An albumin-based immune-nutritional score for predicting complete response, organ preservation, and toxicity after neoadjuvant PD-1-based therapy in low rectal cancer.

BACKGROUND: Programmed death-1 (PD-1) inhibitors are increasingly used in neoadjuvant therapy for low rectal cancer, but complete response and toxicity remain heterogeneous. We developed and validated an albumin-based immune-nutritional score (ABINS) to capture host factors not reflected by tumor biomarkers. METHODS: This multicenter retrospective study included 480 consecutive patients with low rectal adenocarcinoma (≤5 cm from the anal verge) treated with neoadjuvant PD-1-based therapy at three tertiary centers from 2018 to 2023. Patients were assigned to a training cohort (n=276), internal validation cohort (n=87), and geographically separate external validation cohort (n=117). ABINS assigned one point each for albumin <35 g/L, lymphocyte count <1.0×109/L, C-reactive protein >10 mg/L, total cholesterol <3.6 mmol/L, and sex-specific anemia, yielding favorable (0-1), intermediate (2), and poor (3-5) categories. The primary endpoint was composite complete response, defined as sustained clinical complete response under watch-and-wait or pathological complete response. Secondary endpoints included organ preservation, toxicity, immune-related adverse events, and disease-free survival. Logistic and Cox regression evaluated associations, and model performance was assessed by AUC, calibration, and Brier score. RESULTS: Among 480 patients, 166 (34.6%) achieved composite complete response, 280 (58.3%) achieved organ preservation, and 225 disease-free survival events occurred over a median follow-up of 32.4 months. ABINS classified 333 patients (69.4%) as favorable, 87 (18.1%) as intermediate, and 60 (12.5%) as poor. Complete response decreased stepwise across these groups (42.9%, 20.7%, and 8.3%; P<0.001), as did organ preservation (66.4%, 51.7%, and 23.3%; P<0.001), whereas grade ≥3 toxicity increased (7.5%, 19.5%, and 23.3%; P<0.001). After multivariable adjustment, each one-point ABINS increase independently predicted lower complete response (OR 0.51, 95% CI 0.41-0.65), higher grade ≥3 toxicity (OR 1.75, 95% CI 1.37-2.23), and worse disease-free survival (HR 1.42, 95% CI 1.27-1.60; all P<0.001). Adding ABINS improved external-cohort AUC from 0.703 to 0.772 and reduced Brier score from 0.190 to 0.161. CONCLUSIONS: ABINS is a simple, inexpensive host immune-nutritional score that improves risk stratification for complete response, organ preservation, toxicity, and survival after neoadjuvant PD-1-based therapy in low rectal cancer. Prospective validation is warranted.