Objectives: Advances in surgery and perioperative care have contributed to improved outcomes after pancreas transplant. However, the development of peripancreatic infections carries a poor prognosis. It is not clear whether abdominal drainage is helpful in collection prevention.
Materials and Methods: A retrospective review of adult consecutive pancreas transplants at a single institution between January 2017 and December 2018 was undertaken. Postoperative outcomes were compared between patients in whom prophylactic intraoperative drains were placed and patients with no drains.
Results: We identified 83 patients who underwent pancreas transplant with a median age of 45 years; 54.2% were males, and median body mass index was 25.8. Thirty patients had 1 or 2 drains placed (36.1%). There was no difference in the readmission rate (70.0% vs 60.4%; P = .48), reoperation (20.0% vs 30.2%; P = .44), or percutaneous drainage of peripancreatic infections (20.0% vs 15.1%; P = .56) between patients with drains and no drains, respectively. However, prophylactic drainage was associated with a lower rate of reoperation for peripancreatic infections compared with those who were not drained (0.0% vs 13.2%; P < .05). No graft loss occurred in the drain group.
Conclusions: Prophylactic drainage after pancreas transplant may be helpful for reduction in the infection rate after reoperation. The risks of drain placement should be weighed against those of drain avoidance.
Key words : Drains, Morbidity, Peripancreatic infections, Postoperative outcomes
Pancreas transplant (PTx) is an excellent therapy for many patients with diabetes, particularly in the context of renal failure.1 There has been significant improvement in perioperative outcome after PTx since the first successful operation in1966 at the University of Minnesota.2,3 During the first 2 to 3 decades, there was discussion in the literature about the optimal management of pancreatic exocrine secretions, culminating in the current standard for enteric drainage.4,5 Discussion about venous outflow has been ongoing, and most transplants are now performed using systemic rather than portal drainage.6 Despite the many published reports on surgical improvements in transplantation, there is no clear consensus as to whether abdominal drainage of the transplanted pancreas is necessary.
Technical hurdles and historically difficult outcomes have made training in PTx a challenge.7,8 For this reason and others, including reduced referral rates, improved nonsurgical treatments for diabetes, and fear of regulatory scrutiny, rates of PTx have decreased in the United States since 2004.9 Paradoxically, simultaneous pancreas and kidney (SPK) transplant is presently among the safest of all solid-organ transplant procedures.10 Furthermore, PTx provides excellent quality of life and prolongs patient survival compared with kidney transplant alone.11 To this end, reports that help improve access (surgically and medically) to PTx, particularly in the form of SPK transplant, are important for transplant providers.
Pancreas transplant volumes at our institution have increased approximately 200% in 24 months. With the knowledge that rapid volume growth can challenge the system and lead to poor outcomes, our group has critically evaluated each step of the transplant process in order to maintain high quality and to provide access to care.12 Accordingly, amidst our consideration of rates of postoperative complications and opportunities for prevention, we asked the question, Is the use of prophylactic abdominal drainage after PTx necessary?
This is an important unanswered question because pancreas transplant candidates are at high risk for secondary complications of diabetes, from gastroparesis to autonomic dysfunction, and each of these places these patients at a higher postoperative risk.13 Surgery, inflammation, ischemia and reperfusion, and infection may all worsen both gastroparesis and autonomic dysfunction perioperatively, potentially increasing the hospital length of stay, as well as rates of reoperation and readmission.14 Complications such as these may reduce interest in offering PTx to transplant candidates. Thus, any interventions (eg, such as drains) that might decrease postsurgical abdominal infection risk or limit exposure to subsequent complications might also improve short- and long-term outcomes while reducing ongoing health care system needs.
We hypothesized that prophylactic drain placement might decrease rates of surgical peripancreatic fluid collections and that infection avoidance would improve transplant-related outcomes. With the goal of informing our own practice, we retrospectively reviewed our experience to determine whether a protocol for standard drain placement after PTx should be undertaken. We found fewer significant infections in cases for which prophylactic drains were placed.
Materials and Methods
After institutional review board approval, we performed a retrospective review of all consecutive, adult cases of PTx at our center between January 2017 and December 2018. Patients who underwent PTx alone (PTA), pancreas after kidney (PAK) transplant, and SPK transplant were included.
