Portal Venous Versus Systemic Venous Drainage of Pancreas Grafts: Impact on Long-Term Results


Corresponding author: Mark S. Cattral, mark.cattral@uhn.on.ca


Portal venous (PV) and systemic venous (SV) drainage methods are used in pancreas transplantation. The impact of the reconstruction technique on long-term outcome remains unclear. We compared the efficacy and side effects of both methods in 192 recipients who received synchronous pancreas kidney transplants between November 1995 and November 2007. SV and PV drainage were used in 147 and 45 cases, respectively. Pancreas function was determined by hemoglobin A1c levels and annual oral glucose tolerance test. Serum creatinine assessed kidney function. Serum lipid (low-density lipoprotein, high-density lipoprotein and cholesterol) levels and body mass index were measured annually. Patient and graft survival were calculated by log-rank analysis. Pancreas survival for SV versus PV patients was similar after 5 years (81.8% vs. 75.5%) and 10 years (65.1% vs. 60%; p = NS). Similarly, no difference was detected between the groups regarding kidney survival after 5 years (92.9% vs. 84.4%) and 10 years (81.6% vs. 75.5%; p = NS). Patient survival did not differ at 5 years (94.3% vs. 88.8%) and 10 years (85.1% vs. 84.4%; p = NS). Pancreas and kidney function and the lipid profiles were similar in both groups. SV and PV drainage of pancreas grafts offer similar long-term graft survival and function and choice of method should remain the preference of the surgeon.


antithymocyte globulin


body mass index


cold ischemia time




cerebrovascular accidents


end-stage renal disease


insulin-dependent diabetes mellitus


impaired glucose tolerance


intravenous immune globulin


hemoglobin A1c


high-density lipoprotein


human leukocyte antigen


interleukin-2 receptor antagonists


low-density lipoprotein


length of hospitalization


modification of diet in renal disease estimated glomerular filtration rate


panel reactive antibody


pancreas transplants alone


portal venous drainage


synchronous pancreas kidney transplantation


Statistical Package for the Social Sciences


systemic venous drainage


warm ischemia time


Synchronous pancreas kidney transplantation (SPK) has become the treatment of choice for selected patients with insulin-dependent diabetes mellitus and end-stage renal disease (1,2). Two venous drainage options exist for pancreas grafts: portal venous (PV) and systemic venous (SV) drainage. With the PV technique, the portal vein is anastomosed to the recipient superior mesenteric vein, resulting in direct delivery of graft venous outflow to the liver. The SV technique involves connecting the portal vein to the recipient iliac vein or vena cava. PV drainage of pancreas grafts mimics the venous flow of the native pancreas and is therefore considered more “physiologic”. Indeed, several groups have shown that PV drainage maintains the normal portal-systemic insulin gradient and results in less peripheral hyperinsulinemia as compared to SV drainage (3,4). In addition, studies in experimental models and patients suggest that PV provides immunologic advantages with reduced risk of acute rejection and graft loss as compared to SV drainage (4–7).

SV technique is currently the preferred technique for most surgeons in the United States (8). Studies comparing both techniques have failed to show significant differences in various outcome measures with short follow-up. However, the effect of venous drainage on metabolic or immunologic parameters might require a prolonged follow-up to detect differences in graft outcome or in lipid profile. We compared the long-term outcome of patients with SPK receiving either SV or PV of the pancreas graft.


All SPK transplants at the Toronto General Hospital performed between November 1995 and November 2007 were prospectively entered into our transplant database and analyzed retrospectively. The study was approved by the University Health Network's Research Ethics Board (IRB# 09–1013-BE). All transplants were performed according to our standard protocol with enteric drainage (n = 171) and bladder drainage (n = 21). All patients had a negative AHG-CDC T-cell crossmatch at the time of transplantation. Cytomegalovirus (CMV) prophylaxis (valganciclovir) was given for 3 months to all patients who were CMV positive or who received CMV-positive grafts.

