Comparing the performance of QuantiFERON-TB Gold Plus with QuantiFERON-TB Gold in-tube among highly TB exposed gold miners in South Africa

Study setting and population

The study was conducted in the South African goldmines at the Occupational Health Centre (OHC), in Orkney, North West Province among goldminers, attending for their annual medical examination between July 2015 and December 2016.

Study design and procedures – parent study

In a cohort study, herein described as the parent study, we enrolled miners to identify those who were uninfected with Mtb despite being highly exposed to Mtb, to compare epidemiological factors between Mtb uninfected and infected miners and to collect specimens from Mtb uninfected and infected miners to determine gene expression and immunological profiles associated with being Mtb uninfected in future analysis.

Goldminers, attending OHC for their annual medical examination were pre-screened for the study to identify those aged 33-60 years who had worked in the mining industry for at least 15 years. Following informed consent, a full screen was conducted. Miners were included if they did not have symptoms suggestive of TB, no prior or current history of treatment for active TB disease, no history of or not currently taking isoniazid preventive therapy, no silicosis, had body mass index (BMI) >18.5, no serious medical conditions, HIV negative and no current treatment for cancer, no treatment with steroid tablets, inhalers or injections.

Blood samples were collected intravenously by trained professional phlebotomy nurses amongst those who met the inclusion criteria for QFT (QFT-Plus and QFT-GIT; Qiagen, Hilden Germany), peripheral blood mononuclear cell (PBMC), and transcriptomic (PAXgene) testing. Under the first version of the protocol (enrolments from 10 July 2015 to 29 October 2015), participants gave blood samples for QFT at enrolment and were followed up at 90 days where QFT was repeated and TST (RT-23 in Tween-80, Statens Serum Institute, Copenhagen, Denmark, catalogue numbers: XI1177KL and SP18411C) was placed and read by a professional nurses who were trained in placing and reading TST. The protocol was subsequently simplified and under the second version (enrolments after 29 October 2015) participants gave blood samples for QFT at enrolment and TST placed either at enrolment or seven days later. All participants, regardless of protocol version, were followed up to 12 months where QFT and TST were repeated.

A sputum sample was also collected for mycobacterial culture testing (BACTEC MGIT 960 system, BD Diagnostic Systems, Sparks, MD, USA) to exclude subclinical TB. A questionnaire was administered in a private room by a trained research assistant to collect demographic characteristics and information on factors associated with being TB uninfected. Blood samples collected were tested at the Aurum Clinical Research laboratory for LTBI using QFT-GIT and QFT-Plus. A 6 ml of whole blood sample was collected intravenously from each participant into a single lithium heparin tube. Samples were then transported to Aurum Clinical Research laboratory where they were aliquoted into 1 ml tubes: three tubes for QFT-GIT test kit (QFT-GIT nil, QFT-GIT TB, QFT-GIT mitogen with catalogue numbers 0594-0201 and 0594-0501) and four for QFT-Plus (QFT-Plus nil, QFT-Plus TB1, QFT-Plus TB2, QFT-Plus mitogen with catalogue number 622120). Tubes were then placed in a pre-heated 37°C portable incubator for 16-24 hours, within eight hours of collection. Subsequently, samples were centrifuged to separate plasma for same day testing. Alternatively, samples were stored at -80°C for up to 48-72 hours prior testing. Plasma was tested by interferon-gamma enzyme-linked immunosorbent assay (ELISA), performed using Biotek microplate reader model EL x 800 using Gen 5 software. Biotek microplate washer model EL x 508 and Thermostar shaker were also used.

Study design – sub-study

In this sub-study, using cross-sectional data from the parent study, we compared the performance of QFT-Plus with QFT-GIT using QFT and TST measurements all done either at baseline/seven days (participants enrolled under the 2^nd version of protocol) or at 90 days post-enrolment (participants enrolled under the 1^st version of protocol).

Study definitions

QFT-GIT was considered positive if TB1 tested positive; while QFT-Plus was positive if both TB1 and TB2 tested positive or if either TB1 or TB2 tested positive, as per manufacturer's recommendations. TST induration ≥5mm was considered positive. To assess the specific contribution of CD8+ T-cells, we used TB2−TB1 differential values as an indirect estimate. A cut-off value was set at 0.6 in order to reduce the bias of the intrinsic variability of the test⁹.

Statistical methods

All eligible participants from the parent study were included in the analysis. Binary outcomes (positive/negative) from QFT-Plus and QFT-GIT were compared using the percentage agreement and Kappa statistic. TB1 and TB2 responses for those QFT-Plus positive were summarized by QFT-GIT status using median and interquartile range (IQR) and percentage positive (measurement minus nil response>0.35). TST positivity was summarized by QFT-Plus status. Logistic regression was used to identify factors associated with having TB2-TB1 >0.6. Results were summarized using odds ratios (OR) with their corresponding 95% confidence intervals (CI) and p-values. Due to a small number of outcomes a multivariable analysis was not conducted. Data were analyzed using Stata version 15 (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LP.

Ethical statement

The study received ethical clearance from the University of Witwatersrand Human Research Ethics Committee (WHREC Ref: 150217), London School of Hygiene & Tropical Medicine, UK (LSHTM Ethics Ref: 9279), University of Washington, USA (IRB number 33335) and North West Health Research and Ethics Committee (DOH-27-0515-4991). We sought informed consent from all study participants using written informed consent and information sheets available in the most commonly used local languages. Participants who were unable to read or write were asked to make a mark or thumbprint in the presence of a witness. This study was conducted according to Good Clinical Practice guidelines, in accordance with the requirements of the funders and respective ethics committees.

