A preoperative score to predict early death after pancreatic cancer resection

Background This study aimed to develop and validate a preoperative prognostic model for death within one year post-surgery in patients with resectable pancreatic ductal adenocarcinoma (PDAC). Methods A derivation cohort study of 296 patients who underwent surgical resection of PDAC was prospectively enrolled in an observational study. Preoperative predictors of one year mortality were used to develop a risk score which was then validated in an external cohort of 182 patients with resectable PDAC. Results Seventy-eight out of 296 patients (26%) died within the first year. Preoperative independent predictors of one year mortality were: nutritional status (Geriatric Nutritional Risk Index, OR 2.23, 1.14–4.38; p = 0.02), American Society of Anaesthesiologists†score (OR 2.56, 1.1–5.98; p = 0.03), abdominal or back pain at presentation (OR 2.51, 1.05–5.9; p = 0.038) and non metastatic liver disease as comorbidity (OR 4.5, 1.05–19.3; p = 0.043). A score ranging from 0 to 7 points was developed. In the validation cohort, the model was able to predict early mortality (OR 7.1, 3.9–12.7; p < 0.0001), with a predictive ability of 53.5% (Nagelkerke R2), an area under the receiver operating characteristic curve of 88.7% and an acceptable calibration (goodness-of-fit test, p = 0.403). Conclusions Our new simple risk score proved reliable in forecasting one year mortality in patients with resectable PDAC.