Background Reliable biomarkers that predict treatment response and long-term outcomes in neoadjuvant-treated esophageal squamous cell carcinoma (ESCC) remain limited. The Gustave Roussy Immune (GRIm) score and the Controlling Nutritional Status (CONUT) score respectively reflect systemic inflammation and nutritional status. This study evaluated whether integrating these indices (GRIm–CONUT) improves prognostic prediction in ESCC. Methods We retrospectively analyzed 216 patients with resectable ESCC who received neoadjuvant chemotherapy, chemoradiotherapy, or chemoimmunotherapy followed by curative (R0) esophagectomy between June 2016 and December 2021. GRIm and CONUT scores were calculated from pre-treatment laboratory parameters, and patients were classified into three GRIm–CONUT categories (0, 1, 2). Logistic regression identified independent predictors of pathological complete response (pCR). Overall survival (OS) and recurrence-free survival (RFS) were assessed using Kaplan–Meier and Cox proportional hazards models. A nomogram incorporating GRIm–CONUT and ypTNM staging was developed and validated using bootstrap resampling. Results Higher GRIm–CONUT scores were significantly associated with lower pCR rates (p 0.01) and poorer OS and RFS (all p 0.001). Multivariate logistic regression confirmed GRIm–CONUT score, cN stage, and neoadjuvant regimen as independent predictors of pCR. In multivariate Cox analysis, GRIm–CONUT and ypN stage remained independent predictors of OS, while GRIm–CONUT, ypT stage, and ypN stage independently predicted RFS. A GRIm–CONUT–based nomogram demonstrated superior discrimination (C-index 0.717 vs. 0.659 for ypTNM) and offered greater clinical net benefit in decision-curve analysis. Conclusions The GRIm–CONUT composite score is an independent predictor of pCR, OS, and RFS in ESCC patients undergoing neoadjuvant therapy and surgery. As an inexpensive and readily obtainable biomarker, it enables more accurate prognostic stratification and may support personalized perioperative management. Prospective multicenter validation is warranted.
Xie et al. (Thu,) studied this question.