Association of Pretreatment Immune-Inflammatory Biomarkers with Pathological Tumor Regression Following Neoadjuvant...

Background: Predicting tumor regression following neoadjuvant chemoradiotherapy (nCRT) remains a major challenge in the management of locally advanced rectal cancer (LARC). Readily available inflammatory biomarkers may provide useful information regarding treatment response. Methods: This retrospective single-center study included 88 patients with stage II-III rectal adenocarcinoma who underwent neoadjuvant chemoradiotherapy followed by curative-intent surgery between 2017 and 2025. Patients were classified according to the College of American Pathologists tumor regression grading system as CAP 0 (pathological complete response, n = 22), CAP 2 (partial response, n = 50), or CAP 3 (poor/no response, n = 16). Pretreatment C-reactive protein, carcinoembryonic antigen (CEA), neutrophil, platelet, monocyte, and lymphocyte counts, together with NLR, PLR, and PIV, were compared across groups. Receiver operating characteristic and logistic regression analyses were performed for pathological complete response (pCR). Results: None of the evaluated biomarkers differed significantly across CAP groups. The smallest omnibus p-values were observed for neutrophil count (p = 0.052), monocyte count (p = 0.075), and CEA (p = 0.088). Monocyte count showed the highest discriminatory performance for pathological complete response (AUC = 0.663), followed by CEA (AUC = 0.640). In sensitivity analyses adjusted for baseline clinical T stage and receipt of total neoadjuvant therapy, CEA, neutrophil count, and monocyte count were not independently associated with pathological complete response. More favorable tumor regression was associated with lower residual tumor burden and reduced nodal involvement. Conclusions: Pretreatment inflammatory biomarkers showed biologically plausible numerical patterns across tumor regression groups, but their discriminatory and independent predictive performance was limited. These markers should not be considered stand-alone clinical prediction tools and should be validated within larger, multimodal prospective models.