**Background:** Cancer cachexia is a multifactorial syndrome characterized by ongoing skeletal muscle loss that cannot be fully reversed by conventional nutritional support. It affects approximately half of advanced cancer patients, particularly those with digestive tract cancers, and is associated with fatigue, functional impairment, treatment toxicity, poor quality of life, and reduced survival. Despite being a direct cause of death for more than 20% of cancer patients, cancer cachexia is often under-recognized in clinical practice. Existing screening tools have limitations: the Cachexia score (CASCO) lacks disease-state questions and its prognostic value is unknown, while commonly used single measures such as BMI or weight loss have low accuracy when used alone. This study aimed to develop and validate a comprehensive cancer cachexia risk score using multiple clinical, nutritional, and oncological variables to identify digestive tract cancer patients at risk before abdominal surgery.
**Methods:** This large-scale retrospective cohort study was conducted at the Department of General Surgery/Shanghai Clinical Nutrition Research Center, Zhongshan Hospital, Fudan University. Patients aged ≥18 years who underwent abdominal surgery for digestive tract cancer (liver, gallbladder, pancreatic, gastric, or colorectal) between January 2015 and December 2020 were included. Exclusion criteria were incomplete clinical data for cachexia diagnosis, emergency surgery, or previous cancer history. Participants from 2015–2019 were randomly allocated to development (n=8,743) and validation (n=5,828) cohorts in a 6:4 ratio. Those from 2020 formed the application cohort (n=1,693). Cancer cachexia was diagnosed per international consensus: weight loss >5% over 6 months, or any weight loss >2% with BMI <20 kg/m² or skeletal muscle depletion (SMI cut-offs: 37.81 cm²/m² for women, 43.13 cm²/m² for men). Candidate predictors included demographic, clinical, nutritional, and oncologic variables. Univariate analysis identified factors with P<0.05, which were entered into binary logistic backward stepwise regression. A nomogram was constructed, and the optimal cutoff (0.18) was determined by ROC analysis. Model performance was assessed using AUC, calibration curves (Hosmer–Lemeshow test), and decision curve analysis. Kaplan–Meier curves and multivariate Cox regression evaluated survival stratification.
**Key Results:** A total of 16,264 patients were included (median age 64 years; 65.9% male). Cancer cachexia was observed in 2,699 individuals (15.7% in development, 15.6% in validation, 24.3% in application cohorts). Seven independent risk factors were identified: cancer site (gallbladder: OR 7.788, 95% CI 5.418–11.193; pancreas: OR 4.984, 95% CI 3.756–6.615; stomach: OR 2.679, 95% CI 2.012–3.568; liver: OR 3.925, 95% CI 2.755–5.592; all vs. colorectum), cancer stage (stage II: OR 1.311, 95% CI 1.100–1.563; stage III: OR 1.959, 95% CI 1.654–2.320; stage IV: OR 2.496, 95% CI 1.915–3.253; all vs. stage I), time from symptom onset to hospitalization (OR 1.143, 95% CI 1.109–1.178), appetite loss (OR 1.589, 95% CI 1.384–1.825), BMI (OR 0.874, 95% CI 0.854–0.895), SMI (OR 0.977, 95% CI 0.968–0.986), and NLR (OR 1.094, 95% CI 1.069–1.119). The risk score demonstrated AUC values of 0.760 (95% CI 0.747–0.774, P<0.001) in development, 0.743 (95% CI 0.726–0.761, P<0.001) in validation, and 0.751 (95% CI 0.725–0.777, P<0.001) in application cohorts. Calibration was excellent (all Hosmer–Lemeshow P>0.05). Decision curve analysis showed net benefit across risk thresholds. In the application cohort, observed cachexia prevalence was 12.5% in low-risk vs. 42.3% in high-risk groups (P<0.001). The low-risk group had significantly longer overall survival (HR 2.887, 95% CI 1.513–5.508, P<0.001) and relapse-free survival (HR 1.482, 95% CI 1.087–2.020, P=0.01). On multivariate Cox regression, the risk score was an independent prognostic factor for overall survival (HR 7.797, 95% CI 1.431–42.477, P=0.018) and relapse-free survival (HR 4.793, 95% CI 1.798–12.779, P=0.002).
**Clinical Implications:** This risk score provides a practical, evidence-based tool using routinely available clinical data to identify digestive tract cancer patients at high risk of cancer cachexia before abdominal surgery. The inclusion of time from symptom onset to hospitalization—a novel factor not previously studied—highlights the importance of early diagnosis. The score's ability to independently predict survival outcomes suggests it could guide preoperative counseling, nutritional interventions, and treatment planning. Limitations include its retrospective, single-center design and predominantly Chinese population, necessitating prospective multicenter validation in diverse populations. Future studies should also consider incorporating functional measures such as grip strength.