Background: Management of patients with borderline resectable/locally advanced (BR/LA) pancreatic adenocarcinoma is based on knowledge of natural history and patterns of treatment failure, information of great importance to large data registries. Using the SEER database, we examined the survival for patients with BR/LA tumors and critically evaluated the utility of the data.
Methods: T3/T4 tumors from 2004 to 2007 were divided into those that involved the portal vein/superior mesenteric vein/gastroduodenal artery/hepatic artery and those that involved the superior mesenteric artery (SMA) or celiac axis. The control group (CG) included patients who were recommended surgery but did not undergo it. Multivariate disease-specific survival analyses were performed using the Cox proportional hazards model.
Results: Of 3,837 patients, 571 patients (15 %) were recommended surgery, and 323 (8 %) underwent surgical resection. We were unable to separate patients into BR/LA based on current NCCN guidelines. We were able to identify vascular involvement but not those who actually underwent vascular resection. Median survival of patients who underwent surgery with SMA and celiac involvement was 12 and 8 months compared with 7 and 6 months, respectively, in the CG (p = .01). Patients who underwent surgical resection with venous involvement had a longer survival than those with arterial involvement (18 vs 12 months, p = .001).
Conclusions: Analysis of patients with BR/LA pancreatic adenocarcinoma who underwent pancreatic resection in the SEER database yielded limited information. New manuals must focus on obtaining information consistent with current advances in the field; our recommendations for optimizing the SEER database are included.