Log odds of positive lymph nodes may predict survival benefit in patients with node-positive non-small cell lung cancer

The number of positive lymph nodes (npLNs) and the lymph node ratio [LNR; npLNs/number of resected LNs] are useful for predicting survival among patients with non-small cell lung cancer (NSCLC). Here we compared the relative effectiveness of npLNs, LNR, and the log odds of positive lymph nodes (LODDS) to predict overall survival (OS) and cancer-specific survival (CSS) among patients with node-positive NSCLC.We identified 5289 patients with NSCLC and lymph node involvement who had lobectomy or pneumonectomy in 2010-2013 from the Surveillance Epidemiology and End Results (SEER) database. Potential associations between npLNs, LNR, and LODDS with overall survival (OS) and cancer-specific survival (CSS) were assessed with Cox regression analysis. The goodness of fit of npLNs, LNR, and LODDS was compared with the -2 log-likelihood ratio (-2LLR) and by differences in Akaike's information criterion scores (ΔAIC). Tree-based recursive partitioning was applied to split ratio-based variables (LNR and LODDS) into low- and high-risk groups. Kaplan-Meier actuarial estimates of OS and CSS in the various npLNs, LNR, and LODDS subgroups were compared with log-rank tests.Of 5289 patients, 2297 (43.3%) had <10 LNs retrieved and 2992 (56.6%) had ≥10 LNs harvested. Multivariate Cox analysis adjusted for significant factors indicated that LODDS, npLNs, and LNR were independent risk factors for OS and CSS. A LODDS model had the best fit compared with LNR or npLN models in predicting OS and CSS (P < 0.001, ΔAIC = 0). LODDS was slightly superior to LNR for patients with <10 resected LNs, and LNR was slightly superior to LODDS for patients with ≥10 resected LNs (P < 0.001). Higher LODDS was associated with worse OS and worse CSS (log-rank P for both <0.001). LODDS and LNR staging schemes outperformed those of npLNs for predicting OS and CSS.