our fndings reinorce the acceptableuse o these tests as adjuncts in the clinical diagnosiso active TB where resources are sufcient to supportthe use o a test or inection.

    Acknowledgements

    The authors thank the studies authors or kindly providing ad-

    ditional inormation upon request, including I Adetia, S Arend,P Bea, A Bose, T Connell, R Diel, G Dixon, J Dominguez,

    M Grare, K Higuchi, Y Kang, Y Kobashi, B Kampman, Y Lee, W

    Lew, M Losi, M Lucas, P Mantegani, B McKinnon, A Nienhaus,

    R Petrucci, H Pollack and E Tavast. The authors thank M P

    McGraw, Case Western Reserve University, or support to com-

    plete their electronic literature search and K Steingart or helpul

    discussions on QUADAS. This systematic review was made possi-

    ble by the generous support o the American people through the

    United States Agency or International Development (USAID) un-

    der Cooperative Agreement GHN-A-00-08-00004-00. The con-

    tents are the responsibility o the authors and do not necessarily

    reect the views o USAID or the United States Government. DM

    and AB receive salary support rom Fond de la recherche en Sant

    de Qubec. AM receives salary support rom the US Department

    o State Fulbright Senior Scholars program. The unding agencieshad no role in the preparation or submission o this report.

    References

    1 Dolin P J, Raviglione M C, Kochi A. Global tuberculosis inci-

    dence and mortality during 19902000. Bull World Health

    Organ 1994; 72: 213220.

    2 Corbett E L, Watt C J, Walker N, et al. The growing burden o

    tuberculosis: global trends and interactions with the HIV epi-

    demic. Arch Intern Med 2003; 163: 10091021.

    3 Marais B J, Gie R P, Schaa H S, et al. The natural history o

    childhood intra-thoracic tuberculosis: a critical review o liter-

    ature rom the pre-chemotherapy era. Int J Tuberc Lung Dis

    2004; 8: 392402.

    4 Hesseling A C, Cotton M F, Jennings T, et al. High incidence otuberculosis among HIV-inected inants: evidence rom a South

    Arican population-based study highlights the need or improved

    tuberculosis control strategies. Clin Inect Dis 2009; 48: 108

    114.

    5 Anastos K, Kalish L A, Palacio H, et al. Prevalence o and risk

    actors or tuberculin positivity and skin test anergy in HIV-1-

    inected and uninected at-risk women. Womens Interagency

    HIV Study (WIHS). J Acquir Immune Defc Syndr 1999; 21:

    141147.

    6 Anonymous. CDC changing position on anergy testing, ther-

    apy. Centers or Disease Control and Prevention. AIDS Alert1997; 12: 117118.

    7 Graham N M, Nelson K E, Solomon L, et al. Prevalence o tu-

    berculin positivity and skin test anergy in HIV-1-seropositive

    and -seronegative intravenous drug users [Comment]. JAMA

    1992; 267: 369373.

    8 Huebner R E, Schein M F, Hall C A, Barnes S A. Delayed-type

    hypersensitivity anergy in human immunodefciency virus-

    inected persons screened or inection with Mycobacteriumtuberculosis. Clin Inect Dis 1994; 19: 2632.

    9 Karalliedde S, Katugaha L P, Uragoda C G. Tuberculin re-

    sponse o Sri Lankan children ater BCG vaccination at birth.

    Tubercle 1987; 68: 3338.

    10 Klein R S, Flanigan T, Schuman P, Smith D, Vlahov D. Criteria

    or assessing cutaneous anergy in women with or at risk or

    HIV inection. HIV Epidemiologic Research Study Group

    [Comment]. J Allergy Clin Immunol 1999; 103: 9398.

    11 Lavin J, Haidorer C. Anergy testinga vital weapon. RN

    1993; 56: 3132; quiz 3.

    12 Miller W C, Thielman N M, Swai N, et al. Delayed-type hyper-

    sensitivity testing in Tanzanian adults with HIV inection. J

    Acquir Immune Defc Syndr Hum Retrovirol 1996; 12: 303

    308.

    13 Pesanti E L. The negative tuberculin test. Tuberculin, HIV and

    anergy panels. Am J Respir Crit Care Med 1994; 149: 1699

    1709.

