Background Household contacts (HHCs) of pulmonary TB patients are at high risk of Mycobacterium tuberculosis (Mtb) infection and early disease development. Tuberculin skin test (TST) has been traditionally used to identify infected individuals; however, its use is limited by low specificity in populations with high levels of BCG vaccination or significant exposure to non-tuberculosis mycobacteria (NTM), and reduced sensitivity in immunocompromised individuals. Interferon-gamma release assays (IGRAs) such as QuantiFERON-TB Gold In-Tube (QFT-GIT) using Mtb specific antigens provide an alternative to TST for infection detection. IGRAs are now widely used for the detection of Mtb infection and are included in the guidelines of many countries with a low incidence of TB. Despite a growing body of literature on IGRAs, the relationship between the magnitude of T-cell Interferon-γ responses to Mtb specific antigens and risk of progression to disease has not been studied. Objective The main objective of this study was to determine whether HHCs with high (≥10 IU/ml) levels of IFN-γ in response to Mtb specific antigens (ESAT-6, CFP-10 or TB 7.7) in the QFT-GIT assay are at higher risk of developing TB compared to those with low (> 0.35-<10 IU/ml) levels. Other secondary objectives included to determine the following: the performance and operational characteristics of QFT-GIT in a field setting; risk factors associated with positive QFT-GIT results; concordance between the two tests; incidence rates of TB in HHCs with positive and negative QFT-GIT and TST results at baseline as well as positive and negative predictive values. Method This study was nested within a large community randomized trial called ZAMSTAR implemented in 16 communities in Zambia and 8 communities in the Western Cape Province of South Africa. A cohort of HIV-positive and HIV-negative adult (≥ 15 years) HHCs were prospectively followed for 2-4 years. Consenting HHCs had blood drawn for HIV antibodies. QFT-GIT test was performed according to the manufacturer's instructions. TST were performed according to the standard IUATLD protocol. A standardized questionnaire was used to collect information on risk factors for TB and TB treatment information (for those with TB). Results The feasibility studies showed three main findings. Firstly, the sensitivity of QFT-GIT was greater than that of TST overall, at all the standard TST cut-offs and when stratified by HIV status. The sensitivity of QFT-GIT was 85.6% (95%CI: 77.0-91.9) (indeterminate results excluded) compared to that of TST at 51.6% (95% CI: 40.9-62.2) at a cut-off of ≥ 10 mm. Secondly, test-retest reproducibility of QFT-GIT was high at 91.74% (ICC: 0.90; 95% CI 0.82-0.97). Thirdly, in this setting, some biological and operational factors that affected the performance of QFT-GIT were identified such as HIV positivity, low CD4+ T-lymphocytes, delayed incubation of blood samples and power outages. 8 For the main study, the study population at baseline consisted of 1,789 HHCs who were predominantly women (71%); median age was 28 years (IQR: 21-43); HIV positivity rate was 27.9%. Prevalence of tuberculous infection was 63.7% as measured by QFT-GIT and 39.6% by TST. There was a low level of agreement between the tests regardless of TST cut-off point (% agreement=59.7%; kappa=0.24). QFT+/TST- discordance (575/719; 80%) was more frequent than QFT-/TST+ discordance (144/719; 20%) at TST ≥10 mm. Risk factors associated with QFT-GIT positivity were identified at baseline. In multivariable analysis adjusted for sex, age, and community, HIV status was negatively associated with QFTGIT positivity (aOR: 0.48; 95% CI: 0.37–0.63; p<0.001) whereas residing in an urban area (aOR: 2.37; 95% CI: 1.10–5.13; p<0.03), smear status of index (OR: 1.26; 95% CI: 0.91-1.76; p=0.15) and country (aOR: 1.93; 95% CI: 1.48–2.51; p<0.001) were positively associated with QFT-GIT positivity. Similar results were obtained for TST. From a total of 1789 HHCs seen at baseline, 1113 (62.2%) HHCs entered follow-up and were included in the main analysis. The overall incidence rate of TB was 20.96/1000 pyrs (95% CI: 15.93-27.58). TB incidence rate was higher among test positive HHCs compared to those who were negative (IRR for QFT-GIT: 1.65; 95% CI: 0.86-3.37; p=0.06) and for TST (IRR: 1.88; 95%CI: 1.04-3.41; p=0.01). Results