Study groups

A total of 192 patients receiving a SPK transplant were identified. Pancreas-after-kidney transplantrecipients were not included to offer a homogenous study population. The methods of organ recovery, preservation and surgical technique for PV and SV have been previously described (4,9,10). The portal vein was anastomosed end-to-side with 6–0 prolene to the superior mesenteric vein located at the root of the transverse mesocolon for PV drainage and to the distal vena cava for SV drainage. Between 1997 and 2003 PV was the preferred procedure in our program; after 2003, SV became the venous drainage technique of choice. The shift towards SV drainage was influenced by the lack of available evidence showing any metabolic or immunologic disadvantage. The group of surgeons performing pancreas transplantation expanded over the study period. While MC was present during the entire period, IMcG and MS performed pancreas transplantation in Toronto since 2001 and 2005, respectively.

Induction immunosuppression regimens in both groups were analyzed and categorized into three groups: (1) no induction, (2) lymphocyte-depleting therapy, defined by receiving at least one dose of antithymocyte globulin (Thymoglobulin, Genzyme, Mississauga, ON, Canada) and (3) interleukin-2 receptor antagonists therapy, defined by receiving basiliximab (Simulect, Novartis, Mississauga, ON, Canada) or daclizumab (Zenapax, Roche, Mississauga, ON, Canada). Maintenance immunosuppression profile was similar in all patients and consisted of steroids, MMF (CellCept, Roche, 1–2 g/day) (CellCept, Roche, Mississauga, ON, Canada) and tacrolimus (Astellas, Markham, ON, Canada). Steroid therapy began with 500 mg of intravenous methylprednisilone given intraoperatively and was followed by tapered dosage of 200 mg to 20 mg/day over 5 days. Prednisone was continued at 20 mg/day and decreased to 10 mg/day by 1 month and 5 mg/day by 3 months. Tacrolimus was started on postoperative day 3–5 in recipients treated with ATG; dosages were adjusted to maintain whole-blood trough levels of 10–15 μg/L during the first month and 5–10 μg/L thereafter. Cyclosporine was used in 14 patients (SV [n = 9] vs. PV [n = 5]). Some patients in both groups were weaned off prednisone.

Evaluation of graft function

Pancreas graft function was determined by a yearly oral glucose tolerance test (OGTT) in patients who were insulin independent and not taking oral hypoglycemic therapy. Blood samples were drawn immediately before and 2 h after a standard 75 g oral glucose load for determination of blood glucose and serum insulin levels. Blood glucose was measured by the hexokinase method and serum insulin by radioimmunoassay. All missed OGTTs were investigated and the numbers, as well as the reasons for missed tests, were recorded for each group. Impaired glucose tolerance was defined as blood glucose level at 2 h ranging from 7.8 mmol/L to 11.1 mmol/L and diabetes mellitus was defined as level over 11.1 mmol/L (11). In addition, hemoglobin A1c (HbA1c) was determined yearly. Kidney graft function was evaluated yearly by serum creatinine levels. Modification of diet in renal disease estimated glomerular filtration rate (MDRD eGFR) was calculated at the same time points.

Investigation of lipid profile

Low-density lipoprotein (LDL), high-density lipoprotein (HDL) and cholesterol were determined as parameters of the metabolic lipid profile. Body mass index (BMI) was calculated pre- and posttransplant.

Assessment of rejection

Serum creatinine and amylase levels were used to detect kidney and pancreas rejection. When rejection was suspected, a biopsy was performed and graded according to the Banff-97 classification for kidney grafts and the Maryland classification for pancreas grafts (12,13). Acute rejection episodes were treated with steroids initially, and if resistant, with thymoglobulin. Humoral rejections were treated with thymoglobulin, plasmapheresis and intravenous immune globulin as indicated.

Short- and long-term outcome

Short-term outcome was investigated by assessing complications during the first postoperative year, which were graded according to the Clavien score (14). Long-term results were assessed by recipient, as well as pancreas and kidney graft survival. Pancreas graft failure was defined as the need for exogenous insulin therapy or use of oral hypoglycemic agents. Return to chronic dialysis was counted as kidney graft loss.

Statistical analysis

Statistical analysis was performed with the SPSS software version 17.0 (SPSS for Windows 17.0, Chicago, IL, USA), and GraphPad Prism 5.04 for Windows (GraphPad Software, La Jolla CA, USA, www.graphpad.com). A Fisher exact test was used for categorical variables, whereas analysis of variance was performed for continuous parametric variables using a post hoc Bonferoni correction for multiple comparisons. Graft and patient survival were calculated with the Kaplan–Meier survival analysis and compared with the log-rank test. A p-value below 0.05 was considered significant.


