Technical risk factors for portal vein reconstruction thrombosis in pancreatic resection.

Objective: Vascular reconstruction can facilitate pancreas tumor resection, but optimal methods of reconstruction are not well studied. We report our results for portal vein reconstruction (PVR) for pancreatic resection and determinants of postoperative patency.

Methods: We identified 173 patients with PVR in a prospective database of 6522 patients who underwent pancreatic resection at our hospital from 1970 to 2014. There were 128 patients who had >1 year of follow-up with computed tomography imaging. Preoperative, intraoperative, and postoperative factors were recorded. Patients with and without postoperative PVR thrombosis were compared by univariable, multivariable, and receiver operating characteristic curve analyses.

Results: The survival of patients was 100% at 1 month, 88% at 6 months, 66% at 1 year, and 39% on overall median follow-up of 310 days (interquartile range, 417 days). Median survival was 15.5 months (interquartile range, 25 months); 86% of resections were for cancer. Four types of PVR techniques were used: 83% of PVRs were performed by primary repair, 8.7% with interposition vein graft, 4.7% with interposition prosthetic graft, and 4.7% with patch. PVR patency was 100% at 1 day, 98% at 1 month, 91% at 6 months, and 83% at 1 year. Patients with PVR thrombosis were not significantly different from patients with patent PVR in age, survival, preoperative comorbidities, tumor characteristics, perioperative blood loss or transfusion, or postoperative complications. They were more likely to have had preoperative chemotherapy (53% vs 9%; P < .0001), radiation therapy (35% vs 2%; P < .0001), and prolonged operative time (618 ± 57 vs 424 ± 20 minutes; P = .002) and to develop postoperative ascites (76% vs 22%; P < .001). Among patients who developed ascites, 38% of those with PVR thrombosis did so in the setting of tumor recurrence at the porta detected on imaging, whereas among patients with patent PVR, 50% did so (P = .73). Patients with PVR thrombosis were more likely to have had prosthetic graft placement compared with patients with patent PVRs (18% vs 2.7%; P = .03; odds ratio [OR], 7.7; 95% confidence interval [CI], 1.4-42). PVR patency overall was significantly worse for patients who had an interposition prosthetic graft reconstruction (log-rank, P = .04). On multivariable analysis, operative time (OR, 1.01; 95% CI, 1.01-1.02) and prosthetic graft placement (OR, 8.12; 95% CI, 1.1-74) were independent predictors of PVR thrombosis (C statistic = 0.88).

Conclusions: Long operative times and use of prosthetic grafts for reconstruction are risk factors for postoperative portal vein thrombosis. Primary repair, patch, or vein interposition should be preferentially used for PVR in the setting of pancreatic resection.