Secondary Logo

Journal Logo

Translational Medicine Reports

Non-IFNγ Whole Blood Cytokine Responses to Mycobacterium tuberculosis Antigens in HIV-exposed Infants

Anterasian, Christine MD*; Warr, Alex J. MD; Lacourse, Sylvia M. MD, MPH†,‡; Kinuthia, John MBChB, MMed, MPH§; Richardson, Barbra A. PhD, MS‡,¶; Nguyen, Felicia K. BS; Matemo, Daniel HND, MLS§; Maleche-Obimbo, Elizabeth MBChB, MPH; John Stewart, Grace C. MD, PhD, MPH; Hawn, Thomas R. MD, PhD

Author Information
The Pediatric Infectious Disease Journal: October 2021 - Volume 40 - Issue 10 - p 922-929
doi: 10.1097/INF.0000000000003254
  • Free


Tuberculosis (TB) is a leading cause of global childhood morbidity and mortality especially among infants <1 year of age.1–3 The diagnosis of pediatric TB disease remains a challenge due to nonspecific clinical symptoms, specimen collection difficulties and lack of a standard test. Tuberculin skin test (TST) is standard in evaluating for pediatric TB, though interpretation is complicated by cross-reactivity with environmental mycobacteria and Bacillus Calmette–Guérin (BCG) vaccine. Furthermore, detection of Mycobacterium tuberculosis (Mtb) specific responses to ESAT6/CFP10 antigens is hindered by reduced sensitivity of interferon-gamma (IFNγ) release assays (IGRA) in children, especially those <2 years old.4,5

HIV-exposed uninfected (HEU) infants have increased infectious morbidity and mortality compared with HIV-unexposed infants, including increased risk of TB disease.6,7 HEU infants have altered immune responses to many pathogens, increased T-cell activation and memory phenotypes, and decreased T-cell functional responses to BCG vaccination.8–10 Progression from Mtb infection to TB disease can be rapid with approximately 20%–40% of HEU infants having active TB disease at 1 year of age in TB-endemic settings6,7,11–14 compared with <10% in otherwise healthy BCG-vaccinated infants.15

Isoniazid (INH) preventive therapy (IPT) is routinely used to prevent TB in people living with HIV (PLHIV) >1 year old16 but limited data exists on whether IPT protects from primary Mtb infection in high-risk HEU infants living where widespread antiretroviral therapy and IPT for PLHIV is common.7,17 The Infant TB Infection Prevention Study (iTIPS) was a randomized nonblinded trial in western Kenya, an area of high HIV and TB burden.17,18 HEU infants were randomized to receive IPT or no IPT for the first year of life with cumulative incidence of Mtb infection measured at 12 months postrandomization using TST and Quantiferon-Gold Plus (QFT-Plus). IPT was not associated with a significant difference in risk of Mtb infection (hazard ratio = 0.53, P = 0.11), though there was a trend toward lower risk of positive TST among HEU infants randomized to IPT (hazard ratio = 0.46, P = 0.07). No similar trend was detected for QFT-Plus, though analysis was limited due to low QFT-Plus positivity (3/300 QFT-Plus+) resulting in an underpowered study.

Due to TST cross-reactivity with non-TB mycobacteria (NTM) as well as BCG, this trend of lower risk of positive TST in those receiving IPT could be due to lower Mtb infection incidence, lower NTM infection incidence, or a lower response to BCG in the IPT arm. To address this question, we examined whether IPT is associated with non-IFNγ Mtb antigen-specific ESAT6/CFP10 cytokine responses. We measured IL2, IP10, TNF, IL5 and IL15 in QFT-Plus supernatants for broader assessment of infant-acquired immune responses. These cytokines show promise in the diagnosis of Mtb infection,19–21 including in young children22–25 and PLHIV.25–28 In particular, IP10 is a chemokine downstream of IFNγ29 with similar sensitivity and specificity following stimulation of whole-blood with Mtb-specific antigens, can be measured in urine (less specific),30 and is less affected by young age and low CD4 count in PLHIV.31–33 However, Mtb-antigen specific non-IFNγ cytokine outcomes have never been assessed in HEU infants (and only once in an infant population34), in the context of a pediatric TB clinical trial, or as a combined endpoint.

We hypothesized that measuring multiple non-IFNγ cytokines improves the sensitivity of Mtb infection detection in children while retaining the specificity associated with ESAT6/CFP10 antigen-specific immune responses.


