Jennifer F. Tseng, Charlotte C. Sun, Eddie K. Abdalla, Jean-Nicolas Vauthey, Peter W. T. Pisters, Jeffrey E. Lee, Douglas B. Evans, UT/ MD Anderson Cancer Center, Houston, TX

Introduction: Major vascular resection performed at the time of pancreaticoduodenectomy (PD) for adenocarcinoma remains controversial partly due to the lack of standardized preoperative imaging and prospective evaluation of resection margins. As a result, many reports have included patients who have undergone an incomplete (R2) resection precluding accurate analysis of other prognostic factors such as venous resection. Methods: We analyzed all patients who underwent PD for adenocarcinoma of the pancreas between 1990 and 2002. Preoperative imaging criteria for PD included the absence of tumor extension to the celiac axis or superior mesenteric artery; such patients were considered locally advanced and did not undergo PD. Tangential or segmental resection of the superior mesenteric or portal veins was performed when the tumor could not be separated from the vein. The retroperitoneal (RP) margin (soft tissue adjacent to the SMA) was prospectively evaluated at the time of pathologic evaluation and reported as positive (R1) or negative (R0). Results: PD was performed in 280 consecutive patients; 251 (90%) received adjuvant or neoadjuvant therapy. Standard PD was performed in 178 and venous resection (VR) in 102. Vascular resections included tangential resection with vein patch (28), segmental resection with primary anastomosis (28), and segmental resection with autologous interposition graft (46). No patient underwent an R2 resection. An R1 resection occurred in 20 (20%) of the 102 patients who underwent VR, and in 21 (12%) of the 178 patients who underwent standard PD (P=.056). Tumor size was significantly larger in patients who required VR (3.1 vs. 2.5 cm, p<.001). T3 tumors were also more common in the VR group (84% vs. 63%, p=.001). Median survival was 25 months in those who required VR and 26 months in those who underwent standard PD (P= .42). A Cox proportional hazards model was constructed to perform multivariate analysis of the effect of potential prognostic factors (VR, tumor size, T stage, N status, RP margin) on survival. Only the presence of nodal metastases was identified as a predictor of decreased survival (HR 1.53, p=.005). Conclusions: R0 resection is possible in the majority of patients who require venous resection. Properly selected patients with adenocarcinoma of the pancreatic head who require VR have a median survival of approximately 2 years, no different from those who undergo standard PD and superior to historical patients treated with nonoperative therapy.