Meta-analysis: new tests for the diagnosis of latent tuberculosis infection: areas of uncertainty and recommendations for research

Abstract

Background: Until recently, the tuberculin skin test was the only test for detecting latent tuberculosis (TB) infection, but 2 ex vivo interferon-gamma release assays (IGRAs) are now commercially licensed.

Purpose: To estimate sensitivity, specificity, and reproducibility of IGRAs (commercial or research versions of QuantiFERON [QFT] and Elispot) for diagnosing latent TB infection in healthy and immune-suppressed persons.

Data sources: The authors searched MEDLINE and reviewed citations of all original articles and reviews for studies published in English.

Study selection: Studies had evaluated IGRAs using Mycobacterium tuberculosis-specific antigens (RD1 antigens) and overnight (16- to 24-h) incubation times. Reference standards had to be clearly defined without knowledge of test results. DATA EXTRACTION AND QUALITY ASSESSMENT: Specific criteria for quality assessment were developed for sensitivity, specificity, and reproducibility.

Data synthesis: When newly diagnosed active TB was used as a surrogate for latent TB infection, sensitivity of all tests was suboptimal, although it was higher with Elispot. No test distinguishes active TB from latent TB. Sensitivity of the tuberculin skin test and IGRAs was similar in persons who were categorized into clinical gradients of exposure. Pooled specificity was 97.7% (95% CI, 96% to 99%) and 92.5% (CI, 86% to 99%) for QFT and for Elispot, respectively. Both assays were more specific than the tuberculin skin test in samples vaccinated with bacille Calmette-Guérin. Elispot was more sensitive than the tuberculin skin test in 3 studies of immune-compromised samples. Discordant tuberculin skin test and IGRA reactions were frequent and largely unexplained, although some may be related to varied definitions of positive test results. Reversion of IGRA results from positive to negative was common in 2 studies in which it was assessed.

Limitations: Most studies used cross-sectional designs with the inherent limitation of no gold standard for latent TB infection, and most involved small samples with a widely varying likelihood of true-positive and false-positive test results. There is insufficient evidence on IGRA performance in children, immune-compromised persons, and the elderly.

Conclusions: New IGRAs show considerable promise and have excellent specificity. Additional studies are needed to better define their performance in high-risk populations and in serial testing. Longitudinal studies are needed to define the predictive value of IGRAs.