BACKGROUND: Few models about the personalized prognosis evaluation of buccal mucosa cancer (BMC) patients were reported. We aimed to establish predictive models to forecast the prognosis of BMC patients.
METHODS: The complete clinicopathological information of BMC patients from the surveillance, epidemiology and end results program was collected and reviewed retrospectively. Two nomograms were established and validated to predict long-term overall survival (OS) and cancer-specific survival (CSS) of BMC patients based on multivariate Cox regression survival analysis.
RESULTS: 1155 patients were included. 693 and 462 patients were distributed into modeling and validation groups with 6:4 split-ratio via a random split-sample method. Based on the survival analysis, independent prognostic risk factors (variables that can be used to estimate disease recovery and relapse chance) influencing OS and CSS were obtained to establish nomograms. Then, we divided the modeling group into high- and low-risk cohorts. The low-risk cohort had improved OS and CSS compared to the high-risk cohort, which was statistically significant after the Log-rank test (p < 0.05). Furthermore, we used the concordance index (C-index), calibration curve to validate the nomograms, showing high accuracy. The decision curve analyses (DCA) revealed that the nomograms had evident clinical value.
CONCLUSIONS: We constructed two credible nomogram models, which would give the surgeons reference to provide an individualized assessment of BMC patients.