Revision of Pancreatic Neck Margins Based on Intraoperative Frozen Section Analysis Is Associated With Improved Survival in Patients Undergoing Pancreatectomy for Ductal Adenocarcinoma

Objective: To test the hypothesis that complete, tumor-free resection at the pancreatic neck, achieved either en-bloc or non-en-bloc (ie, revision based on intraoperative frozen section [FS] analysis), is associated with improved survival as compared with incomplete resection (IR) in pancreatic ductal adenocarcinoma. Summary Background Data: Given the likely systemic nature of pancreatic ductal adenocarcinoma, the oncologic benefit of achieving a histologically complete local resection, particularly through revision of a positive intraoperative FS at the pancreatic neck, remains controversial. Methods: Clinicopathologic and treatment data were reviewed for 986 consecutive patients with ductal adenocarcinoma at the head, neck, or uncinate process of the pancreas who underwent open pancreatectomy as well as intraoperative FS analysis between 1998 and 2012 at Massachusetts General Hospital and between 1998 and 2013 at the University of Verona. Overall survival (OS) and perioperative morbidity and mortality were compared across 3 groups: complete resection achieved en-bloc (CR-EB), complete resection achieved non-en-bloc (CR-NEB), and IR. Results: The CR-EB cohort comprised 749 (76%) patients, CR-NEB 159 patients (16%), and IR 78 patients (8%). Other than a higher incidence of vascular resection among CR-NEB and IR patients, no demographic, pathologic (eg, tumor grade, lymph node positivity, superior mesenteric artery involvement), or treatment factors (eg, neoadjuvant and adjuvant therapy use) differed between the groups. Median OS was significantly higher in patients with CR-EB (28 mo, P = 0.01) and CR-NEB resections (24 mo, P = 0.02) as compared with patients with IR resections (19 mo). After adjusting for clinicopathologic and treatment characteristics, CR-EB and CR-NEB margin status were found to be independent predictors of improved OS (relative to IR, CR-EB hazard ratio [HR] 0.65, 95% confidence interval [CI] 0.49-0.86; CR-NEB HR 0.69, 95% CI 0.50-0.96). There were no intergroup differences in perioperative morbidity and mortality, including rates of pancreatic fistula. Conclusions: For patients with ductal adenocarcinoma at the head, neck, or uncinate process of the pancreas undergoing pancreatectomy, complete tumor extirpation via either en-bloc or non-en-bloc complete resection based on FS analysis is associated with improved OS, without an associated increased perioperative morbidity or mortality.