We enrolled HIV-infected mothers and their infants in a nonblinded randomized nonblinded trial ( NCT02613169) of IPT to prevent Mtb infection in Kisumu, Kenya (N = 300, 2016–2019).17,18 Eligible infants were 6–10 weeks old, birth weight >2.5 kg, and ≥ 37 weeks of gestation. We excluded infants at enrollment with known TB exposure (including maternal TB diagnosed in the past year) or enrollment in other TB prevention studies. Infants (including HEU infants) routinely receive BCG vaccination at birth in Kenya. Further study information has been previously published.17,18 Infants were randomized 1:1 to 12 months of INH (10 mg/kg dose) with pyridoxine or no INH. Primary endpoint was Mtb infection 12 months after randomization (approximately 14 months old) assessed by QuantiFERON-TB Gold Plus (QFT-Plus) and/or TST. Infants were considered positive for Mtb infection using standard QFT-Plus interpretations from the manufacturer or if TST ≥10 mm induration in concordance with the primary iTIPS clinical trial and pediatric care guidelines. Frozen QFT-Plus supernatants were shipped to Seattle, WA. Multiplexed microbead assay was performed in 2 batches to detect IL-15, IL-5, IL-2, IP-10 and TNF on nil, mitogen, and TB1 tubes with ESAT6/CFP10 antigens. TB1 tubes were selected to assess for Mtb infection by measuring CD4 responses to Mtb-specific antigens. Samples were assayed in duplicate and run using standard protocol.

Samples and cytokine standards were incubated with Luminex microbeads (one unique bead population per cytokine) coated with cytokine-specific antibodies. Beads were washed, incubated with biotinylated cytokine antibodies, washed again, and incubated with phycoerythrin-streptavidin conjugate. After final wash, the assay was read on a Luminex 200 instrument (Luminex Corporation, Austin, TX), classifying each bead by cytokine-specificity and phycoerythrin fluorescence intensity. Phycoerythrin fluorescence of each bead is proportional to cytokine concentration in the samples or standards. A standard curve was generated for each cytokine with sample concentration calculated from these curves.

Defining Positivity of Non-IFNγ Cytokines

Non-IFNγ ESAT6/CFP10 cytokine responses were considered positive if all of the following criteria were met: (1) TB1-nil >90th percentile for cytokine per batch and TB1-nil ≥25% of sample nil; (2) Nil <2× average nil for cytokine per batch (negative control); and (3) Mitogen ≥4.8 pg/mL (Luminex lower limit of detection) (positive control). Non-IFNγ ESAT6/CFP10 cytokine responses were considered negative if the patient did not meet the first criterion above. Otherwise, a sample was considered indeterminate.

Statistical Analysis

Statistical analysis was performed in R version 4.0. Logistic regression was used to assess the effect of IPT on Mtb infection outcomes in HEU infants over the first year of life using an endpoint of any positive IL2, IP10, TNF, IFNγ response to ESAT6/CFP10 (measured by Luminex or QFT-Plus) to define infection ± TST result. Due to small numbers of positive infants when IL2, IP10 and TNF outcomes were analyzed independently, Fisher’s exact tests were used to assess the effect of IPT on Mtb infection outcomes. Kendall rank correlation coefficient (t) was used to assess relationships between non-IFNγ cytokine responses, QFT-Plus responses, TST size and demographic variables.


Among 300 infants, 267 infants had either a QFT-Plus (N = 251) or TST (N = 188) and 229 of 267 infants had IL2, IP10, TNF, IL5 and IL15 measured in supernatants (Table 1). These infants were a median age 6.1 weeks of age at enrollment, median birth weight 3.5 kg, received antiretrovirals (ARV) for prevention of mother to child HIV transmission (99%), were BCG vaccinated (93% with BCG vaccination scar), and were born to mothers of whom 99% received ARVs in pregnancy, 76% had undetectable viral load in pregnancy with median CD4 count 468, and 75% had a history of IPT. 8 infants had a maternal reported of Mtb exposure during the study period, 3 within the household and 5 outside the household. Mtb infection was detected in 3 patients (3/251) by QFT-Plus (0 with reported Mtb exposure) and 25 patients (25/188) by TST (2 with reported Mtb exposure). IL2 responses had low background, strong mitogen responses, and a bimodal distribution for TB1-nil with 7 patients having TB1-nil >0 (TB1-nil range 0–929 pg/mL) (Table 2 and Fig. 1A). IP10 levels had high background (median nil 553 pg/mL) with a small margin between TB1 and nil levels (median TB1-nil 8.9 pg/mL), strong mitogen response, and no clear bimodal distribution (TB1-nil range 0–13,350; IQR 0–75 pg/mL). TNF responses demonstrated moderate background (median 10.6 pg/mL) with a small margin between TB1 and nil levels (median 0 pg/mL), strong mitogen responses, and no clear bimodal distribution (TB1-nil range 0–2985, IQR 0–3.1). Interpretation of IL5 and IL15 was limited due to poor ESAT6/CFP10 and mitogen responses and no further analyses were undertaken. Infant age at enrollment (t = −0.01, 0.01 and −0.02), birth weight (t = −0.05, 0.03 and 0.04), maternal IPT in pregnancy (t = −0.1, −0.02 and −0.01), maternal CD4 count (t = −0.07, 0.001 and −0.02) and maternal viral load (t = 0.01, −0.06 and 0.04) in pregnancy had no correlation with IL2, IP10 and TNF TB1-nil values, respectively.

