Tuberculin skin test reversion following isoniazid preventive therapy reflects diversity of immune response to primary Mycobacterium tuberculosis infection

Abstract

Rationale: Healthy household contacts (HHC) of individuals with Tuberculosis (TB) with Tuberculin Skin Test (TST) conversions are considered to harbor latent Mycobacterium tuberculosis (Mtb), and at risk for TB. The immunologic, clinical, and public health implications of TST reversions that occur following Isoniazid preventive therapy (IPT) remain controversial.

Objectives: To measure frequency of TST reversion following IPT, and variation in interferon-gamma (IFN-γ) responses to Mtb, in healthy Ugandan TB HHC with primary Mtb infection evidenced by TST conversion.

Methods: Prospective cohort study of healthy, HIV-uninfected, TST-negative TB HHC with TST conversions. Repeat TST was performed 12 months following conversion (3 months following completion of 9 month IPT course) to assess for stable conversion vs. reversion. Whole blood IFN-γ responses to Mtb antigen 85B (MtbA85B) and whole Mtb bacilli (wMtb) were measured in a subset (n = 27 and n = 42, respectively) at enrollment and TST conversion, prior to initiation of IPT.

Results: Of 122 subjects, TST reversion was noted in 25 (20.5%). There were no significant differences in demographic, clinical, or exposure variables between reverters and stable converters. At conversion, reverters had significantly smaller TST compared to stable converters (13.7 mm vs 16.4 mm, respectively; p = 0.003). At enrollment, there were no significant differences in IFN-γ responses to MtbA85B or wMTB between groups. At conversion, stable converters demonstrated significant increases in IFN-γ responses to Ag85B and wMtb compared to enrollment (p = 0.001, p<0.001, respectively), while there were no significant changes among reverters.

Conclusions: TST reversion following IPT is common following primary Mtb infection and associated with unique patterns of Mtb-induced IFN-γ production. We have demonstrated that immune responses to primary Mtb infection are heterogeneous, and submit that prospective longitudinal studies of cell mediated immune responses to Mtb infection be prioritized to identify immune phenotypes protective against development of TB disease.

Figure 1. Longitudinal TST measurements among HHC.

TST were placed at study enrollment and every 3-month course of INH preventive therapy and post-conversion TST was placed 12 months following each subject's initial conversion event. Following measurement of post-conversion TST, subjects were classified as either stable converters or reverters. Shown are mean TST measurements (mm) ± 1 SD for all eligible study participants at study enrollment (baseline), conversion, and post-conversion time points (A). As shown in Panel B (numbers in boxes indicate number of conversions per time point), subjects with stable conversions were more likely to experience their initial conversion soon following study entry (mean time to conversion, 4.1 months) while reverters experienced their initial conversion later (mean time to conversion 7.2 months; p = 0.03).

Figure 2. Longitudinal WB-IGRA results among HHC.

Whole blood was obtained from a subset of eligible subjects at study enrollment (baseline) and time of initial TST conversion. WB-IGRA were performed using Mtb Ag85B (A) and whole Mtb H37Ra (B) as T cell stimulus. Boxes indicate the interquartile ranges, horizontal lines transecting boxes indicate medians, and whiskers indicate highest and lowest values. At study entry (baseline), no significant differences were noted between reverters and stable converters for either stimulus. At the time of conversion, subjects with stable conversion produced significantly more IFN-γ than reverters in response to both Mtb Ag85B and whole Mtb H37Ra (A, B). Moreover, IFN-γ production significantly increased among stable converters between enrollment and conversion, while there were no significant changes noted for reverters. Statistical comparison between groups was performed using Wilcoxon rank sum test, and comparison with groups performed with Wilcoxon sign-rank test (p< 0.05 considered significant).