    14 Wright P W, Crutcher J E, Holiday D B. Selection o skin test

    antigens to evaluate PPD anergy. J Fam Pract 1995; 41: 5964.

    15 Hesseling A C, Schaa H S, Gie R P, Starke J R, Beyers N. A

    critical review o diagnostic approaches used in the diagnosiso childhood tuberculosis. Int J Tuberc Lung Dis 2002; 6:

    10381045.

    16 Hesseling A C, Mandalakas A M, Kirchner L H, et al. Highly

    discordant T-cell responses in individuals with recent house-

    hold tuberculosis exposure. Thorax 2008; 64: 840846.

    17 Lienhardt C, Sillah J, Fielding K, et al. Risk actors or tubercu-

    losis inection in children in contact with inectious tuberculo-

    sis cases in the Gambia, West Arica. Pediatrics 2003; 111:

    e608614.

    18 Whiting P, Rutjes A W, Reitsma J B, Bossuyt P M, Kleijnen J.

    The development o QUADAS: a tool or the quality assess-

    ment o studies o diagnostic accuracy included in systematic

    reviews. BMC Med Res Methodol 2003; 3: 25.

    19 Reitsma J B, Rutjes A W S, Whiting P, Vlassow V V, Leeang

    M M G, Deeks J J. Chapter 9: Assessing methodological qua-lity. In: Deeks J J, Bossuyt P M, Gatsonis C, eds. Cochrane hand-

    book or systematic reviews o diagnostic test accuracy version

    1.0.0. London, UK: Cochrane Collaboration, 2009.

    20 Higgins J P, Thompson S G. Quantiying heterogeneity in a

    meta-analysis. Stat Med 2002; 21: 15391558.

    21 Diener M J, Hilsenroth M J, Weinberger J. A primer on meta-

    analysis o correlation coefcients: the relationship between

    patient-reported therapeutic alliance and adult attachment style

    as an illustration. Psychother Res 2009; 19: 519526.

    22 Schlattmann P. Medical applications o fnite mixture models,

    statistics or biology and health. Chapter 7: Investigating and

    analyzing heterogeneity in meta-analysis. Berlin, Heidelberg:

    Springer Verlag, 2009.

    23 Greenland S, Longnecker M P. Methods or trend estimation

    rom summarized dose-response data, with applications tometa-analysis. Am J Epidemiol 1992; 135: 13011309.

    24 Hamza T H, van Houwelingen H C, Stijnen T. The binomial

    distribution o meta-analysis was preerred to model within-

    study variability. J Clin Epidemiol 2008; 61: 4151.

  • 8/4/2019 Mandalakas_IJTLD_2011

    14/16

    14 The International Journal of Tuberculosis and Lung Disease25 Higgins J P, Thompson S G. Controlling the risk o spurious

    fndings rom meta-regression. Stat Med 2004; 23: 16631682.

    26 Adetia I M, Ota M O, Jeries D J, et al. Commercial inter-

    eron gamma release assays compared to the tuberculin skin

    test or diagnosis o latent Mycobacterium tuberculosis inec-tion in childhood contacts in the Gambia. Pediatr Inect Dis J

    2010; 29: 439443.27 Adetia I M O, Lugos M D, Hammond A, et al. Comparison o

    two intereron gamma release assays in the diagnosis oMyco-bacterium tuberculosis inection and disease in The Gambia.BMC Inect Dis 2007; 7: 122.

    28 Arend S M, Thijsen S F, Leyten E M, et al. Comparison o two

    intereron-gamma assays and tuberculin skin test or tracing

    tuberculosis contacts. Am J Respir Crit Care Med 2007; 175:

    618627.

    29 Bakir M, Millington K A, Soysal A, et al. Prognostic value o a

    T-cell-based, intereron-gamma biomarker in children with tu-

    berculosis contact. Ann Intern Med 2008; 149: 777787.