Patient demographics

A total of 192 SPK recipients were transplanted between November 1995 and November 2007 at our institution. SV drainage of the pancreas grafts was used in 147 recipients and PV drainage was used in 45 recipients. Both groups had similar characteristics regarding donor age and gender, recipient age and gender, CMV mismatches, maximum panel reactive antibodies (PRA), human leukocyte antigen (HLA) mismatches and warm and cold ischemia times. No differences were present regarding the immunosuppressive regimen. Median follow-up was 75 months for recipients with SV as compared to 129 months for PV patients (p < 0.0001; Table 1).

Table 1.  Donor and recipient characteristics
 SV (n = 147)PV (n = 45)p-Value
 Age29 ± 1028 ± 90.1
 CMV positive46%55%0.25
 Age41 ± 8 38 ± 60.3
 CMV positive34%20%0.7
Recipient BMI (pre-tx)23.9 (22.7–27.6)26.9 (25.5–28.5)0.4
Pancreas CIT (h)10 ± 311 ± 30.8
Pancreas WIT (min)33 ± 734 ± 50.5
Kidney CIT (h)8.7 ± 3 9 ± 30.9
Kidney WIT (min) 32 ± 12 33 ± 180.9
Years of diabetes28.20 ± 6.9 28.45 ± 6.7 0.8
Peak PRA   7 ± 8%    9 ± 16%0.2
HLA mismatches 4.6 ± 1.1 4.7 ± 0.90.7
CMV mismatches32%42%0.2
 Lymph depleting64%62%0.15
Steroid withdrawal21%15%0.6
Median F/U (months)75129<0.0001

Impact of venous drainage technique on short-term outcome

In both groups, a complication occurred in 38% of the patients during the first postoperative year; the mean Clavien Score was 1 ± 1.1 and 1.2 ± 1.0 in the SV and PV groups, respectively (p = 0.7). No differences were present in the two groups regarding urinary tract infections (12% vs. 11%, p = 0.9), pneumonia (6% vs. 3%, p = 0.2), fungal infections (2% vs. 1%, p = 0.3) and CMV infections (12% vs. 7%, p = 0.8). The median length of hospital stay was 11 days versus 10 days in the SV versus PV groups (p = 0.5), respectively. The 3-months rejection rate was identical in the groups (6%) and the 1-year rejection rate was 12.2 versus 13.3% (p = 0.8) in SV versus PV groups, respectively (Table 2).

Table 2.  Complications during first year
 SV (n = 147)PV (n = 45)p-Value
LOH (days; median, range)11 (7–51)10 (5–45)0.5
Overall complications37.5%38%0.9
Infections UTI12%11%0.9
 Soft tissue3%9%0.06
Rejection 3 month6%6%1
Rejection 1 year12%13%0.8

Effect of venous drainage on pancreas and kidney graft function

The total expected number of OGTTs over the entire duration of follow-up for both groups is 1092. The missing number of OGTTs for the entire population is 164, 127 for the SV drainage group and 37 for the PV drainage group (p = NS). In most patients who missed an OGTT, subsequent OGTTs were performed to assess pancreas function. To ensure that participation in the OGTT was not biased, we contacted patients who missed an OGTT, and verified that none had required insulin or oral hypoglycemic therapy (data not shown).

OGTT results for cohorts at each follow-up year were analyzed (Figure 1). The box in each graph shows the median and the 25th and 75th percentiles. The whiskers of the graph show the largest and smallest values. Baseline and 2 h glucose values remained stable in both groups of patients over the 10 years of follow-up. Systemic insulin levels at baseline and at 2 h were variable but generally higher in the SV group as compared to the PV group. Based on the 2 h glucose value, we classified recipient glucose tolerance as normal, impaired or diabetic. We found that the percentage of abnormal glucose tolerance test were similar in the two groups of patients and did not vary significantly over time (Figure 2). Recipients in both groups also maintained similar HbA1c levels over the 10-year study period (Figure 3).

Figure 1.

Analysis of blood glucose and insulin levels in oral glucose tolerance tests. SV = systemic venous drainage; PV = portal venous drainage.

Figure 2.

