Development of a nomogram predicting perineural invasion risk and assessment of the prognostic value of perineural invasion in colon cancer: a population study based on the Surveillance, Epidemiology, and End Results database.

Abstract

Background: Perineural invasion (PNI) in colon cancer (CC) is widely associated with poor prognosis. In this study, we aimed to develop a predictive model for PNI and to assess its prognostic value in CC patients.

Methods: Data for CC patients with or without PNI were obtained from the Surveillance, Epidemiology, and End Results (SEER) database from 2010 to 2015. Potential features were selected by stepwise logistic regression, and multivariate logistic regression was used to develop the nomogram. Nomogram performance was assessed based on its calibration curve, discrimination ability and clinical utility. The prognostic value of PNI was assessed using Kaplan-Meier analysis, a competing risk model, and a Fine-Gray multivariable regression model.

Results: A total of 51,826 subjects were included in the study. The nomogram consisted of 11 features was constructed, which provided good calibration and discrimination with area under the curve values of 0.787 vs. 0.781 (development cohort vs. validation cohort). Patients with PNI had worse CC-specific survival (P<0.001) and a higher CC-specific death rate (Gray's test, P<0.001) than patients without PNI. Fine-Gray multivariable regression analysis showed that patients with PNI had a higher CC-specific death rate than patients without PNI [hazard ratio (HR) =1.243; 95% confidence interval (CI): 1.183-1.305; P<0.001]. Pathologic stage T4 (pT4) CC patients without PNI treated with chemotherapy (ChemT) plus radiotherapy (RT) had a lower CC-specific death rate than ChemT-treated or non-therapy patients.

Conclusions: The nomogram developed herein has certain clinical application value for predicting PNI risk in CC patients. PNI is a survival predictor for CC patients. pT4 patients without PNI might benefit from combined ChemT and RT.