Background: Though lymph node status may predict long-term outcome of patients with non-metastatic colon cancer, discordant findings exist among various expressions of lymph node status. The present study was designed to assess the prognostic value among these lymph node evaluations.
Methods: The analysis was based on surgical patients with newly diagnosed colon adenocarcinoma registered in the Taiwan Cancer Database from 2003 to 2005. Exclusion criteria included those patients who had stage IV disease, those whose survival period was <1 month, or those whose lymph node information was unavailable. Studied variables included total number of lymph nodes (LNT), number of positive lymph nodes (LNP), number of negative lymph nodes (LNN), ratio of positive lymph nodes (LNR), and log odds of positive lymph nodes (LODDS).
Results: Of 16,790 newly diagnosed colon cancer patients, there were 9,644 (65.4 ± 13.5 years; male 54.9 %) patients with non-metastatic disease who met the criteria. Correlation analyses for patients with stage III disease showed that LNR and LODDS were highly correlated, as were LNT and LNN. By the Cox proportional hazard model, LNT was prognostic of long-term survival in patients with stage II disease, while LNR and LNP were the most powerful prognosticators for patients with stage III disease (p < 0.001). Both the receiver operating characteristics curve analysis and area under the curve indicated that LNR had the best discriminating capability to predict 5-year survival (0.704, 0.700, and 0.709 for overall, disease-free, and disease-specific survival, respectively), followed by LODDS.
Conclusions: For patients undergoing resection for colon cancer, LNR, LODDS, and LNP are better prognostic factors for those with stage III disease than LNT is for patients with stage III disease.