We reviewed demographics and comorbidities, including age, sex, race, type of diabetes mellitus (DM), history of hypertension, prior transplants, prior abdominal surgeries, and body mass index (BMI, calculated as weight in kilograms divided by height in meters squared). Variables in transplanted organs, including cold ischemia time, kidney donor profile index, and import/local status, were considered. Cold ischemia time was calculated from donor cross-clamp time to allograft reperfusion. Operative details were reviewed, including the route of venous drainage (portal vs systemic), drainage of pancreatic enzymes (enteric vs bladder), and transabdominal drain placement. Postoperative laboratory values were evaluated, including peak serum lipase (U/L) and both serum lipase and creatinine (mg/dL) at postoperative day 30. Peak lipase was defined as the highest lipase value during the first week after transplant, and terminal creatinine was the last creatinine value obtained. The upper limit of normal lipase at our institution is 300 U/L. We analyzed the number of pancreas allograft ultrasonographies and computed tomographies of the abdomen and pelvis that were performed within 30 days from the operation. Postoperative outcomes included reoperation status, readmission within 30 days from the discharge date of the index hospitalization, need for percutaneous drain placement, kidney delayed graft function in patients who underwent SPK, pancreas rejection, and graft loss. Kidney delayed graft function was defined as the need for dialysis within the first week after transplant. Graft loss was defined as the occurrence of one or more of the following events: pancreatectomy performed to treat graft thrombosis or infection, patient returned to insulin treatment, or patient death.
Back-bench preparation of the pancreas allograft at the University of Maryland follows standard steps. First, the spleen is removed, the small bowel is stapled to the appropriate length, the bowel staple lines are oversewn, and the mesentery is shortened and oversewn. Thereafter, the donor iliac arterial Y graft is anastomosed to the donor pancreas superior mesenteric and splenic arteries. The recipient operation involves a midline laparotomy with exposure of the retroperitoneum. The intestines are retracted superiorly. In systemically drained allografts, the pancreas is placed on the right with a portocaval anastomosis. The arterial Y graft is anastomosed to the right common iliac artery. For portal drainage, the donor portal vein is anastomosed to the recipient superior mesenteric vein, and an extended arterial Y graft is anastomosed to the right common iliac artery. All allografts are enterically drained by a hand-sewn double-layered side-to-side duodenojejunostomy. In the cases of SPK, a kidney transplant is then performed on the left external iliac artery and vein. Peripancreatic drains are placed after achieving hemostasis and before abdominal closure. Round silicone 19F Jackson-Pratt drains were used. Drains were progressively placed for prophylaxis throughout the study such that by May 2018 nearly all cases incorporated drains. Before standard drain placement, indications for drainage early in the study were bleeding, significant reperfusion injury, and infection prevention.
After reperfusion of the pancreas allograft mean arterial pressure is maintained above 75 mm Hg. In nonuremic patients (ie, PAK and PTA cases), 1 dose of intravenous heparin (2000-5000 U) is used at the time of vessel clamping at the discretion of the surgeon. A mixture of antibiotic irrigation is used at the conclusion of the case and before fascial closure. This antibiotic mixture contains 1 g of vancomycin, 50 mg of gentamicin, and 80 mg of amphotericin B diluted in 1 L of sterile water.
Immunosuppression and postoperative course
In compliance with our institution’s immunosuppression induction protocol, alemtuzumab (Campath-1H), 30 mg given intravenously as a single dose, is used as our standard induction for T-cell depletion. For patients with HIV infection but no evidence of AIDS manifestations, thymoglobulin, 2 mg/kg given intravenously for 3 doses, is used. Basiliximab (Simulect), 20 mg given intravenously in 2 doses, is used in patients who are immunosuppressed iatrogenically (eg, patients with a functioning transplanted organ from a previous operation). Maintenance immunosuppression is maintained with a triple regimen consisting of tacrolimus (goal, 9-11 ng/mL), mycophenolic acid, and a steroid taper protocol for 21 days. Regarding our institution’s prophylactic antibiotic protocol, we use piperacillin/tazobactam (Zosyn) given preoperatively and continued for 48 hours postoperatively. For patients allergic to penicillin, we use aztreonam and clindamycin instead. For patients with history of methicillin-resistant Staphylococcus aureus colonization, vancomycin is added. Fluconazole is an antifungal prophylactic agent and is given once prior to incision and continued for 5 days.
Postoperatively, patients are admitted to the surgical intensive care unit for 12 to 24 hours for monitoring and blood pressure management (mean arterial pressure > 75 mm Hg) and are extubated in the intensive care unit. Aspirin, 81 mg, is initiated on postoperative day 1 if there is no evidence of bleeding and platelet count is > 50 000/μL. Nasogastric decompression is maintained for the first 24 hours, and patients are usually started on clear liquid diet on postoperative day 3. At least one baseline pancreas allograft ultrasonography exam with Doppler is performed before discharge. Drains are maintained with bulb suction. Drain amylase measurement is checked selectively. At our center, pancreatic leak is defined as an elevation of drain amylase > 3 times the serum level. Drains are removed in the first outpatient follow-up visit unless there is evidence of a pancreatic leak or infection.
