Preoperative Prediction of Pathologic Response After Neoadjuvant Chemoradiotherapy for Pancreatic Ductal Adenocarcinoma: A Metabolic Response–Based Predictive Model
Hironobu Suto, Yoshio Shimizu, Takuro Fuke, Hiroyuki Matsukawa, Yasuhisa Ando, Minoru Oshima, Koji Fujita, Kiyoyuki Kobayashi, Hideki Kobara, Keiichi OkanoABSTRACT
Objectives
Neoadjuvant chemoradiotherapy (NACRT) is increasingly employed for resectable and borderline resectable pancreatic ductal adenocarcinoma (PDAC), yet reliable preoperative predictors of pathologic tumor regression are limited. We developed predictive models for pathologic response following NACRT.
Methods
We analyzed prospectively collected Phase II trial data from 169 patients with resectable and borderline resectable PDAC who underwent pancreatic resection after NACRT. Pathologic response was assessed using the College of American Pathologists (CAP) grading system and dichotomized as CAP grades 0–2 versus grade 3. Model A used baseline variables; Model B incorporated post‐treatment preoperative and treatment‐related variables. Performance was evaluated using the area under the curve (AUC).
Results
Among 169 patients, 81 (47.9%) achieved CAP grades 0–2. Model A found no independent baseline predictors (AUC 0.642). In Model B, post‐treatment maximum standardized uptake value (SUVmax) emerged as the sole independent predictor of CAP grades 0–2 (OR: 0.81, 95% CI: 0.67–0.96, p = 0.014). Model discrimination was higher with Model B (AUC: 0.745 vs 0.642; p = 0.006 by DeLong test). Risk stratification indicated CAP grade 0–2 rates of 27%, 42%, and 83% ( p < 0.001).
Conclusions
Model B improved the prediction of pathologic response after NACRT, with post‐treatment SUVmax as the sole independent predictor.