    30 Bamord A R, Crook A M, Clark J, et al. Comparison o

    intereron-gamma release assays and tuberculin skin test in pre-

    dicting active tuberculosis (TB) in children in the UKa paedi-

    atric TB network study. Arch Dis Child 2010; 95: 180186.31 Bea P, Zellweger A, Janssens J P, Wrighton-Smith P, Zellweger

    J P. Indeterminate test results o T-SPOTTM.TB perormed un-der routine feld conditions. Eur Respir J 2008; 31: 842846.

    32 Bergamini B M, Losi M, Vaienti F, et al. Perormance o commer-

    cial blood tests or the diagnosis o latent tuberculosis inection

    in children and adolescents. Pediatrics 2009; 123: e419e424.

    33 Bianchi L, Galli L, Moriondo M, et al. Intereron-gamma re-

    lease assay improves the diagnosis o tuberculosis in children.

    Pediatr Inect Dis J 2009; 28: 510514.

    34 Bruzzese E, Bocchino M, Assante L R, et al. Gamma intereron

    release assays or diagnosis o tuberculosis inection in immune-

    compromised children in a country in which the prevalence o

    tuberculosis is low. J Clin Microbiol 2009; 47: 23552357.

    35 Chun J K, Kim C K, Kim H S, et al. The role o a whole blood

    intereron-gamma assay or the detection o latent tuberculosis

    inection in bacille Calmette-Guerin-vaccinated children. Diagn

    Microbiol Inect Dis 2008; 62: 389394.

    36 Connell T G, Curtis N, Ranganathan S C, Buttery J P. Peror-

    mance o a whole blood intereron gamma assay or detecting

    latent inection with Mycobacterium tuberculosis in children.Thorax 2006; 61: 616620.

    37 Connell T G, Davies M A, Johannisen C, et al. Reversion and

    conversion oMycobacterium tuberculosis IFN-gamma ELISpotresults during anti-tuberculous treatment in HIV-inected chil-

    dren. BMC Inect Dis 2010; 10: 138.

    38 Connell T G, Ritz N, Paxton G A, Buttery J P, Curtis N, Ranga-

    nathan S C. A three-way comparison o tuberculin skin testing,

    QuantiFERON-TB Gold and T-SPOT.TB in children. PLoS One

    2008; 3: e2624.39 Connell T G, Tebruegge M, Ritz N, Bryant P A, Leslie D, Cur-

    tis N. Indeterminate intereron-gamma release assay results in

    children. Pediatr Inect Dis J 2010; 29: 285286.

    40 Davies M A, Connell T, Johannisen C, et al. Detection o tu-

    berculosis in HIV-inected children using an enzyme-linked

    immunospot assay. AIDS 2009; 23: 961969.

    41 Detjen A K, Keil T, Roll S, et al. Intereron-gamma release as-

    says improve the diagnosis o tuberculosis and nontuberculous

    mycobacterial disease in children in a country with a low inci-

    dence o tuberculosis. Clin Inect Dis 2007; 45: 322328.

    42 Dewan P K, Grinsdale J, Kawamura L M. Low sensitivity o a

    whole-blood intereron-gamma release assay or detection o

    active tuberculosis. Clin Inect Dis 2007; 44: 6973.

    43 Diel R, Loddenkemper R, Meywald-Walter K, Niemann S,

    Nienhaus A. Predictive value o a whole blood IFN-gamma as-say or the development o active tuberculosis disease ater re-

    cent inection with Mycobacterium tuberculosis. Am J RespirCrit Care Med 2008; 177: 11641170.

    44 Dogra S, Narang P, Mendiratta D K, et al. Comparison o a

    whole blood intereron-gamma assay with tuberculin skin test-

    ing or the detection o tuberculosis inection in hospitalized

    children in rural India. J Inect 2007; 54: 267276.

    45 Dominguez J, Ruiz-Manzano J, De Souza-Galvao M, et al.

    Comparison o two commercially available gamma intereron

    blood tests or immunodiagnosis o tuberculosis. Clin VaccineImmunol 2008; 15: 168171.

    46 Eisenhut M, Paranjothy S, Abubakar I, et al. BCG vaccination

    reduces risk o inection with Mycobacterium tuberculosis asdetected by gamma intereron release assay. Vaccine 2009; 27:

    61166120.

    47 Ferrara G, Losi M, DAmico R, et al. Use in routine clinical

    practice o two commercial blood tests or diagnosis o inec-

    tion with Mycobacterium tuberculosis: a prospective study.Lancet 2006; 367: 13281334.