Clinical variables for patients with drains and no drains were compared. Summary statistics such as medians, interquartile ranges (IQR), and percentages were compiled for all measured variables. We used the Fisher exact test to analyze binary data. We used the Wilcoxon rank sum test to compare continuous variables between the 2 groups. Statistical significance was determined at the 2-sided α = .05 level (P ≤ .05) for all tests. All statistical analyses were done in R (version 3.5.1).15
During the study period, 83 patients underwent PTx. Forty-five patients were males (54.2%) with a median age at the time of transplant of 45 years (IQR, 40-52 years), and most were White (55.4%). Transplant was for type 1 DM in 66 patients (79.5%). Alemtuzumab was used in 71 patients for induction immunosuppression, thymoglobulin was used in 7 patients, and basiliximab was used in 5 patients.
Forty-one allografts (49.4%) were procured from hospitals within our organ procurement organization. The remainder were shipped from a distance (ie, imported, not local). The median kidney-donor profile index was 25% (IQR, 12% to 40%) with a median cold ischemia time of 8.6 hours (IQR, 5.5-14.3 hours). Eight transplants were performed using organs procured from donations after cardiac death. There were no statistically significant differences between the 2 groups in terms of the organ variables (Table 2).
Drain placement and operative details
Intraoperative, transabdominal drains were placed in 30 patients (36.1%). Of these, 6 patients had 1 drain placed (20.0% of those drained, 7.2% overall) and 24 patients had 2 drains placed (80.0% of those drained, 28.9% overall). Sixty-eight patients (81.9%) received SPK; of these, 25 patients (36.7%) had drains. Ten patients (12.0%) received PAK, of which 4 patients (40.0%) were drained. Five patients had PTA (6% overall), of which only 1 patient (20.0%) was drained. Systemic venous drainage was performed for all but 3 cases (3.6%). All grafts had enteric drainage. Those who received drains and those who did not were similar with regard to demographics and comorbidities. However, there was a higher incidence of type 1 DM in those patients who were not drained (88.7% vs 63.3%; P = .01) (Table 1).
Patient and graft outcome measures
One patient (1.2%) required both allograft nephrectomy and pancreatectomy secondary to bleeding. A second patient (1.2%) had cellular and antibody-mediated rejection that resulted in both renal and pancreas allograft failure after 10 months, but organ removal was not required. Two patients (2.4%) died with functioning grafts. Two patients (2.4%) had pancreatic graft loss secondary to rejection that required graft pancreatectomy. One patient (1.2%) required pancreatectomy for thrombosis. For the second year of the study (2018), the 30-day rate of thrombosis-related graft loss was 0.0%. Of the 68 patients who underwent SPK, 13 (19.1%) developed delayed graft function of the renal allograft; there was no significant statistical difference in the delayed graft function rate between the drain and no-drain group (31.6% vs 19.4%, respectively; P = .53).
Drain-related outcome measures
Patients with drains underwent more pancreas ultrasonography tests compared with patients who did not have drains (3 vs 2 tests; P = .003). There were no statistical differences in peak lipase, mean serum lipase, serum creatinine at 30 days, terminal creatinine, or number of computed tomography scans performed, for those with drains compared with those who did not receive drains (Table 3).
Two patients in the no-drain group were on dialysis at the study endpoint; of these, the first patient underwent allograft nephrectomy secondary to bleeding (resulting from intra-abdominal infection), and the second patient was treated for rejection after 10 months. There was no statistical difference in readmission rate or the need for placement of a percutaneous drain for drainage of a peripancreatic fluid collection (Table 4). Seven patients in the no-drain group had graft loss versus no graft loss in the drain group (13.2% vs 0.0%; P = .04).
The average total daily drain output for the first 7 days was 515, 630, 464, 422, 490, 494, and 527 mL, respectively. Twenty-one patients (70%) had at least 1 measurement of drain amylase, which was elevated in 7 of these patients.
Ten patients (12.0%) developed pancreatic leaks. Of these, 7 patients (70.0%) were prophylactically drained, whereas the remaining 3 patients (30.0%) required subsequent placement of a percutaneous drain by interventional radiology. Subcutaneous octreotide was used in 5 of these patients (50.0%). None of the patients (0.0%) who received organs from donors after cardiac death developed a pancreatic leak.
