Abstract
Lymph node ratio (LNR) and log odds of positive lymph nodes (LOODS) are considered to be potentially more rigorous metastatic lymph node stratification patterns than AJCC N stage due to the integration of the number of examined lymph nodes. Based on the data of colorectal cancer obtained from the Surveillance, Epidemiology, and End Result (SEER) database, comparison of the stratification performance between different prognostic models was validated by statistical methods, includingchi-square test, Cox regression analysis, Kaplan-Meier survival analysis, and Time-dependent Area Under Curve. A prognostic outcome-oriented statistical method based on Log-rank test was performed to determine the cutoff value of LNR. External validation cohort obtained from Zhejiang Cancer Hospital was used to validate the model performance and cutoff value. In the SEER database cohort, the optimal cutoff value for adequate or inadequate lymph node clearance was 14. In performance comparisons, LNR performance was superior to lymph node count alone and similar to LODDS. The optimal cutoff values for LNR were analyzed to be 0.11 (P < 0.001), 0.39 (P < 0.001) and 0.68 (P < 0.001), respectively. The stratification performance of modified N stage phased with the above cutoff value outperforms AJCC N stage. Kaplan-Meier survival analysis of the differences in patient distribution between the two staging systems also demonstrated the excellence of the LNR system. Above results were verified in the external verification cohort. The modified N stage based on LNR has better stratification performance than the AJCC N stage.