Abstract
OBJECTIVE: To evaluate the therapeutic efficacy of adding cindilizumab to the Xeloda-Irinotecan (XELIRI) regimen in patients with advanced colorectal cancer and to identify clinical and molecular biomarkers predictive of treatment response. METHODS: A retrospective analysis was conducted on 197 patients with advanced colorectal carcinoma treated between January 2019 and June 2023. Patients were divided into two cohorts: the standard treatment group receiving XELIRI alone (n=103) and the combined treatment group receiving XELIRI with cindilizumab (n=94). Treatment response was assessed according to RECIST criteria and classified as responsive (complete response [CR] or partial response [PR]) or non-responsive (stable disease [SD] or progressive disease [PD]). Logistic regression analysis was performed to identify independent predictors of treatment response. Adverse events were recorded throughout the treatment course. RESULTS: The experimental cohort demonstrated statistically higher objective response rate (ORR) and disease control rate (DCR) compared to the standard treatment cohort (ORR: 38.30% versus 22.33%, P=0.015; DCR: 80.85% versus 66.02%, P=0.019). The incidence of hypothyroidism and renal impairment was significantly higher in the combination group (P=0.002). Logistic regression identified carcinoembryonic antigen (CEA) (OR=1.336, P<0.001), tumor diameter (OR=2.818, P=0.001), KRAS/NRAS gene status (OR=6.229, P=0.001), and treatment regimen (OR=0.079, P<0.001) as independent predictors of treatment response. Receiver operating characteristic (ROC) curve analysis showed that their combined prediction significantly improved predictive efficacy (AUC=0.881), with high sensitivity and specificity. CONCLUSION: Cindilizumab combined with XELIRI regimen improves ORR and DCR in patients with advanced colorectal cancer but may increase the risk of hypothyroidism and renal impairment. CEA, tumor diameter, KRAS/NRAS gene, and treatment regimen are independent predictors of treatment response. The combined predictive model demonstrates robust diagnostic performance.