Endobronchial ultrasound-guided transbronchial needle aspiration for diagnosis of sarcoidosis in clinically unselected study populations

Literature suggests that ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) has excellent performance characteristics for diagnosis of sarcoidosis. However, many authors challenge the external validity of EBUS-TBNA results, as most studies were performed in referral centres by highly experienced investigators, and included populations with very high sarcoidosis prevalence. We performed a systematic review and meta-analysis to estimate the role of EBUS-TBNA for diagnosis of sarcoidosis in studies enrolling consecutive patients with lymphadenopathy detected at imaging studies, regardless of the suspected underlying clinical aetiology. The Pubmed, Embase, Cinahl, Web of Science and Cochrane Library databases were screened to identify the pertinent literature. Quality of eligible studies was assessed by Quality Assessment, Data Abstraction and Synthesis-2 criteria. Pooled diagnostic yield, sensitivity and specificity were calculated, and a summary receiver operating characteristic curve was constructed. Subgroup analysis was planned to identify possible sources of study heterogeneity. Fourteen studies, collectively involving 2097 patients, fulfilled eligibility criteria. The median prevalence of sarcoidosis was 15%. EBUS-TBNA had a pooled diagnostic yield of 0.79 (standard deviation, 0.24), a pooled sensitivity of 0.84 (95% confidence interval (CI), 0.79-0.88) and a pooled specificity of 1.00 (95% CI, 0.99-1.00). Only subgroup analysis exploring the influence of study design seemed to influence the observed inter-study heterogeneity for sensitivity, retrospective studies showing worst sensitivity than prospective ones. The results of EBUS-TBNA for diagnosis of sarcoidosis in clinically unselected populations are excellent and compare favourably with published results from studies conducted in selected populations. High-quality trials would be needed to evaluate factors possibly explaining the observed heterogeneity in sensitivity.