Portal vein patency after pancreatoduodenectomy for periampullary cancer

Abstract

Abstract Background The fate of the portal vein (PV) after pancreatoduodenectomy, especially its long-term patency and associated complications, has received little attention. The aim of this study was to explore the long-term patency rate of the PV after pancreatoduodenectomy, focusing on risk factors for PV stenosis/occlusion and associated complications. Methods Serial CT images of patients who underwent pancreatoduodenectomy for periampullary cancer between January 2000 and June 2012 in a single institution were evaluated for PV stenosis or occlusion. Results A total of 826 patients were enrolled. The PV stenosis/occlusion rate after pancreatoduodenectomy was 19·6 per cent and the 5-year patency rate 69·9 per cent. The most frequent cause of PV stenosis/occlusion was local recurrence followed by postoperative change and PV thrombosis. Patients who underwent PV resection had a higher PV stenosis/occlusion rate than those who did not (51 versus 17·4 per cent; P < 0·001). The 3-year patency rate was highest in patients with cancer of the ampulla of Vater and lowest in patients with pancreatic cancer (91·9 versus 55·5 per cent respectively; P < 0·001). Multivariable analysis showed that risk factors for PV stenosis/occlusion included primary tumour location, chemoradiotherapy and PV resection. PV stenosis or occlusion without disease recurrence was observed in 17·3 per cent of the patients. PV resection and grade B or C pancreatic fistula were independent risk factors for PV stenosis/occlusion. Among 162 patients with PV stenosis or occlusion, five (3·1 per cent) had fatal recurrent gastrointestinal bleeding. Conclusion PV stenosis or occlusion is common after pancreatoduodenectomy, particularly if the PV has been resected and/or chemoradiotherapy was given after surgery. Although recurrence is the most frequent cause of PV stenosis/occlusion, this complication is found in a significant proportion of patients without disease recurrence.