There was no statistical difference in the reoperation rate for the 2 groups (overall). However, more patients in the no-drain group required reoperation for peripancreatic fluid collections (13.2% for no-drain group vs 0.0% for the drain group; P = .04) (Figure 1). In the drain group, 6 patients (20.0%) required reoperation; however, none of these operations was for peripancreatic infections. Two patients (6.7%, after SPK) required reexploration for urine leak that required revision of ureteral anastomosis. Two patients (6.7%) presented with small bowel obstruction and required reexploration. One patient (3.3%) required reexploration for bleeding in the first 24 hours posttransplant. A single patient (3.3%) required reexploration and revision of the duodenojejunostomy on postoperative day 6 for gastrointestinal bleeding from the enteric anastomosis.
In patients who did not receive drains, 16 patients (30.2%) required reoperation. Six patients (11.3%) were reexplored for presumed bleeding in the early postoperative period. One patient (1.8%) required reexploration for small bowel obstruction. One patient (1.8%) developed a ureteral stricture that required reexploration and revision of ureteral anastomosis after SPK. One patient (1.8%) required reexploration for partial fascial dehiscence. Reoperation for peripancreatic fluid collections was performed in 7 patients (13.2%), and each of these reexplorations occurred within 30 days of transplant. Once reoperation occurred for infection, it was common to initiate additional, serial operations. Of these 7 patients, 2 patients (28.5%) had a temporary open abdomen and 2 patients (28.5%) required a Roux-en-Y reconfiguration. One patient (14.3%) developed a mycotic pseudoaneurysm of the Y graft that led to allograft pancreatectomy and nephrectomy (organs were unilateral, on the right) (Table 5).
Nine of 83 patients (10.8%) had a positive result for an organism cultured from their peripancreatic fluid collection (either obtained operatively or percutaneously). Gram-negative rods were the most common cultured organisms, and some patients had more than 1 organism identified. Candida organism was found in cultures from 2 of 9 patients (22.2%).
Pancreas transplant is an ideal therapy for certain patients with DM. The most common form of PTx is SPK; however, solitary PTx is also an effective treatment for DM, particularly in patients with hypoglycemic unawareness.16 Pancreas transplant improves quality of life, is associated with better self-identified health scores, and results in a lower rate of daily DM-associated frustration.17 In the long term, PTx has been shown to affect the progression of secondary complications of DM.18,19 Furthermore, PTx in SPK and PAK have been shown to protect the renal allograft, and kidney-related outcomes for these patients are better than for those patients who receive a kidney transplant alone.20-22 At 10 years, patient survival is better for recipients of SPK with DM compared with recipients of living donor kidney transplant.11 The survival benefit in SPK is likely attributable to the beneficial effects of the pancreas over time.11
Technical failure has challenged PTx, and reported surgical complications have ranged between 16% and 36% in the 1980s.23 Despite many advances in technical surgery, PTx still carries a 5% to 7% risk of technical failure.10 This nontrivial rate of technical graft failure, in combination with the challenges in postoperative management described in this report, may in part contribute to the decline in the number of PTx surgical procedures in the United States over the last decade.7,8
In the current retrospective analysis, we reported on the role of prophylactic transabdominal drains at the index PTx operation. For patients with prophylactic transabdominal drains, we observed lower rates of reoperation for infection, which helped avoid significant morbidity. This is important because, when infections occur, these may lead to both morbidity and mortality.
Drain placement is widely discussed in general surgery. Some studies are supportive, whereas others suggest drain avoidance.24-26 Pancreas transplant is different from general surgery as well as from kidney and liver transplant. First, unlike general surgery, PTx patients are pharmacologically immunosuppressed. Second, unlike kidney and liver transplant, there is frequently gross contamination of the field in a PTx as a result of the duodenojejunostomy. Finally, in PTx, infection comes at a significant morbidity, as was observed here (eg, open abdomen), whereas drain placement likely does not. Because it may lie adjacent to major vascular structures, peripancreatic abscess after PTx may lead to pseudoaneurysm formation or overt vascular rupture.27,28 Taken together, the risks of drain placement after PTx may be lower than the risks of potentially avoidable infection.
