Derivation of a Modified Novel Pathologic Nodal Classification System for Non-Small Cell Lung Cancer

Abstract

<h3>Purpose/Objective(s)</h3> For decades, nodal staging for non-small cell lung cancer (NSCLC) has been based predominantly on location. Given increasing evidence of the prognostic power of quantitative nodal burden across solid tumors, it is possible that positive lymph node (LN+) count is a primary driver of survival among surgically staged patients with NSCLC. <h3>Materials/Methods</h3> Retrospective cohort analysis was performed on patients with NSCLC undergoing upfront surgery and nodal evaluation, excluding those receiving neoadjuvant chemotherapy in the National Cancer Database (NCDB). Tumor stage was recoded to conform to the current American Joint Committee on Cancer, 8^th Edition (AJCC 8E) based on size and tumor extension. The association between number of positive LNs and survival was modeled with restricted cubic splines in multivariable Cox regression analysis. Recursive partitioning analysis (RPA) was used to develop a modified nodal classification system based on nodal factors independently associated with survival. The proposed nodal classification system was independently validated in patients with NSCLC, now allowing for patients who received neoadjuvant chemotherapy ± radiation, from Surveillance, Epidemiology, and End Results-18 (SEER-18). <h3>Results</h3> Overall, 126,473 patients (62,428 men [49.4%]; mean [SD] age, 67.1 [9.9] years; 32,958 node-positive patients) were included in the NCDB discovery cohort. Mortality risk increased continuously with increasing number of positive LNs in multivariable analysis, with the greatest correlation with the first 3 LN+ (0-3 LN+: HR per LN, 1.38; 95% CI, 1.36-1.41; P<0.001; ≥4 LN+: HR per LN, 1.06; 95% CI, 1.05-1.07; P<0.001). Location of LN (ipsilateral mediastinum) was also a predictor of survival (HR, 1.30; 95% CI, 1.24-1.36; P<0.001). Univariate RPA identified six clusters of patients stratified only by nodal count and not by LN location. On multivariable analysis, these groups demonstrated continuously increasing and non-overlapping mortality risk: N0 (0 LN; reference); N1a (1 LN+; HR, 1.68; 95% CI, 1.61-1.75); N1b (2-3 LN+; HR 1.90; 95% CI, 1.82-1.99); N2a (4 LN+; HR, 2.19; 95% CI, 2.03-2.37); N2b (5-8 LN+; HR, 2.51; 95% CI, 2.35-2.68); N3 (≥9 LN+; HR, 3.35; 95% CI, 3.04-3.70). By contrast, the AJCC 8E nodal classification system spanned a more limited range of mortality risk: N0 (0 LN; reference); N1 (HR, 1.24; 95% CI, 1.09-1.42); N2 (HR, 1.62; 95% CI, 1.41-1.86); N3 (HR, 1.11; 95% CI, 0.75-1.64). The proposed nodal system was validated for overall survival (log-rank P<0.001) and cause-specific survival (k-sample P<0.001) among 112,153 patients with or without neoadjuvant treatment from the SEER-18 validation cohort. <h3>Conclusion</h3> Number of positive LNs is a dominant predictor of outcomes in patients with NSCLC undergoing surgical resection and nodal evaluation, whereas LN location has more limited independent prognostic information in this context. Thus, nodal count should serve as a backbone for pathologic nodal staging in patients with NSCLC.