    48 Ferrara G, Losi M, Meacci M, et al. Routine hospital use o a

    new commercial whole blood intereron-gamma assay or the

    diagnosis o tuberculosis inection. Am J Respir Crit Care Med

    2005; 172: 631635.

    49 Girardi E, Loredo M, Alessandrini A, Anzidei G, Goletti D. A

    two-step approach or screening contacts o active tuberculo-sis. Inection 2007; 35: 122123.

    50 Grare M, Derelle J, Dailloux M, Laurain C. QuantiFERON-

    TB Gold In-Tube as help or the diagnosis o tuberculosis in a

    French pediatric hospital. Diagn Microbiol Inect Dis 2010;

    66: 366372.

    51 Hansted E, Andriuskeviciene A, Sakalauskas R, Kevalas R, Sit-

    kauskiene B. T-cell-based diagnosis o tuberculosis inection in

    children in Lithuania: a country o high incidence despite a

    high coverage with bacille Calmette-Guerin vaccination. BMC

    Pulm Med 2009; 9: 41.

    52 Haustein T, Ridout D A, Hartley J C, et al. The likelihood o an

    indeterminate test result rom a whole-blood intereron-gamma

    release assay or the diagnosis oMycobacterium tuberculosisinection in children correlates with age and immune status.

    Pediatr Inect Dis J 2009; 28: 669673.

    53 Herrmann J L, Belloy M, Porcher R, et al. Temporal dynamics

    o intereron gamma responses in children evaluated or tuber-

    culosis. PLoS One 2009; 4: e4130.

    54 Hesseling A C, Mandalakas A M, Kirchner H L, et al. Highly

    discordant T-cell responses in individuals with recent exposure

    to household tuberculosis. Thorax 2009; 64: 840846.

    55 Higuchi K, Harada N, Fukazawa K, Mori T. Relationship be-

    tween whole-blood intereron-gamma responses and the risk

    o active tuberculosis. Tuberculosis (Edinb) 2008; 88: 244248.

    56 Higuchi K, Harada N, Mori T, Sekiya Y. Use o QuantiFERON-

    TB Gold to investigate tuberculosis contacts in a high school.

    Respirology 2007; 12: 8892.

    57 Higuchi K, Kondo S, Wada M, et al. Contact investigation in a

    primary school using a whole blood intereron-gamma assay.J Inect 2009; 58: 352357.

    58 Higuchi R, Mori M, Ozawa R, et al. Whole blood intereron-

    gamma assay or tuberculosis in children in Japan. Pediatr

    Intern 2009; 51: 97102.

    59 Kampmann B, Whittaker E, Williams A, et al. Intereron-gamma

    release assays do not identiy more children with active tuber-

    culosis than the tuberculin skin test. Eur Respir J 2009; 33:

    13741382.

    60 Kang Y A, Lee H W, Hwang S S, et al. Useulness o whole-

    blood intereron-gamma assay and intereron-gamma enzyme-

    linked immunospot assay in the diagnosis o active pulmonary

    tuberculosis. Chest 2007; 132: 959965.

    61 Kobashi Y, Mouri K, Miyashita N, et al. QuantiFERON TB-

    2G test or patients with active tuberculosis stratifed by age

    groups. Scan J Inect Dis 2009; 41: 841846.62 Kobashi Y, Mouri K, Obase Y, Fukuda M, Miyashita N, Oka

    M. Clinical evaluation o QuantiFERON TB-2G test or

    immunocompromised patients. Eur Respir J 2007; 30: 945950.

  • 8/4/2019 Mandalakas_IJTLD_2011

    15/16

    IGRAs and childhood TB: systematic review 15

    63 Latorre I, De Souza-Galvao M, Ruiz-Manzano J, et al. Quanti-

    tative evaluation o T-cell response ater specifc antigen stimu-

    lation in active and latent tuberculosis inection in adults and

    children. Diagn Microbiol Inect Dis 2009; 65: 236246.