Process flow

We approached 25,627 miners, 17,030 (66.5%) agreed to be pre-screened, of whom 3,534 (20.8%) satisfied the pre-screen criteria and were eligible for full screening, following informed consent (Figure 1). Overall, 2,980 (84.3%) were offered consent, of whom 1,749 (58.7%) consented and 1,231 (41.3%) declined to take part in the study; 554 (15.7%) were not offered consent as they were lost in the OHC queue.

Figure 1. Participant flow chart at enrolment.

*Data from the daily attendance register. Includes repeat attendances during this time period. §Screen out questions were asked in stages; if screened out at a stage no screening was conducted for subsequent stages. Within a stage more than one screen-out reason may apply. Yrs, years; IPT, isoniazid preventive therapy; BMI, body mass index; QFT-Plus, QuantiFERON-TB Gold-Plus; QFT-GIT, QuantiFERON-TB Gold In-Tube assay; TST, tuberculin skin test.

Following the full screen, 349/1749 (20.0%) met the inclusion criteria and were enrolled into the study. Of the 349, 304 had a baseline QFT-Plus and QFT-GIT while 234 had QFT-GIT, QFT-Plus and TST measured within seven days of each other. For the latter, which is restricted to having a TST measurement within seven days of the QFT measurements, participants enrolled under the 1st version of the protocol contribute QFT-Plus and QFT-GIT measurements from their 90-day visit.

Demographic characteristics

Of the 349 participants enrolled into the study, the median age was 48 years (IQR 45, 53 years), median years in the workface was 24 (IQR 18, 28 years) and 98.6% (344) were male (Table 1). Overall, 92.0% (321) were of Black/African ethnicity, 72.5% (253) had a BCG scar present, 66.8% (233) were born in South Africa and a minority lived in a mine hostel (28.7%; 100). Subsamples of the 349 participants who either had (i) a baseline QFT-Plus and QFT-GIT result (n=304) or (ii) QFT-Plus and TST reading (n=234) had similar demographic characteristics to the overall sample¹⁰.

Comparison of QuantiFERON-TB Gold-Plus vs. Gold in-tube

Of the 304 who had QFT-Plus and QFT-GIT results, 214 (70.4%) had a positive QFT-Plus result, 205 (67.4%) were positive on both assays; 83 (27.3%) were negative on both assays, while 16 (5.3%) had discordant results (seven QFT-Plus negative/QFT-GIT positive; nine QFT-Plus positive/QFT-GIT negative) (Table 2). Overall, the agreement between QFT-Plus and QFT-GIT was 94.7% (288/304) and Kappa was 0.87.

Two hundred and thirty-four participants had QFT-Plus, QFT-GIT and TST measured at the same time/or within seven days of each other. Overall, 54.7% (128) had a TST response ≥5mm, 11.1% (26) had a TST response >0mm and <5mm and 34.2% (80) had zero TST response. The median TST response was 11mm (IQR 6-15mm), among those with a TST>0mm. Of the 234, 160 (68.4%) were positive on both assays, of whom 61.3% (98/160) were TST-positive, and 62 (26.5%) were negative on both assays, of whom 40.3% (25/62) TST-positive. Furthermore, 101 (43.2%) were QFT-Plus positive/TST positive; 41 (17.5%) were QFT-Plus negative/TST negative; 65 (27.8%) were QFT-Plus positive/TST negative; and 27 (11.5%) were QFT-Plus negative/TST positive. However, 12 (5.1%) had discordant results [six QFT-GIT positive/QFT-Plus negative, 33.3% (2/6) TST positive; six QFT-GIT negative/QFT-Plus positive, 50.0% (3/6) TST positive]. When analysis was restricted to a subsample of 234, the percentage agreement between QFT-GIT and QFT-Plus was 94.9%, similar to before.

Assessing the specific contribution of CD8+ T-cells

Of the 214 who had positive QFT-Plus results, 202 (94.4%) were positive on TB1, while 205 (95.8%) were positive on TB2 (Table 2). The median (IQR) values for TB1 and TB2 among those who were QFT-Plus positive were 3.06 (1.31, 7.00) and 3.25 (1.53, 8.02), respectively. The median difference between the TB1 and TB2 was 0.00 (-0.17, 0.39). A >0.6 difference between TB2 and TB1 was observed in 16.4% (35/214) of those who were positive on QFT-Plus (Table 2). Only BMI status was found to be associated with TB2-TB1 >0.6; unadjusted OR 1.94 (95% CI: 0.74–5.05) for BMI>30 kg/m² versus BMI between 18.5-24.9 (Table 3).

Study limitations

This sub-analysis was based on a cross-sectional sample and did not include follow-up to confirm the LTBI status. However, in this population of goldminers with long duration of service, equating to prolonged exposure to Mtb infection, it is unlikely that the LTBI status would change over a relatively short period of follow-up. The sample size of 304 was relatively small to make strong inferences. In addition, the study was conducted among HIV-negative goldminers, selected for those most likely to have experienced Mtb exposure, thus the results may not be generalizable. The lack of a gold-standard test for Mtb infection means that where results were discordant, we cannot know which (if either) was correct.