We observed higher rates of reoperation in patients who were not drained (30.2% vs 20.0%), but this did not reach statistical significance. Differences in reoperation were observed when we analyzed those who required reoperation for infection. The development of peripancreatic infections carried a poor prognosis in our cohort, as some of the patients required serial explorations with open abdomen and potential graft loss. Notably, the overall reoperation rate was high. This higher-than-expected rate was the impetus for taking a critical look at our perioperative practice. This may have caused the drop in reoperation rates over the course of this 2-year study to less than 10% in the final 6 months and may reflect a center learning curve in the context of quality improvement implementation and in the context of rapid rise in surgical volume. Indeed, using a system of “internal conversion” to offer kidney transplant candidates SPK versus kidney alone has been effective for treating patients quickly, with high-quality organs, thus increasing surgical volumes.29
Bleeding is also a concern after PTx. We observed that 6 of 53 patients (11.3%) in the no-drain group underwent reoperation for bleeding. In contrast, only 1 patient (3.3%) in the drained group required reoperation for bleeding. This may suggest that avoidance of drains is associated with a lower threshold to return to the operating room, perhaps because drain effluent (ie, blood) aids in diagnosis.
Graft loss was relatively uncommon in our cohort. In this study, 7 patients (8.4%) lost their grafts. In 2017, 6 of 41 patients (14.6%) had graft loss, whereas in 2018, only 1 of 42 patients had graft loss (2.3%). All graft losses happened in the no-drain group, and 2 of these patients had prior evidence of peripancreatic infections requiring exploration that eventually led to graft loss. One graft loss was a death with a functioning graft in a patient who required multiple operations, the first of which was an infected peripancreatic infection. This case was eventually complicated by an enteric leak that required conversion to a Roux-en-Y enteric drainage. The second graft loss was for allograft pancreatectomy related to bleeding. None of the other 5 graft losses were specifically related development of peripancreatic infections.
Immunologic graft loss, in the context of published rates (1.8% to 6.0%) within the first year, was uncommon.30 We had only 1 early graft loss (1.2%) due to technical failure resulting in graft thrombosis on postoperative day 1. Two patients had late graft thrombosis that required graft pancreatectomy with evidence of rejection on the final pathology report. This low rate of graft loss from immunologic causes may be related to a conservative approach to organ acceptance. Our group avoids transplant in the presence of donor-specific antibody and relies heavily on virtual crossmatch, wherein recipient mean fluorescent intensity values for discordant antigens are generally all less than 1000. Given the attendant surgical risks of PTx, our group feels that compound risk of multiple factors such as donor-specific antibody or high mean fluorescent intensity values puts the patient at unacceptable risk. Alternatively, our heavy reliance on alemtuzumab is also a likely explanation.
Our study has several weaknesses. First, this was a retrospective study and it carried the attendant limitations of drawing a causal relationship between prophylactic drain placement and reoperation for peripancreatic infections. However, there does appear to be an association between drain placement and reduced reoperations for infection. Second, volume and outcome are known to be interrelated.31,32 Our center experienced a rapid rise in surgical volume during the study period.29 This may have affected our observations. Finally, we only captured the patients who developed peripancreatic infections and required reexplorations and patients who required percutaneous drainage. We failed to capture the patients who were treated nonoperatively (eg, with antibiotics alone); however, it is uncommon to have image-proven infection that is not treated surgically or with percutaneous drainage.
Overall, the prophylactic placement of transabdominal drains may reduce the rate of peripancreatic infections that would require surgical reexploration. A prospective study could more definitively address this important topic. In the opinion of the authors, the potential benefits of prophylactic drain placement for PTx justify the risk.
Volume : 19
Issue : 1
Pages : 64 - 71
DOI : 10.6002/ect.2020.0010
From the 1Division of Transplantation, Department of Surgery, University of
Maryland School of Medicine Baltimore, Maryland, USA; and the 2Biomedical and
Health Informatics, Children’s Hospital of Philadelphia, Philadelphia,
Acknowledgements: The authors have not received any funding or grants in support of the presented research or for the preparation of this work and have no declarations of potential interest. Author contributions are as follows: Scalea and Goussous designed the study; Goussous and Scalea wrote the manuscript; Goussous, St. Michel, Mcdade, Borth, and Gaines collected the data; Bromberg, Al-Qaoud, and Barth revised the manuscript; and Dawany analyzed the data.
Corresponding author: Joseph R. Scalea, Division of Transplantation, Department of Surgery, University of Maryland, School of Medicine, 29 S. Greene Street, Baltimore, MD 21201, USA
Phone: +410 328 0038
Table 1. Demographics and Comorbidities of Patients
Table 2. Organ Details
Table 3. Postoperative Investigations
Table 4. Patient Outcomes
Table 5. Reasons for Reexploration
Figure 1. Reoperation Rates for the Group With Drains Compared With the Group Without Drains