    64 Lee J Y, Choi H J, Park I N, et al. Comparison o two com-

    mercial intereron-gamma assays or diagnosing Mycobacte-

    rium tuberculosis inection. Eur Respir J 2006; 28: 2430.65 Lew W J, Jung Y J, Song J W, et al. Combined use o Quanti-

    FERON-TB Gold assay and chest computed tomography in a

    tuberculosis outbreak. Int J Tuberc Lung Dis 2009; 13: 633

    639.

    66 Liebeschuetz S, Bamber S, Ewer K, Deeks J, Pathan A A, Lal-

    vani A. Diagnosis o tuberculosis in South Arican children with

    a T-cell-based assay: a prospective cohort study. Lancet 2004;

    364: 21962203.

    67 Lighter J, Rigaud M, Eduardo R, Peng C H, Pollack H. Latent

    tuberculosis diagnosis in children by using the QuantiFERON-

    TB Gold In-Tube Test. Pediatrics 2009; 123: 3037.

    68 Lucas M, Nicol P, McKinnon E, et al. A prospective large-scale

    study o methods or the detection o latent Mycobacteriumtuberculosis inection in reugee children. Thorax 2010; 65:

    442448.69 Mandalakas A M, Hesseling A C, Chegou N N, et al. High

    level o discordant IGRA results in HIV-inected adults and

    children. Int J Tuberc Lung Dis 2008; 12: 417423.

    70 Mantegani P, Piana F, Codecasa L, et al. Comparison o an in-

    house and a commercial RD1-based ELISpot-IFN-gamma as-

    say or the diagnosis oMycobacterium tuberculosis inection.Clin Med Res 2006; 4: 266272.

    71 Moilicotti P, Bua A, Mela G, et al. Perormance o Quanti-

    FERON-TB testing in a tuberculosis outbreak at a primary

    school. J Pediatrics 2008; 152: 585586.

    72 Moyo S, Isaacs F, Gelderbloem S, et al. Tuberculin skin test and

    QuantiFERON assay in young children investigated or tuber-

    culosis in South Arica. Int J Tuberc Lung Dis. [In press]

    73 Nakaoka H, Lawson L, Squire S B, et al. Risk or tuberculosis

    among children. Emerg Inect Dis 2006; 12: 13831388.

    74 Neira-Munoz E, Smith J, Cockcrot P, Basher D, Abubaker I.

    Extensive transmission oMycobacterium tuberculosis amongchildren on a school bus. Pediatr Inect Dis J 2008; 27: 836

    837.

    75 Nicol M P, Davies M A, Wood K, et al. Comparison o

    T-SPOT.TB assay and tuberculin skin test or the evaluation oyoung children at high risk or tuberculosis in a community

    setting. Pediatrics 2009; 123: 3843.

    76 Nicol M P, Pienaar D, Wood K, et al. Enzyme-linked immuno-

    spot assay responses to early secretory antigenic target 6, cul-

    ture fltrate protein 10, and purifed protein derivative among

    children with tuberculosis: implications or diagnosis and moni-

    toring o therapy. Clin Inect Dis 2005; 40: 13011308.

    77 Nienhaus A, Schablon A, Diel R. Intereron-gamma release assayor the diagnosis o latent TB inectionanalysis o discordant

    results, when compared to the tuberculin skin test. PLoS One

    2008; 3: e2665.

    78 Nsutebu E, Moftt S J, Mullarkey C, Schweiger M S, Collyns

    T, Watson J P. Use o QuantiFERON-TB Gold test in the inves-

    tigation o unexplained positive tuberculin skin tests. Public

    Health 2008; 122: 12841287.

    79 Okada K, Mao T E, Mori T, et al. Perormance o an intereron-

    gamma release assay or diagnosing latent tuberculosis inec-

    tion in children. Epidemiol Inect 2008; 136: 11791187.

    80 Ozekinci T, Ozbek E, Celik Y. Comparison o tuberculin skin

    test and a specifc T-cell-based test, T-Spot.TB, or the diag-nosis o latent tuberculosis inection. J Intern Med Res 2007;

    35: 696703.

    81 Pai M, Joshi R, Dogra S, et al. T-cell assay conversions and re-versions among household contacts o tuberculosis patients in

    rural India. Int J Tuberc Lung Dis 2009; 13: 8492.