were similar in univariable analysis (QFT-GIT: 1.66 (95%CI: 0.88-3.11; p=0.11) and TST: 1.89 (95%CI: 1.09-3.28; p=0.02)) and multivariable analysis adjusted for sex, age and HIV (QFT-GIT: 2.20 (95%CI: 1.14-4.25; p=0.02) and TST: 2.19 (95%CI: 1.24-3.86; p=0.007)). Overall, PPV for QFT-GIT was 5.38% (95%CI: 3.84-7.31), compared to TST, 6.57% (95% CI: 4.41- 9.36). Overall for QFT-GIT, the IRR was higher among HIV negative HHCs (IRR: 3.85; 95%CI: 0.90-34.51; p=0.07) compared to HIV positives (IRR; 1.93; 95%CI: 0.88-4.57; p=0.04). Overall for TST, the IRR for HIV negatives (IRR: 2.21; 95%CI: 0.78-6.72; p=0.05) was similar to that among HIV positives (IRR: 2.32; 95%CI: 1.09-5.00; p=0.009). Univariable analysis showed similar results for both tests. In multivariable analyses adjusted for age, sex and country as an effect modifier, the HR for developing TB was 4.72 (95%CI: 1.35-16.46; p=0.01) in HIV positive QFT-GIT positives compared to 2.13 (95%CI: 0.81-5.60; p=0.12) in HIV positives TST positive HHCs. Risk factors for TB were identified. In multivariable analyses, adjusted for age, sex , HIV status and country there was strong evidence that occasional smoking, (HR: 4.07; 95%CI:1.31-12.63), HIV positivity (HR: 4.60; 95%CI:2.48-8.56), smear positivity of the index (HR: 2.00 ; 95%CI:1.04- 3.87) and country (HR: 1.79 ; 95%1.02-3.15; p=0.04) ) were associated with incidence of TB. Out of the 1,113 HHCs who entered follow-up, 406 HHCs had IFN-γ levels <0.35 IU/ml and were excluded leaving 707 HHCs in analysis for the primary objective. Out of these 536 (75.8%) had IFN-gamma levels ≥ 0.35 and <10 IU/ml (low IFN-γ levels) while 171 (24.2%) HHCs had ≥ 10 IU/ml (high IFN-γ levels). Out of the 707 HHCs that entered follow-up, 38 (5.4%) HHCs developed active TB over 1558.0 person-years (pyrs) of follow-up, giving an incidence rate of 24.39/1000 pyrs (95% CI: 17.75- 33.52).TB incidence rates were 24.51/1000 pyrs (9 cases/367.2 pyrs) in HHCs with high levels and 24.35 (29 cases/1190.7 pyrs) among those with low levels of IFN-γ, giving an IRR of 1.0 (95% CI: 0.42-2.18; p=0.48). Overall, unadjusted HR in HHCs with high IFN-γ levels was 1.02 (95%CI: 0.48-2.15; p=0.96) while in multivariable analysis adjusted for age, sex, country and HIV as an effect modifier, HR was 1.74 (95%CI: 0.63-4.79; p=0.29). TB incidence rates in HIV positives was 51.94/1000 pyrs (3 cases/57.8 pyrs) in HHCs with high levels and 65.29/1000 pyrs (19 cases/291.0 pyrs) among those with low levels of IFN-γ, giving an IRR of 0.79 (95%CI: 0.15-2.70; p=0.38).TB incidence rates in HIV negatives were 19.56/1000 pyrs (6 cases/306.7 pyrs) in HHCs with high levels and 11.47 (10 cases/871.7 pyrs) among those with low levels of IFN-γ, giving an IRR of 1.70 (95%CI: 0.51-5.18, p=0.16). Unadjusted HR among HIV negative HHCs was 1.73 (95%CI: 0.63-4.77; p=0.29) and 0.75 (95%0.22-2.55; p=0.65) among HIV positive ones respectively. In multivariable analysis adjusted for age, sex and country, the HR remained similar as unadjusted analysis for both HIV negatives and positives. For all the groups used for sensitivity analysis of the primary question, HHCs with the highest IFN-γ levels had increased IRRs ranging from 1.5 to 2 compared to the reference sub-group. For HIV negatives, HHCs with the highest IFN-γ levels had the highest IRRs in all groups apart from one group. HIV negative HHCs with the highest IFN-γ levels had increased IRRs ranging from 4 to 5-fold compared to the reference sub-group. In comparison, HIV positive HHCs with the highest IFN-γ levels had increased IRRs ranging from 1.6 to 2.6 compared to the reference sub-group. Conclusions The principal finding in this study is that there was no difference in incidence rates between HHCs with low and high levels (overall IRR: 1.0 (95% CI: 0.42-2.18)). Another principal finding was that there was strong evidence of a five-fold increased risk of TB in HIV positive QFT-GIT positive HHCs compared to HIV positive QFT-GIT negative ones (aHR : 4.72; 95%CI: 1.35-16.46; P=0.01). For all the groups used in the sensitivity analysis of the primary question, HHCs with the highest IFN-γ levels had increased IRRs ranging from 1.5 to 2 compared to the reference sub-group. The feasibility studies emphasized the need for stringent sample collection and processing techniques to ensure the accuracy of QFT-GIT results.