Abstract

Background: The reported accuracy of transbronchial needle aspiration (TBNA) for mediastinal staging in non-small cell lung cancer (NSCLC) varies widely. We performed a meta-analysis to estimate the accuracy of TBNA for mediastinal staging in NSCLC.

Methods: Medline, Embase, and the bibliographies of retrieved articles were searched for studies evaluating TBNA accuracy with no language restriction. Meta-analytical methods were used to construct summary receiver-operating characteristic curves and to pool sensitivity and specificity.

Results: Thirteen studies met inclusion criteria, including six studies that surgically confirmed all TBNA results and enrolled at least 10 patients with and without mediastinal metastasis (tier 1). Methodological quality varied but did not affect diagnostic accuracy. In tier 1 studies the median prevalence of mediastinal metastasis was 34%. Using a random effects model, the pooled sensitivity and specificity were 39% (95% CI 17 to 61) and 99% (95% CI 96 to 100), respectively. Compared with tier 1 studies, the median prevalence of mediastinal metastasis (81%; p = 0.002) and pooled sensitivity (78%; 95% CI 71 to 84; p = 0.009) were higher in non-tier 1 studies. Sensitivity analysis confirmed that the sensitivity of TBNA depends critically on the prevalence of mediastinal metastasis. The pooled major complication rate was 0.3% (95% CI 0.01 to 4).

Conclusions: When properly performed, TBNA is highly specific for identifying mediastinal metastasis in patients with NSCLC, but sensitivity depends critically on the study methods and patient population. In populations with a lower prevalence of mediastinal metastasis, the sensitivity of TBNA is much lower than reported in recent lung cancer guidelines.