Dynamic, quantitative ESAT6-CFP10 skin test for tuberculosis risk prediction: a large-scale, multi-center,...

Predicting which individuals with M. tuberculosis (M. tb) infection will progress to active disease remains challenging. We evaluated whether a pragmatic serial, quantitative ESAT6-CFP10 (EC) antigen-based skin test strategy improves risk stratification for targeted prevention in a large-scale prospective cohort.

We enrolled 73,761 contacts identified during school-based tuberculosis outbreaks in Jiangsu, China, from 2020 to 2024. Participants underwent baseline EC test, chest radiography, and symptom screening. EC-negative individuals were retested after 8-12 weeks. Incident tuberculosis was identified through active surveillance and registry linkage. We compared single vs. serial test strategies using Cox models, receiver operating characteristic (ROC) curves and precision-recall (PR) curves.

Among 73,761 close contacts, 108 had prevalent tuberculosis and 190 developed incident cases (overall incidence 151.2 per 100,000 person-years). EC response size predicted incident tuberculosis in a steep, dose-dependent manner. Each 1-mm increase in the maximum response diameter was associated with a 7% higher hazard. The serial test strategy, utilizing the maximum response from two measurements, substantially outperformed single test [C-statistic: 0.806 vs. 0.722]. At the ≥5 mm threshold, this combined strategy yielded a sensitivity of 65.0% and specificity of 96.1%. The subgroup of recent converters had a PPV of 3.4% (95% CI: 2.8-4.1), corresponding to an NNT of approximately 30, and a hazard ratio (HR) of 45.12 (95% CI: 32.50-62.63). Preventive treatment completion was strongly protective (aHR 0.17; 95% CI: 0.11-0.25).

The "test twice, take maximum" EC strategy provides superior risk stratification for tuberculosis prevention. This approach identifies high-risk contacts for targeted intervention. Despite limited sensitivity, these results suggest that quantitative EC skin testing can provide a practical alternative for programmatic risk stratification. In settings where IGRAs are constrained by cost or infrastructure, this approach may enable more efficient targeting of preventive treatment.

National Natural Science Foundation of China (82504476, 82473693, 82574173); Jiangsu Province Preventive Medicine Research Project (Ym2023039); The Special Scientific Research Project for Talent Introduction of the First Affiliated Hospital of Wannan Medical College (KY2960YR2530); Jiangsu Province Postgraduate Research and Innovation Project (KYCX24_2061).