Postoperative malnutrition predicts severe chemotherapy toxicity after gastrectomy: evaluation of an inflammation-anemia phenotype.

Abstract

Objectives: Severe toxicity during adjuvant chemotherapy after curative gastrectomy is common. We evaluated whether postoperative malnutrition predicts severe toxicity, whether an inflammation-anemia (IA) composite adds incremental information, and whether a simple risk model offers clinical utility.

Methods: In this single-center prospective cohort, adults received adjuvant chemotherapy after curative gastrectomy. Malnutrition was defined as albumin <35 g/L, BMI <18.5 kg/m², or low skeletal muscle index; IA as CRP ≥1.5 mg/L plus hemoglobin <120 g/L. A ridge-penalized logistic model included age, pathologic stage, regimen (S-1 vs. oxaliplatin-based multi-agent), malnutrition, IA, and a malnutrition × regimen interaction. Nested 5-fold cross-validation and decision curve analysis evaluated model performance and net benefit.

Results: Of 222 screened patients, 219 were analyzed. Severe toxicity occurred in 48/219 (21.9 %), more often with multi-agent therapy than S-1 (23.5 vs. 8.7 %). Malnutrition was strongly associated with toxicity and remained an independent predictor (penalized OR 3.42, 95 % CI 1.75-6.68), whereas IA was not. Out-of-sample AUROC was 0.695, Brier score 0.156, and calibration slope 0.86. At a 0.30 threshold, net benefit was higher for the full model than for base + malnutrition.

Conclusions: Postoperative malnutrition is a key predictor of severe adjuvant chemotherapy toxicity after gastrectomy. This parsimonious model may support risk-adapted supportive care and treatment planning.