Combined inflammation-related biomarkers and clinicopathological features for the prognosis of stage II/III colorectal cancer by machine learning

Abstract

Background: Inflammation-related biomarkers, such as systemic inflammation score (SIS) and neutrophil-lymphocyte ratio (NLR), are associated with colorectal cancer prognosis. However, the combined role of SIS, NLR, and clinicopathological factors in stage II/III colorectal cancer remains unclear. This study developed a nomogram to predict long-term prognosis for these patients.

Methods: This retrospective study included 1540 patients (training set) from the First Affiliated Hospital of Kunming Medical University and 152 patients (testing set) from The Honghe Third People's Hospital. Cox regression identified independent prognostic factors, and machine learning established predictive models. Model performance was evaluated by the C-index, area under the curve (AUC), and decision curve analysis (DCA).

Results: In the training set, a total of 1540 patients with stage II/III colorectal cancer were included. More than 70 years old (HR = 1.830, p = 0.000); SIS = 2 (HR = 1.693, p = 0.002); Preoperative CEA more than 5 ng/mL (HR = 1.614, p = 0.000); and Moderately differentiated (HR = 1.438, p = 0.011); or Low/undifferentiated (HR = 2.126, p = 0.000); The pN1 (HR = 2.040, p = 0.000) and pN2 (HR = 3.297, p = 0.000) stages were considered independent prognostic risk factors of stage II/III colorectal cancer. Negative perineural invasion (HR = 0.733, p = 0.014) and NLR less than 4 (HR = 0.696, p = 0.022) were considered independent prognostic protective factors of stage II/III colorectal cancer. A nomogram was established based on SIS, NLR, and the clinicopathological results for predicting and validating the overall survival in the training and testing sets. The C-index of the training set was 0.746, and the C-index of the testing set was 0.708, indicating the high prediction efficiency of the nomogram.

Conclusions: A nomogram combining SIS, NLR, and clinicopathological factors provides an effective, cost-efficient tool for predicting the prognosis of stage II/III colorectal cancer. Future studies will validate its long-term predictive performance in larger, multicenter cohorts.

Keywords: Cliniopathological factors; Colorectal cancer; NLR; Nomogram; SIS.

Fig. 1

Flow chart of the study. Illustrates the selection process of patients included in the study, detailing inclusion and exclusion criteria and the final number of patients analyzed

Fig. 2

Survival prognosis of colorectal cancer. A-G Kaplan-Meier curves of OS according to SIS, NLR, age, CEA, pN stage, perineural invasion and tumor differentiated in the training set from the cohort of first affiliated hospital of kunming medical university(log-rank test: p < 0.05)

Fig. 3

Risk models for training sets. A List of prediction models generated via machine learning and the calculation of the C-index of each model across all validation datasets. B Nomogram model for predicting the 1-, 3-, and 5-year OS rate of stage II/III CRC patients. C ROC curve of stage II/III CRC prognosis in the training set. D.Kaplan-Meier curves of OS according to the median score in the training set (p < 0.05). E. DCA of the stage II/III CRC model

Fig. 4

Validate the risk model of the set. A ROC curve of stage II/III CRC prognosis in the testing set from the cohort of third people's hospital of Honghe prefecture. B Kaplan–Meier curves of OS according to the model median score in the testing set (p < 0.01). C The calibration curve of the testing set for predicting the 1-, 3-, and 5-year OS rates in stage II/III CRC patients