Routine CT reporting of vascular involvement in pancreatic ductal adenocarcinoma: completeness and association with margin-positive resection.

backgroundNational Comprehensive Cancer Network (NCCN) guidelines emphasize detailed reporting of tumor-vessel relationships for pancreatic ductal adenocarcinoma (PDAC), but the completeness and prognostic significance of vascular findings documented in routine clinical CT reports remain uncertain in the contemporary era of neoadjuvant therapy (NAT). PURPOSE: To assess the completeness of NCCN-relevant vascular reporting in routine preoperative CT examinations and determine whether CT-reported vascular findings and derived NCCN resectability categories are associated with margin-positive (R1) resection. MATERIALS AND

methodsIn this retrospective single-center study, 528 patients who underwent resection for pathologically confirmed PDAC between 2011 and 2026 were identified. For each patient, the most recent preoperative staging CT report and corresponding surgical pathology report were linked. Tumor-vessel relationships involving the superior mesenteric artery (SMA), celiac axis, common hepatic artery (CHA), superior mesenteric vein (SMV), and portal vein (PV) were extracted from routine clinical reports via a large language model, which underwent validation against 2 readers. Multivariable logistic regression was used to evaluate associations between vascular findings, NCCN resectability category, and R1 resection after adjustment for age, sex, tumor size, tumor location, and NAT status.

resultsAmong 528 patients (mean age, 69.2 years ±9.9; 277 men), 257 (48.7%) received NAT and 172 (32.6%) had R1 resection. Complete reporting of all five NCCN vessels was present in 224/528 (42.4%) reports. NCCN resectability category could be assigned in 322 patients, including 150 (46.6%) resectable, 123 (38.2%) borderline resectable, and 49 (15.2%) locally advanced tumors. R1 rates increased progressively across categories (26.7%, 45.5%, and 53.1%, respectively; p < 0.001). After adjustment, SMA abutment (OR, 2.85; p = 0.001), SMA encasement (OR, 2.49; p = 0.023), celiac encasement (OR, 2.81; p = 0.021), CHA encasement (OR, 2.31; p = 0.031), PV abutment (OR, 2.31; p = 0.007), and SMV encasement (OR, 2.50; p = 0.014) were independently associated with R1 resection. In sensitivity analysis assuming unreported vessels had no tumor contact, borderline resectable tumors (OR, 2.68; 95% CI, 1.60-4.50) and locally advanced tumors (OR, 4.77; 95% CI, 2.22-10.23) demonstrated progressively higher odds of R1 resection relative to resectable tumors.

conclusionComplete NCCN-compliant vascular reporting was uncommon in routine PDAC CT reports, limiting standardized resectability classification. Nevertheless, CT-reported vascular findings and NCCN resectability categories derived from routine reports retained significant associations with margin-positive resection, supporting the continued clinical relevance of vascular staging information communicated in everyday radiology practice.