Background: Guidelines recommend endosonography with fine-needle aspiration for mediastinal nodal staging in non-small-cell lung cancer, but most do not specify whether this should be through endobronchial endoscopy (EBUS), oesophageal endoscopy (EUS), or both. We assessed the added value and diagnostic accuracy of the combined use of EBUS and EUS.
Methods: For this systematic review and random effects meta-analysis, we searched MEDLINE, Embase, BIOSIS Previews, and Web of Science, without language restrictions, for studies published between Jan 1, 2000, and Feb 25, 2016. We included studies that assessed the accuracy of the combined use of EBUS and EUS in detecting mediastinal nodal metastases (N2/N3 disease) in patients with lung cancer. For each included study, we extracted data on the age and sex of participants, inclusion criteria regarding tumour stage on imaging, details of the endoscopic testing protocol, duration of each endoscopic procedure, number of lymph nodes sampled, serious adverse events occurring during the endoscopic procedures, the reference standard, and 2 × 2 tables for EBUS, EUS, and the combined approach. We evaluated the added value (absolute increase in sensitivity and in detection rate) of the combined use of EBUS and EUS in detecting mediastinal nodal metastases over either test alone, and the diagnostic accuracy (sensitivity and negative predictive value) of the combined approach. This study is registered with PROSPERO, number CRD42015019249.
Findings: We identified 2567 unique manuscripts by database search, of which 13 studies (including 2395 patients) were included in the analysis. Median prevalence of N2/N3 disease was 34% (range 23-71). On average, addition of EUS to EBUS increased sensitivity by 0·12 (95% CI 0·08-0·18) and addition of EBUS to EUS increased sensitivity by 0·22 (0·16-0·29). Mean sensitivity of the combined approach was 0·86 (0·81-0·90), and the mean negative predictive value was 0·92 (0·89-0·93). The mean negative predictive value was significantly higher in studies with a prevalence of 34% or less (0·93 [95% CI 0·91-0·95]) compared with studies with a prevalence of more than 34% (0·89 [0·85-0·91]; p=0·013). We found no significant differences in mean sensitivity and negative predictive value between studies that did EBUS first or EUS first, or between studies that used an EBUS-scope or a regular echoendoscope to do EUS.
Interpretation: The combined use of EBUS and EUS significantly improves sensitivity in detecting mediastinal nodal metastases, reducing the need for surgical staging procedures.
Funding: No external funding.
Copyright © 2016 Elsevier Ltd. All rights reserved.