TABLE 1. - HEU Infant Demographics in the iTIPS Clinical Trial With a Clinical Endpoint Assessing the Effect of IPT
All Participants (N=267)
Characteristic Median (IQR) or n (%)
Infant characteristics
 Infant age (weeks) 6.1 (6.0 - 6·6)
 Median weight (kg) 3.5 (3.0 – 3.8)
 Male 140 (52)
 Breast-fed 266 (99.6)
 Received ARVs for PMTCT 264 (99)
 BCG vaccination scar 249 (93)
 Infant IPT 132 (49)
 Infant TB exposure 8 (3)
Maternal characteristics
  Maternal age 27 (24-31)
  Maternal ARVs
  Initiated before pregnancy 199 (74)
  Initiated during pregnancy 66 (25)
  Initiated after pregnancy 2 (1)
  Maternal CD4 (cells/mm3) 478 (349-663)
   Maternal HIV RNA (copies/ml) 0 (0-0)
   HIV viral load undetectable 203 (76)
   HIV viral load >1000 (copies/ml) 11 (4)
  Maternal history of IPT 201 (75)
  Current IPT 54 (20)
  History of TB 25 (9)
 QFT-Plus at 1-year postenrollment
 Pos 3 (1)
 Neg 245 (98)
 Ind 3 (1)
TST ≥10 mm at 1-year postenrollment
 Pos at 1 year 25 (13)
 Neg at 1 year 163 (87)
 Median TST size (mm) 0 (0-5)
PMTCT indicates prevention of mother-to-child transmission of HIV.

TABLE 2. - Non-IFNγ Cytokine Levels in QFT-Plus Supernatants From HEU Infants at ~14 Months of Age
Nil (pg/mL) Mitogen-nil (pg/mL) TB1-nil (pg/mL)*
Number 229 229 229 229 229 229 229 229 229
Num. >0 (%) 5 (2) 228 (99) 160 (70) 184 (80) 225 (98) 224 (98) 7 (3) 125 (55) 70 (31)
Median 0 553 10.6 20 4302 3014 0 8.9 0
IQR 0–0 365–890 0–66 6.6–60 3004–6068 1114–5855 0–0 0–75 0–3.1
Min/max 0/151 0/17,414 0/2334 0/1947 0/28,285 0/32,310 0/929 0/13,350 0/2985
*TB1 antigen tube has ESAT6/CFP10 antigens.

Non-IFNγ cytokine levels were obtained from QFT-Plus supernatants and stratified by positive/negative/indeterminate classification. A: QFT-Plus was obtained from HEU infants at ~14 months of age. IL2, IP10 and TNF concentrations (pg/mL) were measured from the QFT-Plus supernatant of the nil, mitogen and TB1 (ESAT6/CFP10 stimulated) tubes by Luminex assay. B: For each non-IFNγ cytokine measured from QFT-Plus supernatant, infants were classified as having a negative, positive or indeterminate result for Mtb infection at ~14 months based on the following criteria which mimic that of the QFT-Plus. Positive responses met the following criteria. (1) TB1-nil >90th percentile for cytokine and TB1-nil ≥ 25% of sample nil; (2) Nil <2x average nil for cytokine (negative control); (3) Mitogen ≥ 4.8 pg/mL (lower limit of detection of Luminex) (positive control). Non-IFNγ ESAT6/CFP10 cytokine responses were considered negative if the patient did not meet the first condition. Otherwise, a sample was considered indeterminate. C: Number of positive, negative and indeterminate infants using various Mtb infection endpoint definitions

To define positivity thresholds for non-IFNγ Mtb antigen-specific (ESAT6/CFP10) cytokine responses in this cohort of HEU infants, we modeled positive/negative/indeterminate criteria on the QFT-Plus with a conservative 90th percentile positivity threshold for batch1/batch2 of 0/0, 198/249 and 57/21 pg/mL, for IL2, IP10 and TNF, respectively (Fig. 1B). Six (2.6%), 4 (1.7%) and 9 (3.9%) infants were considered positive for IL2, IP10 and TNF, respectively, as compared with 3 (1.2%) by standard QFT-Plus interpretation, none of which had a reported Mtb exposure. Loosening the positivity cutoff to the 75th percentile identified an additional 6 participants with possible Mtb infection [total positive 21 infants: IL2, 6 (2.6%); IP10, 4 (1.7%); TNF, 16 (7.0%)] but did not meaningfully impact other statistical analyses described below.

Using a composite endpoint of any ESAT6/CFP10 response among IFNγ, IL2, IP10 and TNF, 15 of 229 (6.6%) infants had evidence of an ESAT6/CFP10-specific response (Fig. 1C). Five of 229 (2.2%) infants had either an IFNγ-response or 2 non-IFNγ cytokine responses. Together, these data suggest that non-IFNγ cytokine endpoints detect a higher prevalence of Mtb infection in HEU infants compared with an IFNγ endpoint.

