Importance: There are currently no clinically relevant criteria to predict a futile up-front pancreatectomy in patients with anatomically resectable pancreatic ductal adenocarcinoma.
Objectives: To develop a futility risk model using a multi-institutional database and provide unified criteria associated with a futility likelihood below a safety threshold of 20%.
Design, setting, and participants: This retrospective study took place from January 2010 through December 2021 at 5 high- or very high-volume centers in Italy. Data were analyzed during April 2024. Participants included consecutive patients undergoing up-front pancreatectomy at the participating institutions.
Exposure: Standard management, per existing guidelines.
Main outcomes and measures: The main outcome measure was the rate of futile pancreatectomy, defined as an operation resulting in patient death or disease recurrence within 6 months. Dichotomous criteria were constructed to maintain the futility likelihood below 20%, corresponding to the chance of not receiving postneoadjuvant resection from existing pooled data.
Results: This study included 1426 patients. The median age was 69 (interquartile range, 62-75) years, 759 patients were male (53.2%), and 1076 had head cancer (75.4%). The rate of adjuvant treatment receipt was 73.7%. For the model construction, the study sample was split into a derivation (n = 885) and a validation cohort (n = 541). The rate of futile pancreatectomy was 18.9% (19.2% in the development and 18.6% in the validation cohort). Preoperative variables associated with futile resection were American Society of Anesthesiologists class (95% CI for coefficients, 0.68-0.87), cancer antigen (CA) 19.9 serum levels (95% CI, for coefficients 0.05-0.75), and tumor size (95% CI for coefficients, 0.28-0.46). Three risk groups associated with an escalating likelihood of futile resection, worse pathological features, and worse outcomes were identified. Four discrete conditions (defined as CA 19.9 levels-adjusted-to-size criteria: tumor size less than 2 cm with CA 19.9 levels less than 1000 U/mL; tumor size less than 3 cm with CA 19.9 levels less than 500 U/mL; tumor size less than 4 cm with CA 19.9 levels less than 150 U/mL; and tumor size less than 5 cm with CA 19.9 levels less than 50 U/mL) were associated with a futility likelihood below 20%. Both disease-free survival and overall survival were significantly longer in patients fulfilling the criteria.
Conclusions and relevance: In this study, a preoperative model (MetroPancreas) and dichotomous criteria to determine the risk of futile pancreatectomy were developed. This might help in selecting patients for up-front resection or neoadjuvant therapy.