Recursive partitioning staging system based on the log odds of the negative lymph node/T stage ratio in colon mucinous adenocarcinoma

Abstract

Background: This study aimed to investigate the prognostic significance of the log odds of negative lymph nodes/T stage ratio (LONT) and develop an efficient prognostic staging system using LONT in patients with colon mucinous adenocarcinoma (MAC).

Methods: This study included 5,236 patients diagnosed with colon MAC obtained from the Surveillance, Epidemiology, and End Results database. The Kaplan-Meier method, subgroup analysis, receiver operating characteristic (ROC) curve, and Cox proportional hazard regression model were used to determine the clinical outcomes. Recursive partitioning analysis (RPA) was used to develop a novel prognostic system.

Results: The 1-, 3-, and 5-year ROC curves, used to predict cancer-specific survival (CSS) and overall survival (OS), demonstrated that the areas under the ROC curve for LONT were superior to those of pT, pN, and pTNM stages. Additionally, a lower LONT was correlated with worse clinical outcomes. The LONT classification efficiently differentiated the prognosis of patients in terms of OS and CSS. Multivariate Cox analyses revealed that LONT was an independent prognostic factor for both CSS and OS. Based on the pT stage and LONT, a novel prognostic staging system was developed using RPA, demonstrating a good prognostic predictive performance.

Conclusion: A lower LONT was associated with worse survival in patients with colon MAC. The pT stage and LONT-based prognostic staging system facilitated risk stratification in these patients.

Keywords: LONT; colon mucinous adenocarcinoma; prognosis; recursive partitioning analysis; staging system.

Figure 1

ROC curves of 1-, 3-, and 5-year CSS and OS prediction of (A, E) LONT; (B, F) pT stage; (C, G) pN stage; (D, H) pTNM stage. LONT demonstrated superior predictive performance compared to the pT stage, pN stage, and pTNM stage (AUC in the bottom right of the figure). ROC, receiver operating characteristic curve; CSS, cancer-specific survival; OS, overall survival; LONT, the log odds of negative lymph nodes/T stage ratio; SEER, Surveillance, Epidemiology, and End Results; AUC, area under the ROC curve.

Figure 2

Kaplan–Meier curves for (A) CSS and (B) OS according to the LONT classification, and survival differences were observed (P < 0.05). CSS, cancer-specific survival; OS, overall survival; LONT, the log odds of negative lymph nodes/T stage ratio; SEER, Surveillance, Epidemiology, and End Results.

Figure 3

LONT differentiated the presentation of CSS in subgroup analyses. Kaplan–Meier curves evaluating CSS according to LONT classification in (A–D) pT stages I–III, (E–I) pN stages N0–N2b, and (J–L) pTNM stages I–III. CSS, cancer-specific survival; LONT, the log odds of negative lymph nodes/T stage ratio; TNM, tumor-node-metastasis.

Figure 4

LONT differentiated the presentation of OS in the subgroup analyses. Kaplan–Meier curves evaluating OS according to LONT classification in (A–D) pT stages I–III, (E–I) pN stages N0–N2b, and (J–L) pTNM stages I–III. OS, overall survival; LONT, the log odds of negative lymph nodes/T stage ratio; TNM, tumor-node-metastasis.

Figure 5

Development of an RPA staging system for prognostic stratification. (A) RPA for patients with colon MAC. (B) The RPA staging system. MAC, mucinous adenocarcinoma.

Figure 6

The prognostic evaluation capability of the recursive partitioning analysis staging system, and survival differences were observed across staging subgroups. (A) Kaplan–Meier curves evaluating CSS according to the novel staging system. (B) ROC curve of the 1-, 3-, and 5-year CSS prediction of the novel staging system (AUC in the bottom right of the figure). ROC, receiver operating characteristic curve; CSS, cancer-specific survival; AUC, the area under the ROC curve.