We next examined the correlations across different Mtb immunologic endpoints. There was no association between TST size (as a continuous variable) and background-subtracted IL2, IP10 and TNF cytokine concentrations (t = 0.001, 0.04 and −0.05, respectively). There was minimal correlation between QFT-Plus results and the IL2 (t = 0.10), IP10 results (t = 0.08) and TNF results (t = 0.01). However, IL2/IP10 (t = 0.25) results had moderate correlation. Together, these data show that non-IFNγ endpoints are not highly correlated with traditional QFT-Plus and TST measurements.

We then examined whether IPT was associated with protection from Mtb infection defined by non-IFNγ or combined endpoints in 229 infants with non-IFNγ cytokine data where 109 received IPT and 120 did not (Table 3). Receipt of IPT was not associated with Mtb infection prevalence at 14 months when analyzed using ESAT6/CFP10 responses for IL2 [odds ratio (OR) 1.1 (95% CI: 0.2–8.8), P = 1.0], IP10 [OR 1.1 (95% CI: 0.1–15), P = 1] or TNF [OR 0.3 (95% CI: 0–1.6), P = 0.17] independently (Fig. 2). IPT was not associated with Mtb infection as measured by any positive ESAT6/CFP10-specific cytokine response (IL2, IP10, TNF or IFNγ) [OR = 0.7 (95% CI: 0.2–2.1), P = 0.54] or any non-IFNγ cytokine response [OR = 0.6 (95% CI: 0.2–1.8), P = 0.37]. This also held true when the endpoint was narrowed to include either a positive QFT-Plus or any 2 non-IFNγ cytokines positive [OR = 1.7 (95% CI: 0.3–13), P = 0.58]. We next examined whether IPT was associated with protection from Mtb infection, defined by either a positive TST or any positive ESAT6/CFP10-specific cytokine response across all 267 infants (Table 3). Forty of 267 (15%) of infants were considered to have Mtb infection. Logistic regression of TST alone trended toward significance [OR 0.4 (95% CI: 0.2–1.0), P = 0.07] similar to the parent iTIPS study17,18 and did not change significantly with the reclassification of the 15 infants with ESAT6/CFP10 specific responses [OR 0.5 (95% CI: 0.2–1.0), P = 0.05]. The addition of possible confounding variables to the regression model, including infant Mtb exposure [OR 0.5 (95% CI: 0.2, 1.0), P = 0.05], ARV in pregnancy [OR 0.5 (95% CI: 0.2–1.0), P = 0.05], prior maternal history of TB [OR 0.5 (95% CI: 0.2–1.0), P = 0.05], maternal CD4 count [OR 0.6 (95% CI: 0.3–1.2), P = 0.17], viral load in pregnancy [OR 0.5 (95% CI: 0.2–1.0), P = 0.05] and whether the infant had a BCG scar [OR 0.5 (95% CI: 0.2–1.0), P = 0.05] did not change the model. Further studies analyzing the relationship between infant Mtb exposure and infant Mtb infection outcomes did not reveal an association (TST P = 0.18, QFT P = 1, TST/QFT P = 0.12, non-IFNγ cytokine P = 1, any positive: P = 0.28, measured by Fisher’s exact test).

TABLE 3. - Effect of IPT on Mtb Infection
Endpoint N #Pos (%) #Neg (%) OR 95% CI P
TST+ or any ESAT6/CFP10 response 267 40 (15) 227 (85) 0.5 0.2–1.0 0.05
 TST+ 188 25 (13) 163 (87) 0.4 0.2–1.0 0.07
 Any ESAT6/CFP10 response 229 15 (7) 214 (93) 0.7 0.2–2.1 0.54
  QFT-Plus (IFNγ)* 251 3 (1) 245 (98) 2.2 0.1–130 0.61
  Any non-IFNγ cytokine 229 14 (6) 214 (93) 0.6 0.2–1.8 0.37
   IL2* 229 6 (3) 178 (78) 1.1 0.1–8.8 1
   IP10* 229 4 (2) 202 (88) 1.1 0.1–15 1
   TNF* 229 9 (4) 194 (85) 0.3 0.03–1.6 0.17
 Any 2 ESAT6/CFP10 response or QFT+ 229 5 (2) 224 (98) 1.7 0.3–13 0.58
*Fisher’s exact test used for individual QFT, IL2, IP10, TNF analysis due to small number of positive cases.

Effect of IPT on IFNγ, IL2, IP10 and TNF Mtb-specific antigen responses. A: Number of HEU infants with any Mtb-antigen specific (ESAT6/CFP10) cytokine response and at least 2 Mtb-antigen specific (ESAT6/CFP10) cytokine responses stratified by whether the infant received IPT for the first year of life. B: Cytokine concentration (TB1-nil) stratified by whether the infant received IPT for the first year of life.


