Remnant cholesterol inflammatory index for predicting pathological complete response to neoadjuvant therapy in breast cancer: A multicenter external validation study

Abstract

Background Pathological complete response (pCR) after neoadjuvant therapy (NAT) is associated with long-term outcomes in breast cancer (BC). The remnant cholesterol inflammatory index (RCII) integrates lipid-related and inflammatory components, but its role in pCR prediction remains unclear. Methods This multicenter retrospective study included women with invasive BC treated with NAT between January 2022 and January 2026. Prespecified multivariable logistic regression models incorporating baseline inflammatory, metabolic, and clinicopathologic variables were developed in a training cohort and evaluated in an independent external validation cohort without recalibration. Model performance was assessed using discrimination and calibration metrics. Results Among 422 patients, 157 (37.2%) achieved pCR. In the clinically adjusted model, triple-negative BC (OR 7.56, 95% CI 2.74–20.86), HER2-positive BC (OR 4.13, 95% CI 2.15–7.92), and lower log-RCII (OR 0.48, 95% CI 0.33–0.69) were independently associated with pCR. External validation yielded an AUC of 0.84 (95% CI 0.76–0.91). The calibration slope was 0.94, while the calibration intercept of −0.60 indicated systematic overprediction of pCR probability. Sensitivity and internal validation analyses supported the robustness of the primary findings. Conclusion Log-RCII was independently associated with pCR and provided incremental predictive information beyond standard clinicopathologic factors. The model showed favorable discrimination in the external validation cohort, but systematic overprediction indicates that further prospective validation and potential recalibration are warranted before clinical implementation.