    82 Petrucci R, Amer N A, Gurgel R Q, et al. Intereron gamma,

    intereron-gamma-induced-protein 10, and tuberculin re-

    sponses o children at high risk o tuberculosis inection.

    Pediatr Inect Dis J 2008; 27: 10731077.

    83 Piana F, Ruo Codecasa L, Baldan R, Miotto P, Ferrarese M,

    Cirillo D M. Use o T-SPOT.TB in latent tuberculosis inec-tion diagnosis in general and immunosuppressed populations.

    New Microbiol 2007; 30: 286290.84 Ruhwald M, Petersen J, Kooed K, et al. Improving T-cell as-

    says or the diagnosis o latent TB inection: potential o a

    diagnostic test based on IP-10. PLoS One 2008; 3: e2858.

    85 Soysal A, Millington K A, Bakir M, et al. Eect o BCG vac-

    cination on risk oMycobacterium tuberculosis inection inchildren with household tuberculosis contact: a prospective

    community-based study. Lancet 2005; 366: 14431451.

    86 Soysal A, Turel O, Toprak D, Bakir M. Comparison o posi-

    tive tuberculin skin test with an intereron-gamma-based as-

    say in unexposed children. Jpn J Inect Dis 2008; 61: 192

    195.

    87 Stavri H, Ene L, Popa G L, et al. Comparison o tuberculin

    skin test with a whole-blood intereron gamma assay and

    ELISA, in HIV-positive children and adolescents with TB.

    Roum Arch Microbiol Immunol 2009; 68: 1419.88 Stean D C, Dippenaar A, Detjen A K, et al. Intereron-gamma

    release assays or the detection oMycobacterium tuberculo-sis inection in children with cancer. Int J Tuberc Lung Dis2010; 14: 689694.

    89 Taylor R E B, Cant A J, Clark J E. Potential eect o NICE

    tuberculosis guidelines on paediatric tuberculosis screening.

    Arch Dis Child 2008; 93: 200203.

    90 Tsiouris S J, Austin J, Toro P, et al. Results o a tuberculosis-

    specifc IFN-gamma assay in children at high risk or tuber-

    culosis inection. Int J Tuberc Lung Dis 2006; 10: 939941.

    91 Wang J Y, Chou C H, Lee L N, et al. Diagnosis o tuberculosis

    by an enzyme-linked immunospot assay or intereron-gamma.

    Emerg Inect Dis 2007; 13: 553558.

    92 Warier A, Gunawathi S, Venkatesh S, John K R, Bose A. T-cell

    assay as a diagnostic tool or tuberculosis. Indian Pediatr 2009;

    47: 9092.

    93 Winje B A, Otung F, Korsvold G E, et al. School-based

    screening or tuberculosis inection in Norway: comparison

    o positive tuberculin skin test with intereron-gamma release

    assay. BMC Inect Dis 2008; 8: 140.

    94 Diel R, Forssbohm M, Loytved G, et al. [Recommendations

    or background studies in tuberculosis]. Pneumologie 2007;

    61: 440455. [French]

    95 Mazurek M, Jereb J, Vernon A, LoBue P, Goldberg S, Castro

    K. Updated guidelines or using intereron gamma release as-

    says to detect Mycobacterium tuberculosis inectionUnitedStates, 2010. MMWR Recomm Rep 2010; 59: 125.

    96 Updated recommendations on intereron gamma release as-

    says or latent tuberculosis inection. An Advisory CommitteeStatement (ACS). Can Commun Dis Rep 2008; 34: 113.

    97 Pai M, Zwerling A, Menzies D. Systematic review: T-cell-

    based assays or the diagnosis o latent tuberculosis inection:

    an update. Ann Intern Med 2008; 149: 177184.

    98 Menzies D, Pai M, Comstock G. Meta-analysis: new tests or

    the diagnosis o latent tuberculosis inection: areas o uncer-

    tainty and recommendations or research. Ann Intern Med

    2007; 146: 340354.

    99 World Health Organization. Guidance or national tuber-

    culosis programmes on the management o tuberculosis in

    children. WHO/HTM/TB/2006.371. WHO/FCH/CAH/2006.7.