This study is the first to assess non-IFNγ cytokine responses to Mtb specific-antigens in HEU infants, and among the first to assess these responses in infants overall. It also provides additional information to complement the results of the iTIPS clinical trial. Our primary findings indicate that measuring non-IFNγ cytokines identified an additional 12 infants (5.2%) with possible Mtb infection beyond the 3 who tested positive by QFT-Plus (1.1%). None of these 15 infants had a positive TST result. These data suggest that non-IFNγ cytokines may be more sensitive than IFNγ in detecting infant Mtb infection.

When we measured Mtb infection by the presence of Mtb-antigen specific immune responses (either QFT-Plus or non-IFNγ cytokines) and excluded TST results, we did not detect a trend toward protection against Mtb infection in BCG-vaccinated HEU infants who received IPT as was seen in the iTIPS study, which used a combined TST/IGRA endpoint.17,18 However, we did detect a trend for IPT protection against Mtb infection when using a composite infection endpoint of a positive TST or any ESAT6/CFP10 cytokine response to define positivity. As the relationship between Mtb infection outcome and IPT was driven by the non-Mtb specific TST, the possibility remains that the association between IPT and TST size reflects either NTM infection or an effect of IPT administration on BCG efficacy (despite initiating IPT 6 weeks of postvaccination) rather than Mtb infection. Given limitations in detecting infant Mtb infection related to low infant Th1-cytokine responses and TST cross-reactivity to BCG vaccine, there is a critical need for further research investigating alternative Mtb diagnostic strategies in this population. In this study, we used multiple diagnostic approaches (QFT, TST and non-IFNγ cytokines) to maximize test sensitivity and capture more infants with possible Mtb-infection. While measurement of non-IFNγ cytokines in this study did not yield additional evidence to suggest potential IPT protection against Mtb infection in HEU infants compared with the original prespecified clinical trial endpoints, this study does provide evidence that assessing non-IFNγ cytokine responses in addition to QFT and TST outcomes may be helpful in evaluating the efficacy of novel Mtb vaccine candidates in a clinical trial setting.

In addition to no standard for diagnosing infant Mtb infection, another limitation of this study is the lack of clearly defined cutoffs for non-IFNγ cytokine positivity. Interpreting non-IFNγ cytokine responses is limited by the lack of bimodal distributions for IP10 and TNF and inability to use a standard of proven TB disease to make an AUC curve in an infant population to rigorously define positivity cutoffs. The latter is necessary before implementing diagnostic policy changes. IFNγ responses measured by QFT similarly lack bimodal distributions, and QFT positivity cutoffs (conventionally 0.35 IU/mL) continue to be controversial especially in the pediatric population.35,36 We attempted to recapitulate the structure of the QFT in our determination of positive/negative/indeterminate non-IFNγ outcomes, adjusting for background and mitogen responses, and though we acknowledge that ultimately the decision for positivity cutoff (ie, 90th percentile) was arbitrary, we attempted to be conservative in defining positive cases. Trial of a different cutoff of 75th percentile did not meaningfully change the statistical results.

Another limitation is low statistical power when using individual cytokine endpoints (IFNγ, IL2 and IP10) to assess the effect of IPT on Mtb infection due to small numbers of positive outcomes. We partially addressed this limitation by using a combined endpoint of any ESAT6/CFP10 cytokine response among IFNγ, IL2, IP10 and TNF, though we remain unable to draw conclusions about individual cytokines. Additionally, as we did not enroll HIV-unexposed infants in the original clinical trial, we are unable to determine the effect of HIV exposure on infant T-cell responses, nor are we able to account for TB exposure unknown to the mother.

Non-IFNγ cytokine responses did not correlate with TST size. Although QFT and TST have high concordance in TB endemic settings, a percentage of values are often discordant since the assays do not measure the same immunologic response. For example, among non-US born children, those <2 years old had discordant TST/QFT results 67% of the time with the vast majority TST positive and IGRA negative.37 Our study highlights further discordance with a non-IFNγ endpoint. This combination endpoint of any ESAT6/CFP10 cytokine response was discordant with TST results in 22% of the time, approximately twice the rate reported in the literature for QFT/TST discordance in BCG-vaccinated children living in endemic settings.38,39 Poor sensitivity of QFT and TST in children, especially those <2 years old is a primary contributor for this discordance,40,41 and is likely exacerbated by the young age of participants in our study at the time of Mtb outcome evaluation. Correlations among all tested cytokines were low to moderate, suggesting each additional cytokine measurement provided novel information about ESAT6/CFP10 responses and identified additional patients with probable Mtb infection.

