Interferon release assay for the diagnosis of uveitis associated with tuberculosis: a Bayesian evaluation in the absence of a gold standard

Aim To evaluate an interferon release assay in the diagnosis of uveitis associated with tuberculosis (TB), in the absence of a gold standard. Methods We conducted a prospective 1-year study on consecutive patients with a new diagnosis of uveitis presenting with ocular signs suggestive of TB-associated uveitis (TAU). All subjects underwent ocular and systemic review with investigations, including T-SPOT.TB assay and tuberculin skin test (TST). The main outcome measures were sensitivity and specificity; positive and negative predictive values (PPV, NPV) of T-SPOT.TB were estimated using Bayesian statistics (presented with 95% Bayesian credible intervals (CrI)). Results Analysis after exclusion of patients with indeterminate results and other diseases included 191 patients, with mean age of 47.2 +/- 15.3years. The majority were Chinese (n=110, 57.6%) and women (n=102, 53.4%). The TST (0.70, 0.65 to 0.75) was more sensitive compared to T-SPOT.TB (0.53, 0.42 to 0.66) but the T-SPOT.TB (0.91, 0.88 to 0.93) was more specific than TST (0.71, 0.58 to 0.83). The T-SPOT.TB had a higher PPV (0.88, 0.80 to 0.93) compared to the TST (0.76, 0.57 to 0.90), but a lower NPV (0.60, 0.38 to 0.78) than the TST (0.65, 0.46 to 0.78). Patients who are TST and T-SPOT.TB positive are very likely (95.1% likelihood) to have TAU (39/41 95% CrI 35 to 41). Based on statistical decision theory, in populations where the prevalence of TAU is high the TST should be the first-choice test; the T-SPOT.TB should be performed if the prevalence of TB is low. Conclusions The T-SPOT.TB assay is more specific but less sensitive than the TST and should be used in preference to the TST in low-TB-prevalence populations. Used in conjunction, the likelihood of TAU is greatest if the T-SPOT.TB and TST are both positive.