Single-gene transcripts for subclinical tuberculosis: an individual participant data meta-analysis.

Summary

Across seven datasets (6544 samples), five single-gene transcripts (BATF2, FCGR1A/B, ANKRD22, GBP2, SERPING1) achieved AUCs of 0.75–0.77, equivalent to the best multi-gene signature for detecting subclinical TB over 12 months. Single-gene tests showed setting-invariant sensitivity and specificity, and decision curve analysis indicated higher net benefit than IGRAs in high-burden settings, while combined testing offered the highest benefit in low-burden settings.

Key Findings

• Five single-gene transcripts (BATF2, FCGR1A/B, ANKRD22, GBP2, SERPING1) matched the best multi-gene signature with AUC 0.75–0.77 over 12 months.
• IGRA specificity was low in high-burden settings (about 32%), whereas single-gene tests maintained more consistent sensitivity and specificity across settings.
• Decision curve analysis favored single-gene testing in high-burden settings and combined IGRA+single-gene testing in low-burden settings for preventive therapy stratification.
• No transcript met WHO minimum target product profile for TB progression tests, underscoring room for improvement.

Clinical Implications

In high-burden settings, single-gene blood transcript tests may better guide preventive therapy than IGRAs; in low-burden settings, combining IGRAs with single-gene tests maximizes net benefit when aiming to treat fewer individuals per prevented case.

Limitations

• Despite strong performance, signatures did not meet WHO TPP for TB progression tests.
• Heterogeneity in source datasets and observational designs may introduce spectrum bias.

Future Directions

Prospective, implementation-focused trials in varied burden settings to refine thresholds, evaluate combination strategies with IGRA, and assess real-world impact on preventive therapy outcomes.