Studies looking at non-IFNγ infant responses to ESAT6/CFP10 are very limited. A recent study by Lubyayi et al34 in healthy HIV-unexposed Uganda infants born to LTBI+ and LTBI− mothers and followed with serial whole blood Luminex assays through 1 year of life also demonstrated detectable IL2, IP10 and TNF responses in addition to IFNγ, IL5, IL13, IL10 and IL17 among others. IP10 and TNF responses were approximately 10-fold higher, though whether this was due to different sample types (whole blood vs. QFT-Plus supernatant) or the fact that these infants were HIV-unexposed requires further study.

When comparing our IL2, TNF and IP10 cytokine levels with those reported in other pediatric studies22–25 assessing non-IFNγ cytokines in QFT supernatants, we detected lower cytokine levels in nil and TB1 tubes. This may be due to decreased infant Th1-cytokine responses compared with adults and young children,42,43 especially as median age in other pediatric studies ranged from 2.5 to 10 years old. Additionally, in utero HIV exposure may adversely impact cytokine secretion, especially as QFT supernatants from children with HIV/Mtb co-infection contain less detectable IFNγ and IL2 compared with HIV-negative Mtb-infected children.25 Also, many studies23–25,44 use samples from children in nonendemic settings referred due to concern for TB disease, which may be associated with higher cytokine levels. Finally, our samples are from a Kenyan cohort with different NTM background exposures that may cross-react with ESAT6/CFP10, altering measured cytokine levels.

Similar to other pediatric studies,22–25 we demonstrate that IL2, IP10 and TNF may differentiate Mtb-infected from uninfected children and that using multiple cytokines improves detection of Mtb infection.45 However, in contrast to other studies22–25 showing high sensitivity of IP10 and IL2 in detecting Mtb infection in the pediatric population, including children with HIV, we find that TNF captured the highest number of additional infants with ESAT6/CFP-10 specific cytokines responses. Whether this is reflective of underlying infant T-cell biology in the context of Mtb requires further study.

Presence of Mtb sensitization at birth, demonstrated in multiple studies,34,46–49 may also confound interpretation of infant TST and ESAT6/CFP10 responses in evaluation for Mtb infection, even after 1 year of life. Infant ESAT6/CFP10 responses may be due to Mtb infection or maternal sensitization. The mechanisms underlying maternal transfer of Mtb-specific antigens and/or immune responses, and the extent to which this process impacts Mtb diagnostic results in infants and young children, is unknown. Mtb specific-antigen induced IFNγ, IP10 and TNF have been detected in cord blood at birth, regardless of maternal Mtb infection status; kinetics demonstrated that these cytokines increased over the first 24 weeks of life before plateauing and remained elevated at 1 year of age.34 Research is needed to assess these early infant Mtb-antigen specific responses on a cellular level and elucidate the mechanisms and effects of maternal transfer.

In conclusion, this study demonstrates that IL2, TNF and IP10 are promising biomarkers in diagnosing primary Mtb infection in HEU infants and young children. Further studies are needed to optimize the diagnosis of Mtb infection in these age groups.


We acknowledge iTIPS Study Clinic Staff, Kisumu and Siaya County Directors of Health, and UW-Kenya and Kenyatta National Hospital Research and Programs health facility and operational staff. We thank Qiagen for discounted QFT-Plus kits and Kenya Medical Research Institute CDC for performing QFT-Plus assays. We thank Madison Jones, Lara Joudeh, and Carlo Hernandez and our colleagues at Fred Hutch for assistance with Luminex assays. Most importantly, we thank the families who participated in the study.


