Abstract
Background: Mediastinal staging in non-small cell lung cancer with endosonography (EUS-FNA plus EBUS-TBNA) followed by mediastinoscopy is more sensitive to detect nodal metastasis as compared to mediastinoscopy alone (ASTER trial, JAMA 2010;304:2245). However 11 patients need to undergo a mediastinoscopy to detect one with N2/3 missed by endosonography. We analysed if lymph node size measured on CT scan can identify patients in whom the mediastinoscopy can be omitted.
Methods: In ASTER, 123 patients were randomized to endosonography followed by mediastinoscopy if the former did not show mediastinal metastasis. Sensitivity, negative predictive value (NPV) and number of mediastinoscopies needed to detect one false negative endosonography were calculated in the cases with complete data (n=120; 98%).
Results: With CT, 74 patients had enlarged mediastinal nodes (≥10mm), the prevalence of N2/3 was 65 (54-75)%. Sensitivity and NPV of endosonography was 86 (74-93) and 77 (60-88)%. Adding mediastinoscopy increased sensitivity and NPV to 96 (87-99) and 93 (77-98)%. In the 46 patients without enlarged mediastinal nodes (N0/1) the prevalence of N2/3 was 39 (26-54)%. Sensitivity and NPV of endosonography was 89 (67-97) and 93 (79-98)%. Adding mediastinoscopy did not improve this. In patients with enlarged vs normal-sized mediastinal nodes, the number of mediastinoscopies needed to detect one false negative endosonography is 6 vs. infinite (p=0.026).
Conclusions: A negative endosonography should be followed by a mediastinoscopy in patients with enlarged mediastinal nodes on CT. In the absence of enlarged nodes, a mediastinoscopy following a negative endosonography can be omitted.
- © 2011 ERS