Development and validation of a preoperative systemic inflammation-based nomogram for predicting surgical site infection in patients with colorectal cancer

Surgical site infection (SSI) represents a significant postoperative complication in colorectal cancer (CRC). Identifying associated factors is therefore critical. We evaluated the predictive value of clinicopathological features and inflammation-based prognostic scores (IBPSs) for SSI occurrence in CRC patients.

We retrospectively analyzed data from 1445 CRC patients who underwent resection surgery at Wuhan Union Hospital between January 2015 and December 2018. We applied two algorithms, least absolute shrinkage and selector operation (LASSO) and support vector machine-recursive feature elimination (SVM-RFE), to identify key predictors. Participants were randomly divided into training (n = 1043) and validation (n = 402) cohorts. A nomogram was constructed to estimate SSI risk, and its performance was assessed by calibration, discrimination, and clinical utility.

Combining the 30 clinicopathological features identified by LASSO and SVM-RFE, we pinpointed seven variables as optimal predictors for a pathology-based nomogram: obstruction, dNLR, ALB, HGB, ALT, CA199, and CA125. The model demonstrated strong calibration and discrimination, with an area under the curve (AUC) of 0.838 (95% CI 0.799–0.876) in the training cohort and 0.793 (95% CI 0.732–0.865) in the validation cohort. Decision curve analysis (DCA) showed that our models provided greater predictive benefit than individual clinical markers.

The model based on simplified clinicopathological features in combination with IBPSs is useful in predicting SSI for CRC patients.