1. Dodd PJ, Yuen CM, Sismanidis C, et al. The global burden of tuberculosis mortality in children: a mathematical modelling study. Lancet Glob Health. 2017;5:e898–e906.
2. Jenkins HE, Yuen CM, Rodriguez CA, et al. Mortality in children diagnosed with tuberculosis: a systematic review and meta-analysis. Lancet Infect Dis. 2017;17:285–295.
3. World Health Organization. Global tuberculosis report 2019. 2019. Available at: Accessed July 8, 2020.
4. Machingaidze S, Wiysonge CS, Gonzalez-Angulo Y, et al. The utility of an interferon gamma release assay for diagnosis of latent tuberculosis infection and disease in children: a systematic review and meta-analysis. Pediatr Infect Dis J. 2011;30:694–700.
5. Mandalakas AM, Detjen AK, Hesseling AC, et al. Interferon-gamma release assays and childhood tuberculosis: systematic review and meta-analysis. Int J Tuberc Lung Dis. 2011;15:1018–1032.
6. Cotton MF, Schaaf HS, Lottering G, et al.; PACTG 1041 Team. Tuberculosis exposure in HIV-exposed infants in a high-prevalence setting. Int J Tuberc Lung Dis. 2008;12:225–227.
7. Madhi SA, Nachman S, Violari A, et al.; P1041 Study Team. Primary isoniazid prophylaxis against tuberculosis in HIV-exposed children. N Engl J Med. 2011;365:21–31.
8. Abu-Raya B, Kollmann TR, Marchant A, et al. The immune system of HIV-exposed uninfected infants. Front Immunol. 2016;7:383. doi: 10.3389/fimmu.2016.00383
9. Hesseling AC, Jaspan HB, Black GF, et al. Immunogenicity of BCG in HIV-exposed and non-exposed infants following routine birth or delayed vaccination. Int J Tuberc Lung Dis. 2015;19:454–462.
10. Kidzeru EB, Hesseling AC, Passmore JA, et al. In-utero exposure to maternal HIV infection alters T-cell immune responses to vaccination in HIV-uninfected infants. AIDS. 2014;28:1421–1430.
11. Marais BJ, Gie RP, Schaaf HS, et al. The clinical epidemiology of childhood pulmonary tuberculosis: a critical review of literature from the pre-chemotherapy era. Int J Tuberc Lung Dis. 2004;8:278–285.
12. Martinez L, Lo NC, Cords O, et al. Paediatric tuberculosis transmission outside the household: challenging historical paradigms to inform future public health strategies. Lancet Respir Med. 2019;7:544–552.
13. Moyo S, Verver S, Mahomed H, et al. Age-related tuberculosis incidence and severity in children under 5 years of age in Cape Town, South Africa. Int J Tuberc Lung Dis. 2010;14:149–154.
14. le Roux SM, Abrams EJ, Donald KA, et al. Infectious morbidity of breastfed, HIV-exposed uninfected infants under conditions of universal antiretroviral therapy in South Africa: a prospective cohort study. Lancet Child Adolesc Health. 2020;4:220–231.
15. Tameris MD, Hatherill M, Landry BS, et al.; MVA85A 020 Trial Study Team. Safety and efficacy of MVA85A, a new tuberculosis vaccine, in infants previously vaccinated with BCG: a randomised, placebo-controlled phase 2b trial. Lancet. 2013;381:1021–1028.
16. World Health Organization. Latent TB Infection: Updated and consolidated guidelines for programmatic management. 2018. Available at: Accessed July 5, 2020.
17. LaCourse SM, Richardson BA, Kinuthia J, et al. A randomized controlled trial of isoniazid to prevent Mycobacterium tuberculosis infection in Kenyan HIV-exposed uninfected infants. Clin Infect Dis. 2021;73:e337–e344.
18. LaCourse SM, Richardson BA, Kinuthia J, et al. Infant TB Infection Prevention Study (iTIPS): a randomised trial protocol evaluating isoniazid to prevent M. tuberculosis infection in HIV-exposed uninfected children. BMJ Open. 2020;10:e034308.
19. Biselli R, Mariotti S, Sargentini V, et al. Detection of interleukin-2 in addition to interferon-gamma discriminates active tuberculosis patients, latently infected individuals, and controls. Clin Microbiol Infect. 2010;16:1282–1284.
20. Kim SY, Park MS, Kim YS, et al. The responses of multiple cytokines following incubation of whole blood from TB patients, latently infected individuals and controls with the TB antigens ESAT-6, CFP-10 and TB7.7. Scand J Immunol. 2012;76:580–586.
21. Rubbo PA, Nagot N, Le Moing V, et al. Multicytokine detection improves latent tuberculosis diagnosis in health care workers. J Clin Microbiol. 2012;50:1711–1717.
22. Armand M, Chhor V, de Lauzanne A, et al. Cytokine responses to quantiferon peptides in pediatric tuberculosis: a pilot study. J Infect. 2014;68:62–70.
23. Hur YG, Crampin AC, Chisambo C, et al. Identification of immunological biomarkers which may differentiate latent tuberculosis from exposure to environmental nontuberculous mycobacteria in children. Clin Vaccine Immunol. 2014;21:133–142.
24. Lighter-Fisher J, Peng CH, Tse DB. Cytokine responses to QuantiFERON® peptides, purified protein derivative and recombinant ESAT-6 in children with tuberculosis. Int J Tuberc Lung Dis. 2010;14:1548–1555.
25. Jeljeli M, Guérin-El Khourouj V, Pommelet V, et al. Cytokine/chemokine secretion for detecting tuberculosis in quantiferon supernatants from HIV+ and HIV- children. J Infect. 2017;75:77–80.