Stratification of recipients according to 2 h blood glucose (BG) value from oral glucose tolerance tests. S = systemic venous drainage; P = portal venous drainage. Normal (BG < 7.8 mmol/L); Impaired glucose tolerance (BG 7.8–11.1 mmol/L) and Diabetes (BG > 11.1 mmol/L).

Figure 3.

HbA1c values in recipients with SV or PV pancreas grafts.

Kidney graft function was investigated by serum creatinine and calculated MDRD eGFR. Serum creatinine levels in patients with SV versus PV grafts were identical after 1 year (125 ± 20 μmol/L vs. 124 ± 55 μmol/L; p = 0.9), 3 years (130 ± 71 μmol/L vs. 126 ± 50 μmol/L; p = 0.7), 5 years (132 ± 64 μmol/L vs. 156 ± 85 μmol/L; p = 0.2), 7 years (140 ±114 μmol/L vs. 162 ± 110 μmol/L; p = 0.1) and 10 years (142 ±120 μmol/L vs. 168 ± 150 μmol/L; p = 0.09). MDRD eGFR was also similar between SV versus PV recipients after 1 year (56.82 ± 20.32 mL/min/1.73 m2 vs. 54.62 ± 22.19 mL/min/1.73 m2; p = 0.5), 3 years (54.84 ± 19 mL/min/1.73 m2 vs. 58.92 ± 20.36 mL/min/1.73 m2; p = 0.2), 5 years (57.42 ± 21.41 mL/min/1.73 m2 vs. 52.97 ± 22.09 mL/min/1.73 m2; p = 0.3), 7 years (58.12 ± 20.97 mL/min/1.73 m2 vs. 52.96 ± 21.46 mL/min/1.73 m2; p = 0.2) and 10 years (49.43 ±17.74 mL/min/1.73 m2 vs. 57.78±27.20 mL/min/1.73 m2; p = 0.2).

Does the venous drainage affect metabolic parameters after pancreas transplantation?

We investigated whether the venous drainage type affected lipid metabolism. At 5, 7 and 10 years, serum cholesterol, LDL and HDL values were similar in the SV and PV groups (Figure 4). Of note, a high proportion of recipients in both groups received lipid-lowering therapy (95.5%, SV vs. 94.5%, PV) at a similar daily dosage (19.2 ± 10.5 mg/day, SV vs. 20.5 ± 13.6 mg/day, PV; p = 0.6). Steroid administration was comparable at 5 mg/day throughout the follow-up period for both groups, as was the rate of steroid cessation (21%, SV vs. 15%, PV; p = 0.6). Analysis of BMI showed no difference between SV and PV groups after 3, 5, 7 and 10 years (Figure 5).

Figure 4.

Lipid profile in recipients with SV or PV pancreas grafts. Number of observations analyzed comparing SV versus PV groups, respectively, during follow-up intervals of 5, 7 and 10 years, respectively, for cholesterol (85 vs. 37, 54 vs. 34 and 18 vs. 23), HDL (87 vs. 35, 55 vs. 31 and 18 vs. 23) and LDL (85 vs. 34, 54 vs. 34 and 18 vs. 22).

Figure 5.

Recipient body mass index (BMI) with SV or PV pancreas grafts.

Does the type of venous drainage affect long-term rejection rate, graft or patient survival?

Rejection-free survival was similar in the SV and PV groups over 10 years (Figure 6A). Pancreas graft survival for both groups was similar after 1 year (92.5% vs. 91.1%), 5 years (81.8% vs. 75.5%), 7 years (70.9% vs. 68.8%) and 10 years (65.1% vs. 60%; p = NS; Figure 6B). The most common reasons for graft loss in both groups were: death with functioning graft (25%), graft thrombosis (13%), rejection (11%) and duodenal leak (9%).

Figure 6.

Kaplan–Meier plots after SPK transplantation with SV or PV pancreas grafts. (A) Rejection-free survival of kidney and pancreas grafts. (B) Death-uncensored pancreas graft survival. (C) Death-uncensored kidney graft survival. (D) Patient survival (the respective number of patients at risk at 1, 3, 5, 7 and 10 years follow-up in the SV vs. PV groups are 144 vs. 44, 133 vs. 42, 101 vs. 41, 63 vs. 40 and 26 vs. 31).

