Late anastomotic leaks in pancreas transplant recipients - clinical characteristics and predisposing factors

Background: Anastomotic leaks after pancreas transplants usually occur early in the postoperative course, but may also be seen late post-transplant. We studied such leaks to determine predisposing factors, methods of management, and outcomes. Results: Between January 1, 1994 and December 31, 2002, a total of 25 pancreas transplant recipients at our institution experienced a late leak (defined as one occurring more than 3 months post-transplant). We excluded recipients with an early leak or with a leak seen immediately after an enteric conversion. The mean recipient age was 40.3 yr; mean donor age, 31.3 yr. The category of transplant was as follows: simultaneous pancreas-kidney (n = 5, 20%), pancreas after kidney (n = 10, 40%), and pancreas transplant alone (n = 10, 40%). At the time of their leak, most recipients (n = 23, 92%) had bladder-drained pancreas grafts; only two recipients (8%) had enteric-drained grafts. The mean time from transplant to the late leak was 20.5 months (range = 3.5-74 months). A direct predisposing event or risk factor occurring in the 6 wk preceding leak diagnosis was identified in 10 (40%) of the recipients. Such events or risk factors included a biopsy-proven episode of acute rejection (n = 4, 16%), a history of blunt abdominal trauma (n = 3, 12%), a recent episode of cytomegalovirus infection (n = 2, 8%), and obstructive uropathy from acute prostatitis (n = 1, 4%). Non-operative or conservative care (Foley catheter placement with or without percutaneous abdominal drains) was the initial treatment in 14 (56%) of the recipients. Such care was successful in nine (64%) of the 14 recipients; the other five (36%) required surgical repair after failure of conservative care at a mean of 10 d after Foley catheter placement. Of the 25 recipients, 11 underwent surgery as their initial leak treatment: repair in nine and pancreatectomy because of severe peritonitis in two. After appropriate management (conservative or operative) of the initial leak, five (20%) of the 25 recipients had a recurrent leak; the mean length of time from initial leak to recurrent leak was 5.6 months. All five recipients with a recurrent leak ultimately required surgery. Conclusions: Late anastomotic leaks are not uncommon; they may be more common with bladder-drained grafts. One-third of the recipients with a late leak had experienced some obvious preceding event that predisposed to the leak. For two-thirds of our stable recipients with bladder-drained grafts, non-operative treatment of the leak was successful.