26. Ruhwald M, Bodmer T, Maier C, et al.; TBNET. Evaluating the potential of IP-10 and MCP-2 as biomarkers for the diagnosis of tuberculosis. Eur Respir J. 2008;32:1607–1615.
27. Alsleben N, Ruhwald M, Rüssmann H, et al. Interferon-gamma inducible protein 10 as a biomarker for active tuberculosis and latent tuberculosis infection in children: a case-control study. Scand J Infect Dis. 2012;44:256–262.
28. Goletti D, Raja A, Syed Ahamed Kabeer B, et al. Is IP-10 an accurate marker for detecting M. tuberculosis-specific response in HIV-infected persons? PLoS One. 2010;5:e12577.
29. Tuuminen T, Salo E, Kotilainen H, et al. Evaluation of the filter paper IP-10 tests in school children after exposure to tuberculosis: a prospective cohort study with a 4-year follow-up. BMJ Open. 2012;2:e001751.
30. Petrone L, Cannas A, Aloi F, et al. Blood or Urine IP-10 cannot discriminate between active tuberculosis and respiratory diseases different from tuberculosis in children. Biomed Res Int. 2015;2015:589471.
31. Ruhwald M, Aabye MG, Ravn P. IP-10 release assays in the diagnosis of tuberculosis infection: current status and future directions. Expert Rev Mol Diagn. 2012;12:175–187.
32. Bihari S, Cavalcanti N, Correia JB, et al. Interferon gamma-induced protein-10 concentrations in children with previous tuberculosis infections and disease. Pediatr Infect Dis J. 2012;31:1089–1091.
33. Yassin MA, Petrucci R, Garie KT, et al. Can interferon-gamma or interferon-gamma-induced-protein-10 differentiate tuberculosis infection and disease in children of high endemic areas? PLoS One. 2011;6:e23733.
34. Lubyayi L, Mawa PA, Nabakooza G, et al. Maternal latent Mycobacterium tuberculosis does not affect the infant immune response following BCG at birth: an observational longitudinal study in Uganda. Front Immunol. 2020;11:929.
35. Jonsson J, Westman A, Bruchfeld J, et al. A borderline range for Quantiferon Gold In-Tube results. PLoS One. 2017;12:e0187313.
36. Lighter J, Rigaud M, Eduardo R, et al. Latent tuberculosis diagnosis in children by using the QuantiFERON-TB Gold In-Tube test. Pediatrics. 2009;123:30–37.
37. Ahmed A, Feng PI, Gaensbauer JT, et al.; TUBERCULOSIS EPIDEMIOLOGIC STUDIES CONSORTIUM. Interferon-γ release assays in children <15 years of age. Pediatrics. 2020;145:e20191930.
38. Altet-Gómez N, De Souza-Galvao M, Latorre I, et al. Diagnosing TB infection in children: analysis of discordances using in vitro tests and the tuberculin skin test. Eur Respir J. 2011;37:1166–1174.
39. Pavić I, Katalinić-Janković V, Čepin-Bogović J, et al. Discordance between tuberculin skin test and interferon-γ release assay in children younger than 5 years who have been vaccinated with Bacillus Calmette-Guérin. Lab Med. 2015;3:200–206
40. Moyo S, Isaacs F, Gelderbloem S, et al. Tuberculin skin test and QuantiFERON® assay in young children investigated for tuberculosis in South Africa. Int J Tuberc Lung Dis. 2011;15:1176–81, i.
41. Jenum S, Selvam S, Mahelai D, et al. Influence of age and nutritional status on the performance of the tuberculin skin test and QuantiFERON-TB gold in-tube in young children evaluated for tuberculosis in Southern India. Pediatr Infect Dis J. 2014;33:e260–e269.
42. Jeljeli M, Guérin-El Khourouj V, Pédron B, et al. Ontogeny of cytokine responses to PHA from birth to adulthood. Pediatr Res. 2019;86:63–70.
43. Decker ML, Grobusch MP, Ritz N. Influence of age and other factors on cytokine expression profiles in healthy Children-A systematic review. Front Pediatr. 2017;5:255.
44. Jones CE, Hesseling AC, Tena-Coki NG, et al. The impact of HIV exposure and maternal Mycobacterium tuberculosis infection on infant immune responses to Bacille Calmette-Guérin vaccination. AIDS. 2015;29:155–165.
45. Lundtoft C, Awuah AA, Nausch N, et al. Alternative Quantiferon cytokines for diagnosis of children with active tuberculosis and HIV co-infection in Ghana. Med Microbiol Immunol. 2017;206:259–265.
46. Jeljeli M, Guérin-El Khourouj V, Pommelet V, et al. Immune response to Mycobacterium tuberculosis in young contacts with discordant immunological test results. J Infect. 2016;73:517–520.
47. Van Rie A, Madhi SA, Heera JR, et al. Gamma interferon production in response to Mycobacterium bovis BCG and Mycobacterium tuberculosis antigens in infants born to human immunodeficiency virus-infected mothers. Clin Vaccine Immunol. 2006;13:246–252.
48. van den Biggelaar AH, Prescott SL, Roponen M, et al. Neonatal innate cytokine responses to BCG controlling T-cell development vary between populations. J Allergy Clin Immunol. 2009;124:544–50, 550.e1.
49. Kampmann B, Tena GN, Mzazi S, et al. Novel human in vitro system for evaluating antimycobacterial vaccines. Infect Immun. 2004;72:6401–6407.

tuberculosis; HIV-exposed uninfected infants; diagnostic testing; immune response

Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.