P1.065. Lymph Node Yield as a Prerequisite for Prognostic Superiority of Ratio-Based Nodal Staging in Esophageal Cancer: A Multi-Institutional Validation Study

Abstract

Abstract Topic Esophageal Cancer: Surgical Treatment of Esophageal Cancer Background Lymph node ratio (LNR) and log odds of positive lymph nodes (LODDS) have been proposed as yield-independent alternatives to pN staging for esophageal cancer. However, their prognostic superiority over pN has not been validated across institutions with varying lymph node (LN) yields. We hypothesized that the discriminative advantage of ratio-based metrics is conditional on adequate lymphadenectomy. Methods A total of 2,490 patients who underwent curative-intent esophagectomy across four specialized centers (2008–2020) were included. Centers A–D differed substantially in median LN yield (22–48 nodes). For each center, the prognostic discrimination of pN staging, LNR, and LODDS for overall survival (OS) was quantified using Harrell's C-statistic, with pairwise comparisons performed by DeLong's test. A threshold analysis evaluated the relationship between LN yield and the incremental C-statistic improvement (ΔC) of LODDS over pN across the full range of institutional yields. Cox proportional-hazards regression was used to test for interaction between LODDS and LN yield category. Results LODDS showed statistically significant superiority over pN only at Center A, a Japanese high-volume center with median LN yield of 48 (IQR 38–59) (C-statistic: 0.703 vs. 0.673; ΔC=+0.030; DeLong p=0.023). At Centers B, C, and D—where median LN yield ranged from 22 to 32—LODDS provided no measurable improvement and showed marginal decrements at the two lowest-yield centers (ΔC=−0.002 to −0.006; all p>0.58). Pooled analysis revealed no significant overall benefit (ΔC=+0.006; p=0.18). Threshold analysis identified a critical LN yield of approximately 25–30 nodes, below which LODDS offered no discriminatory advantage (ΔC≤0.001), and above which benefit increased monotonically with yield (ΔC=+0.031 at LNY≥45; interaction p=0.001). These findings remained robust in sensitivity analyses restricted to patients with ≥15 harvested nodes. Conclusion The prognostic superiority of LODDS over pN staging in esophageal cancer is not universal but yield-dependent: it emerges only when lymph node harvest is adequate, with a critical threshold of approximately 25–30 nodes. Ratio-based staging metrics should not be adopted as universal replacements for pN without first ensuring adequate lymphadenectomy quality. These findings have direct implications for the standardization of lymph node staging across institutions and health systems with divergent surgical practices.