Similarly, kidney graft survival was not different between the two groups after 1 year (97.2% vs. 95.5%), 5 years (92.9% vs. 84.4%), 7 years (88% vs. 82.2%) and 10 years (81.6% vs. 75.5%; p = NS; Figure 6C). The most common cause of kidney graft loss in both groups was death with functioning graft (61%) and rejection (11%).

No differences were observed regarding patient survival at 1 year (97.9% vs. 95.5%), 5 years (94.3% vs. 88.8%), 7 years (89.4% vs. 86.6%) and 10 years (85.1% vs. 84.4%; p = NS; Figure 6D). Most common causes of death in both groups were myocardial infarction (35%), cerebrovascular accidents (CVA; 13%) and cancer (13%). A total of eight patients died of myocardial infarction; five in the SV group and three in the PV group. Three patients died of CVA; one in the SV group and two in the PV group.


The optimal type of venous drainage in pancreas transplantation has been controversially discussed in the past. In this study, we show that the SV and PV drainage methods provide similar long-term outcome regarding graft survival, graft function, rejection rate and metabolic profile. Although previous studies suggested similar patient and graft survival outcome with either method (15), long-term metabolic and immunologic outcomes were not clearly demonstrated. PV drainage maintains the first pass effect of the liver on insulin uptake, which removes about half of the insulin from the circulation. This effect reduces the systemic hyperinsulinemia that is commonly observed in SV recipients. Hyperinsulinemia may adversely affect lipid metabolism and promote atherosclerosis (16,17), however, its impact in SV drainage recipients has been difficult to delineate in short-term studies (18,19). Our series is one of the first of its kind to study the long-term effects of PV drainage on metabolic parameters. As expected, we documented higher insulin levels with OGTT in the SV group as compared to the PV group at baseline and at 2 h, which was sustained over the 10-year study period. This difference had little effect on lipid profiles or on death from cardiovascular diseases, however. We acknowledge that the use of lipid-lowering medications in both groups of patients may have obscured any difference related to the method of pancreas venous drainage. It is also possible that lipid metabolism adapts to increased systemic levels of insulin leading to peripheral insulin resistance and resistance to the antilipolytic action of insulin (20). Interestingly, this form of systemic peripheral resistance is not accompanied by the hepatic insulin resistance that is commonly seen in patients with insulin resistance syndrome (21).

Improved immunologic tolerance of pancreas grafts resulting in decreased rejection rates has been cited as a second potential benefit of the PV method (6,7). This benefit may arise from the specialized microenvironment of the liver on antigen presentation by Kupffer cells and liver dendritic cells, which can promote immunologic responsiveness (7,22,23). In a nonimmunosuppressed pig model of pancreas transplantation, Tang et al. (24) found that PV grafts had less severe and delayed rejection as compared with SV grafts. In a series of 119 pancreas transplant patients reported by Philosophe et al. (7), the 3-year rejection rate in patients with SV grafts was significantly higher than those with PV grafts (52% vs. 21%, p < 0.001). The series represents an early experience of pancreas transplantation and the rejection rate observed in their study was higher than what is generally experienced today. In addition, the authors did not specify how rejection was diagnosed, an immunologic risk profile of the patients was not provided (HLA mismatches, peak PRA) and pancreas transplants alone as well as SPK transplants were combined. In our study, we did not observe differences in the rejection rate between the two methods of venous drainage. The differences between our results and the previous findings could be related to the different patient population (SPK only in our series) or differences in immunosuppressive regimen. Similar results were obtained by Petruzzo et al. (18) who found in a randomized controlled trial similar graft survival for both methods. Unfortunately, data describing rejection, graft function and metabolic profile was not investigated by the authors.

Our study has several limitations, such as the retrospective study design and the relatively small sample size. In addition, the follow-up was longer for PV patients. We compensated for these potential shortcomings by choosing a homogeneous study group restricted to SPK transplantation. In addition, our protocol for immunosuppression and patient care was constant over the study period making an era effect unlikely. Furthermore, less than 10% of the patients were lost in follow-up minimizing observation bias. Both patient groups had identical immunologic risk profiles making the groups comparable for the rejection analysis.

In summary, we determined that the long-term outcome is similar in SV and PV drained pancreas recipients. Graft function and survival, rejection, lipid profile and cardiovascular outcomes were similar between the two different drainage techniques. The choice of venous drainage should be surgeon preference because it does not influence short- or long-term outcome.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.