    Geneva, Switzerland: WHO, 2006.

    100 Hill P C, Jackson-Sillah D J, Fox A, et al. Incidence o tuber-

    culosis and the predictive value o ELISPOT and Mantoux

    tests in Gambian case contacts. PloS ONE 2008; 3: e1379.101 World Health Organization. Report o the Tenth Meeting o

    the Strategic and Technical Advisory Group or Tuberculosis

    (STAG-TB). Geneva, Switzerland: WHO, 2010.

  • 8/4/2019 Mandalakas_IJTLD_2011

    16/16

    IGRAs and childhood TB: systematic review i

    C O NT E XT E : Les enants inects par Mycobacterium

    tuberculosis encourent un risque signifcati de dve-

    lopper une tuberculose (TB) et une thrapie prventive

    leur est bnfque.

    O B JE C T I VE : Evaluer la valeur des tests de libration de

    linterron-gamma (IGRA) et des tests cutans tubercu-

    liniques (TST) pour le diagnostic de linection et de la

    maladie TB chez les enants.

    M T HO DE S : On a inclus 33 tudes valuant les IGRA

    commerciaux (QuantiFERON-TB [QFT] et T-SPOT.TB)

    ainsi que le TST. Les standards de rrence de linection

    ont t la TB incidente ou lexposition la TB. On a

    valu la perormance des tests pour le diagnostic de la

    maladie dans les tudes valuant des enants ayant une

    TB confrme et/ou diagnostique cliniquement par com-

    paraison aux enants o la TB a t exclue.

    RSU LTATS : Deux petites tudes ont mesur la TB inci-dente chez les enants au moyen de QFT et ont trouv

    une aible valeur prdictive positive. Lassociation de la

    rponse au test avec lexposition (caractrise soit de

    manire dichotomique, soit sous orme dun gradient) a

    t similaire pour tous les tests. Pour le diagnostic de la

    maladie, la sensibilit et la spcifcit de lensemble des

    tests ont t semblables. Une analyse stratife a suggr

    une sensibilit plus aible de lensemble des tests chez les

    jeunes enants inects par le virus de limmunodfcience

    humaine.

    C O NC L US I O NS : Les donnes disponibles suggrent que

    le TST et les IGRA ont une prcision similaire pour la

    dtection de linection TB ou pour le diagnostic de la

    maladie chez les enants. Lhtrognit des mthodolo-

    gies limite le caractre comparable des tudes et linter-

    prtation des rsultats. Une approche rigoureuse et stan-

    dardise de lvaluation des tests diagnostiques de la TB

    simpose chez les enants.

    MAR C O DE R E F E R E NC I A : Los nios inectados por

    Mycobacterium tuberculosis presentan un riesgo im-

    portante de padecer tuberculosis (TB) y se benefcian

    con el tratamiento preventivo.

    O B JE T I VO : Evaluar la utilidad de las pruebas de libera-

    cin de interern gama (IGRA) y de la reaccin cutnea

    a la tuberculina (TST) en el diagnstico de la ineccin

    y de la enermedad tuberculosa en los nios.M T O DO S : Se incluyeron en el anlisis 33 estudios de

    evaluacin de las IGRA (QuantiFERON-TB [QFT] y

    la prueba T-SPOT.TB) y de la reaccin TST. Se usaron

    como criterios de reerencia de ineccin un caso nuevo

    de TB o la exposicin a la TB. El rendimiento diagns-

    tico de las pruebas se evalu en los estudios que incluye-

    ron nios con TB clnica o confrmada, comparados con

    nios en quienes se haba descartado el diagnstico de

    enermedad tuberculosa.

    RESULTADOS: En dos estudios pequeos se midieron

    los casos nuevos de TB en nios en quienes se practic

    una prueba QFT y se encontr un bajo valor pronstico

    del resultado positivo. La asociacin entre la respuesta a

    la prueba y la exposicin (por categoras dicotmicas o

    como un gradiente) ue equivalente con todas las prue-

    bas. La sensibilidad y la especifcidad con respecto al

    diagnstico de la enermedad ueron anlogas con todas

    las pruebas. El anlisis estratifcado indic una menor sen-sibilidad en los nios de