Serial T-SPOT.TB responses in Tanzanian adolescents: Transient, persistent and irregular conversions

Abstract

Background: Prospective studies of interferon-gamma release assays (IGRA) on healthy subjects in tuberculosis-endemic regions have not examined the long-term variability of serial assays. This issue is relevant to the interpretation of tuberculosis (TB) vaccine trials based on prevention of infection.

Methods: T-SPOT.TB assays were performed manually on healthy adolescents during a tuberculosis vaccine trial in Tanzania at 5 intervals over 3 years. Assay results were defined as negative, positive, borderline or invalid. Subsequently, microtiter plates were analyzed by an automated reader to obtain quantitative counts of spot forming cells (SFCs) for the present analysis.

Results: 3387 T-SPOT.TB samples were analyzed from 928 adolescents; manual and automated assay results were 97% concordant. Based on the quantitative results 143 (15%) participants were prevalent IGRA-positives at baseline, were ineligible for further study. Among the remaining IGRA-negative participants, the annual rate of IGRA conversion was 2·9%. Among 43 IGRA converters with repeat assays 12 (28%) were persistent converters, 16 (37%) were transient converters, and 15 (35%) comprised a new category defined as irregular converters (≥2 different subsequent results). ESAT-6 and CFP-10 responses were higher in prevalent than incident positives: 53 vs 36 for CFP-10 (p < 0·007); 44 vs 34 for ESAT-6 (p = 0·12).

Conclusions: Definitions of IGRA conversion, reversion, and persistence depend critically on the frequency of testing. Multiple shifts in categories among adolescents in a TB-endemic country may represent multiple infections, variable host responses in subclinical infection, or assay variation. These findings should to be considered in the design and interpretation of TB vaccine trials based on prevention of infection. Household contact studies could determine whether even transient IGRA conversion might represent exposure to an active case of M. tuberculosis disease.

Fig 1. Analysis groups based on quantitative IGRA results.

Among 928 subjects with ≥1 qIGRA result, 161 (17.3%) failed to meet strict criteria for inclusion in an analysis group, including 107 (11.5%) who were negative at baseline, but did not participate in the vaccine study and had no subsequent samples, 23 (2.5%) who had borderline or invalid baseline qIGRA, 19 (2.0%) whose only post-baseline non-negative qIGRA was borderline, and 12 (1.3%) who were excluded from the vaccine study because their baseline mIGRA was borderline or positive, but their qIGRA was not positive.

Fig 2. Distribution of SPCs comparing prevalent positives and first positive in converters.

Prevalent positives (n = 143, red diamonds) and IGRA converters (n = 45, blue squares) at time of first positive IGRA (including both early and late converters as defined in the text). For converters with more than one positive qIGRA, only the results of the first positive qIGRA are included. p-values are calculated using the Mann-Whitney U-test. See text for details of potential impact of vaccine on the ESAT-6 results.