Conversion or Reversion of Interferon γ Release Assays for Mycobacterium tuberculosis Infection: A Systematic Review and Meta-analysis

Background Interferon γ release assays (IGRAs) are widely used for diagnosis of latent tuberculosis infection. However, with repeated testing, IGRA transformation (conversion or reversion) may be detected and is challenging to interpret. We reviewed the frequency of and risk factors for IGRA transformation. Methods We screened public databases for studies of human participants that reported the frequency of IGRA transformation. We extracted study and participant characteristics, details of IGRA testing and results. We calculated the pooled frequency of IGRA transformation (and transient transformation) and examined associated risk factors. Results The pooled frequency of IGRA conversion or reversion from 244 studies was estimated at 7.3% (95% confidence interval [CI], 6.1%-8.5%) or 22.8% (20.1%-25.7%), respectively. Transient conversion or reversion were estimated at 46.0% (95% CI, 35.7%-56.4%) or 19.6% (9.2%-31.7%) of conversion or reversion events respectively. Indeterminate results seldom reverted to positive (1.2% [95% CI, .1%-3.5%]). IGRA results in the borderline-positive or borderline-negative range were associated with increased risk of conversion or reversion (pooled odds ratio [OR] for conversion, 4.15 [95% CI, 3.00-5.30]; pooled OR for reversion, 4.06 [3.07-5.06]). BCG vaccination was associated with decreased risk of conversion (OR, 0.70 [95% CI, .56-.84]), cigarette smoking with decreased risk of reversion (0.44 [.06-.82]), and female sex with decreased risk of either conversion or reversion (OR for conversion, 0.66 [.58-.75]; OR for reversion, 0.46 [.31-.61]). Conclusions IGRA conversion is less common than reversion, and frequently transient. Research is needed to determine whether individuals with reversion would benefit from tuberculosis-preventive treatment. Retesting of people with indeterminate results is probably not indicated, because indeterminate results seldom revert to positive.