Construction of a modified TNM staging system and prediction model based on examined lymph node counts for gastric cancer patients at pathological stage N3.

BACKGROUND

Examined lymph node (ELN) count is a critical factor affecting the number of metastatic lymph nodes (MLNs). The impact of the ELN number on survival and staging remains unclear.

METHODS

This study included 4,291 stage N3 GC patients from the SEER database (training cohort) and 567 stage N3 GC patients from the FAHZZU database (validation cohort). The optimal ELN count and stage migration were investigated, and a modified TNM (mTNM) staging system including the ELN count was proposed. LASSO regression and random forest analyses were used to screen and evaluate the variables associated with survival, and an mTNM-based nomogram was constructed. The performance of the mTNM staging system and mTNM-based nomogram were compared with that of the 8th edition of the TNM staging system.

RESULTS

The optimal threshold of the ELN count was identified as 21. An insufficient number of ELNs (≤ 21) was associated with poorer survival outcomes and led to stage migration in all N3 patients. A new mTNM staging system was proposed, integrating the ELN count into the TNM staging system (8th edition). LASSO regression analysis revealed that age, tumor size, adjuvant chemotherapy, adjuvant radiotherapy, and the mTNM system were associated with overall survival (OS) outcomes, and random forest analysis revealed that the mTNM system was the most important variable for predicting survival. An mTNM-based nomogram was constructed to predict 1-, 3-, and 5-year OS rates. Compared with the TNM staging system (8th edition), the mTNM staging system and mTNM-based nomogram showed superior prognosis discriminative ability, better predictive accuracy, and greater net improvement in survival outcomes.

CONCLUSIONS

The optimal ELN count for N3 GC patients was 21. The mTNM staging system and mTNM-based nomogram showed superior discriminative ability, predictive accuracy, and greater net benefit for OS outcomes.