Study population
A total of 230 TB-exposed children were enrolled. Of these, 76 (33%) had been exposed in the last 6 months (e.g. retrospective cases) and 154 (67%) children were recruited prospectively as new index cases were diagnosed.
All the children enrolled had received the BCG vaccine in the first two months of life. Of the 230 children recruited, 52.1% were male and 47.9% were female. Moreover, the median age of the participants was 8.5 (range 1-17) years, and 4.8% were less than 24 months of age. Demographic characteristics and symptoms identified at the time of screening are shown in Table 1. Through clinical and microbiological evaluation, one household contact was diagnosed with active TB at baseline, one child 12 months after the diagnosis of the index case, and another 18 months after the diagnosis of the index case. Accordingly, 227 children (98.7%) remained asymptomatic.
TST and QFT-Plus results
Totally, 36% (80/230) of child TB contacts were TST positive in our survey. Exhaustively, 12.6% (29/230), 8% (16/201), 13.5% (29/185), and 8.3% (13/156) of the evaluated contacts showed positive TST results, respectively, at baseline, three months, 12 months, and 18 months after the diagnosis of the index case. QFT-Plus results indicated that the positive rate of LTBI for the child contacts was 37.8% (87/230), in which 20.9% (48/230) of the participants were positive at the initiation of screening. Moreover, among the contacts with baseline QFT-Plus-negative results and valid follow-up QFT-Plus data, 13.7% (25/182), 6.4% (10/157), and 2.7 (4/147), showed an increase over the baseline IFN-γ value after 3, 12, and 18 months, respectively.
According to the TST/QFT-Plus results and the absence of clinical signs/radiological evidence of active TB, 45.2% (104/230) of children were identified as LTBI during our study. Our findings revealed that 28.3% (65/230) of the household pediatric contacts were considered as LTBI at the baseline, while, the rate of LTBI conversion was 17% (39/230). Exhaustively, 2.3% (12/165), 14.4% (22/153), and 2.3% (5/131) of contacts were identified as LTBI after 3, 12, and 18 months, respectively.
IP-10 Results
Significantly increased IP-10 levels were found in LTBI patients (Mean ± SEM: 2134 ± 362.5 for TB1 and 2379 ± 408.5 for TB2) compared to healthy contacts (TST-/QFT-) (Mean ± SEM: 208.6 ± 90.58 for TB1 and 158.4 ± 72.56 for TB2) in response to both TB1 and TB2 stimulation (Figure 1). To define IP-10 results as positive or negative, a cut-off value for IP-10 was calculated using the ROC curve. For TB1 a cut-off of 67.2 pg/mL identifies LTBI with 60.2% sensitivity (50.5%-69.1%) and 90% specificity (83.12%-94.38%); similarly, for TB2, an IP-10 level > 142.3 pg/mL predicted LTBI with 53.4% sensitivity (43.8%-62.7%) and 93.7% specificity (87.5%- 96.9%). Accordingly, 59.6% (62/104) of LTBI contacts and 11.3% (14/123) of healthy contacts were considered as IP-10-TB1-positive, respectively. In addition, the proportion of LTBI and healthy children with positive IP-10-TB2 was, respectively, 53.8% (56/104) and 8.9% (11/123).
Discussion
Children in close contact with an active TB case are at high risk of LTBI. The WHO recommends prophylactic treatment for young children exposed to a TB source case without examining whether the child is infected (28, 29). Recognizing children with active TB or LTBI and either treating disease or targeting prophylaxis to high-risk groups are strategies recommended in high TB burden regions. However, the identification of LTBI cases is more challenging in low–middle-income countries.
To the best of our knowledge, this study is the most comprehensive investigation on the prevalence of LTBI among household child TB contacts in Iran. Estimating the prevalence of LTBI among child contacts of TB cases in the absence of a gold standard assay remains a challenge. This is of special relevance in the current study for using different tests for the identification of LTBI cases, in which it is possible to determine if an individual is really infected.
In addition, since it is problematic to accurately identify at what time an individual was infectious, we performed our screening from the time at which the index case showed symptomatology compatible with TB until 18 months later and at specified intervals. Our method provided useful insights into the prevalence and incidence of LTBI among child household contacts of TB cases in two high burden provinces in Iran.
Our findings revealed that the prevalence and incidence of LTBI among Iranian children exposed to an index TB case with sputum smear and/or culture-positive adults were 28.3% and 17%, respectively. These children are at high risk of developing active TB which is very worrying. Our findings are similar to other investigations in which only under half of child household contacts have evidence of LTBI (30, 31). The high prevalence of LTBI found in our study supports that children are very susceptible to exposure to an active TB case and are at a high risk of LTBI (31). Moreover, we detected three patients with active TB, in which two cases did not have clinical evidence of active disease at the beginning of the screening, thus were considered as incident cases. However, we cannot claim conclusively that the source of the infection in this population was the index case.
