Use of an interferon-γ (IFN-γ) release assay or tuberculin skin test for detection and management of latent tuberculosis infection is controversial. For both types of test, we assessed their predictive value for the progression of latent infection to active tuberculosis disease, the targeting value of preventive treatment, and the necessity of dual testing.

In this systematic review and meta-analysis, we searched PubMed, Embase, Web of Science, and the Cochrane Library, with no start date or language restrictions, on Oct 18, 2019, using the keywords (“latent tuberculosis” OR “latent tuberculosis infection” OR “LTBI”) AND (“interferon gamma release assays” OR “Interferon-gamma Release Test” OR “IGRA” OR “QuantiFERON®-TB in tube” OR “QFT” OR “T-SPOT.TB”) AND (“tuberculin skin test” OR “tuberculin test” OR “Mantoux test” OR “TST”). We included articles that used a cohort study design; included information that individuals with latent tuberculosis infection detected by IFN-γ release assay, tuberculin skin test, or both, progressed to active tuberculosis; reported information about treatment; and were limited to high-risk populations. We excluded studies that included patients with active or suspected tuberculosis at baseline, evaluated a non-commercial IFN-γ release assay, and had follow-up of less than 1 year. We extracted study details (study design, population investigated, tests used, follow-up period) and the number of individuals observed at baseline, who progressed to active tuberculosis, and who were treated. We then calculated the pooled risk ratio (RR) for disease progression, positive predictive value (PPV), and negative predictive value (NPV) of IFN-γ release assay versus tuberculin skin test.

We identified 1823 potentially eligible studies after exclusion of duplicates, of which 256 were eligible for full-text screening. From this screening, 40 studies (50 592 individuals in 41 cohorts) were identified as eligible and included in our meta-analysis. Pooled RR for the rate of disease progression in untreated individuals who were positive by IFN-γ release assay versus those were negative was 9·35 (95% CI 6·48–13·49) compared with 4·24 (3·30–5·46) for tuberculin skin test. Pooled PPV for IFN-γ release assay was 4·5% (95% CI 3·3–5·8) compared with 2·3% (1·5–3·1) for tuberculin skin test. Pooled NPV for IFN-γ release assay was 99·7% (99·5–99·8) compared with 99·3% (99·0–99·5) for tuberculin skin test. Pooled RR for rates of disease progression in individuals positive by IFN-γ release assay who were untreated versus those who were treated was 3·09 (95% CI 2·08–4·60) compared with 1·11 (0·69–1·79) for the same populations who were positive by tuberculin skin test. Pooled proportion of disease progression for individuals who were positive by IFN-γ release assay and tuberculin skin test was 6·1 (95% CI 2·3–11·5). Pooled RR for rates of disease progression in individuals who were positive by IFN-γ release assay and tuberculin skin test who were untreated versus those who were treated was 7·84 (95% CI 4·44–13·83).

IFN-γ release assays have a better predictive ability than tuberculin skin tests. Individuals who are positive by IFN-γ release assay might benefit from preventive treatment, but those who are positive by tuberculin skin test probably will not. Dual testing might improve detection, but further confirmation is needed.

National Natural Science Foundation of China and Natural Foundation of Yunnan Province.