A meta-analysis including 95 studies from low- and middle-income settings estimated the prevalence of LTBI among household contacts to be 51.5% (95% CI 47.1-55.8%, I(2)=98.9%), with the highest incidence among children between 5 and 14 years of age (31). In another study in Laos, Nguyen et al reported that the prevalence of LTBI among child household contacts was 31.1% and increased with age from 26.0% in children below 5 years to 35.7% in children between 6 – 15 years (32). So far, there is only limited data available on prevalence of LTBI among household contacts of TB cases in Iran. Previously, Shamaei et al. reported that 27.5% of close-family contacts of hospitalized TB patients in Iran were diagnosed with LTBI (33). Therefore, our data reinforce the crucial importance of systematic screening and close follow-up of child household contacts, as well as a precise targeting of preventive treatment in this population.
TST is used as the only screening method to identify LTBI children in several high burden TB countries with limited resources, but the low sensitivity of TST, its poor specificity due to cross-reactions with environmental non-tuberculous mycobacteria and the BCG vaccine, as well as its unreliability in children with immunosuppression and malnutrition could limit the applicability of this approach (34, 35). On the other hand, although IGRAs have emerged as promising alternatives to the TST, the limitations of these tests are gradually becoming apparent with their widespread use (36, 37). For instance, previous surveys in both high and low TB burden regions confirmed that IGRA conversions and reversions could occur and there is some confusion about how to interpret such results (38-40). Accordingly, several national guidelines (e.g. UK, Canada, Germany, the Netherlands, Italy, Switzerland, Norway, Spain, Ireland, and Korea) recommend a two-step approach of TST first, followed by an IGRA (36). In the current study, we used both TST and IGRA assay to investigate LTBI cases among close contacts to an index TB case. In addition, we rechecked TST or IGRA in the subjects after 3, 12, and 18 months when the primary result was negative. Our findings indicated that there was a disagreement between TST and QFT-plus results. For contacts in whom TST and QFT-plus results were discordant, it is impossible to know which test was correct because there is no reference standard. In our study, BCG vaccination status might affect TST positivity in adjusted multivariable analysis. The quality of BCG vaccination among contacts and the percentage of subjects with booster vaccinations could contribute to the different degrees of concordance between TST and QFT-plus results. Moreover, although TST was performed by trained personnel following standard procedures, the disagreement between TST and QFT-plus results could be due to the variability in reading TST in the pediatric population. Some authors previously reported that the disagreement rate between TST and IGRA tests was about 10%, regardless of BCG vaccination (41, 42).
In the present study, we also evaluated the accuracy of IP-10 assay in the plasma of QFT-Plus for the diagnosis of LTBI among Iranian household pediatric contacts of TB cases. The results of some investigations suggest that IP-10 assay could be an alternative to IGRA, may be useful for diagnosing LTBI cases as it would be easier to use (21, 43). Our findings indicated that IP-10 levels are significantly increased in contacts with LTBI compared to healthy contacts in response to both TB1 and TB2 antigens (P value < 0.0001). Considering the IP-10 results, no significant differences were observed between the TB1 and TB2. Therefore, the IP-10-based assay showed a high discriminating power and could help to distinguish between child contacts with LTBI from healthy contacts. In addition, a high concordance between the IP-10 assay and TST and QFT-plus tests was found.
IP-10 assay was previously thought to have significant advantages over IGRAs, including the possibility to use smaller blood volumes and to be detected in urine samples. It is also possible to measure IP-10 in dried plasma spots in filter papers allowing cheap and simple mail transportation at room temperature (21, 24, 44, 45). Therefore, IP-10 assay would be useful as an uncomplicated and inexpensive alternative test for LTBI diagnosis in child household contacts of TB cases, especially in low-resource settings. Although in our study IP-10 assay showed an acceptable specificity (90% for TB1 and 93.7% for TB2), it had a low sensitivity (60.2% for TB1 and 53.4% for TB2) for LTBI diagnosis. Importantly, it should be noted that the combination of the IGRAs and IP-10 assay has been reported to increase sensitivity (21). Moreover, early studies suggest that the IP-10-based assay could not help to discriminate between active TB and LTBI among children (46).
The advantage of this study is that we used various approach for identification of LTBI among child household contacts, including TST, QFT, and IP-10 assay. Hence, our methodology allowed calculation of the prevalence of LTBI among child household contacts which, so far, has not been performed since LTBI cannot easily be demonstrated using a single method. However, some limitations could be noted in our study. For instance, investigator bias was further limited by the fact that TST could be interpreted differently by independent experts.