Telephone: 412-647 4932; FAX: 412-647-9268
Posttransplant biliary complications in the pre- and post-model for end-stage liver disease era
Article first published online: 28 MAR 2011
Copyright © 2011 American Association for the Study of Liver Diseases
Volume 17, Issue 4, pages 428–435, April 2011
How to Cite
Sundaram, V., Jones, D. T., Shah, N. H., de Vera, M. E., Fontes, P., Marsh, J. W., Humar, A. and Ahmad, J. (2011), Posttransplant biliary complications in the pre- and post-model for end-stage liver disease era. Liver Transpl, 17: 428–435. doi: 10.1002/lt.22251
- Issue published online: 28 MAR 2011
- Article first published online: 28 MAR 2011
- Accepted manuscript online: 10 JAN 2011 09:03AM EST
- Manuscript Accepted: 8 DEC 2010
- Manuscript Received: 11 MAY 2010
Biliary complications remain a cause of morbidity after liver transplantation. The aim of this study was to determine whether changes in clinical practice in the era of the Model for End-Stage Liver Disease (MELD) has affected biliary complications after liver transplantation. We retrospectively reviewed all deceased donor liver transplants at a single center. Patients were categorized as pre- or post-MELD (transplant before or after February 28, 2002). A total of 1798 recipients underwent deceased donor liver transplants. Biliary stricture was more common in the post-MELD era (15.4% versus 6.4%, P < 0.001). The strongest risk factors for stricture development were donor age (odds ratio [OR] = 1.01), presence of a prior bile leak (OR = 2.24), use of choledochocholedochostomy (OR = 2.22), and the post-MELD era (OR = 2.30). Bile leak was more common in the pre-MELD era (7.5% versus 4.9%, P = 0.02), with use of a T-tube as the strongest risk factor (OR = 3.38). Surgical factors did not influence the biliary complication rate. In conclusion, even when employing multivariate analysis to allow for factors that may influence biliary strictures, transplant in the post-MELD era was an independent predictor for stricture development. Further studies are warranted to determine the etiology of this increase. Liver Transpl, 2011. © 2011 AASLD.
The occurrence of biliary strictures after liver transplant is a significant cause of morbidity in the liver transplant recipient.1 Despite advances in surgical technique, studies in the 1990s suggested that up to 15% of deceased donor liver transplants (DDLT) are complicated by biliary strictures.2-7
Since February 2002, the Model for End-Stage Liver Disease (MELD) score has been used to determine organ allocation in the United States, by prioritizing organs to sicker patients. A recent small study has suggested that biliary complications are perhaps more common in the MELD era due to several donor, recipient, and technical factors.7
Multiple surgical and nonsurgical risk factors for the development of biliary strictures have been postulated. Technical factors that have been implicated include placement of a T-tube,5, 7-10 occurrence of bile leak,3, 11 hepatic arterial complications,2, 12, 13 and the type of biliary reconstruction.12, 14, 15 Patient factors include older donor age,7 cytomegalovirus (CMV) status, and donor-recipient Rhesus incompatibility.16-20 Many of these observations were noted at a time of considerable change in transplant surgical technique and immunosuppression regimens.
Liver transplantation has excellent outcomes, with 1- and 5-year survival rates of 85%-90% and 70%-80%, respectively.21-25 However, the procedure is limited by the availability of deceased donor organs. With no foreseeable decrease in patients on the liver transplant waiting list, maximizing the life of each allograft is critical. Allograft loss can be due to posttransplant biliary complications, and identifying factors associated with these complications may enable corrective measures to be taken, as is the case with avoidance of older donors in patients undergoing liver transplant for hepatitis C.26
The advent of the MELD system has resulted in sicker patients undergoing transplantation, but it has also coincided with the acceptance in the transplant community of the use of more extended criteria donor or marginal organs. This paradigm shift has reduced the number of patients dying while waiting for liver transplant,27 but runs the risk that certain patients may have a worse outcome after undergoing transplant.28, 29
In light of these findings, the aim of the present study was to determine whether the introduction of the MELD system has affected the development of biliary complications, primarily anastomotic biliary stricture, in a large cohort of patients who underwent DDLT at a single center.
PATIENTS AND METHODS
We retrospectively reviewed the records of all adult patients who underwent DDLT using a prospectively collected database at a single center from January 1, 1997, through December 31, 2008. MELD era was defined by the introduction of the MELD allocation system on February 28, 2002. All transplants prior to this date were noted as being in the pre-MELD era and all transplants after were noted as being in the post-MELD era.
More than 20 independent variables were collected regarding potential donor, recipient, and technical risk factors for biliary complications. This included recipient age, sex, ethnicity, and body mass index (BMI); donor age, sex, ethnicity, and BMI; etiology of liver disease; Child-Turcotte-Pugh score; MELD score; type of surgical anastomosis; type of suture (running versus interrupted); surgeon performing the biliary anastomosis; use of a T-tube; donation after cardiac death (DCD); cold ischemia time (CIT); warm ischemia time (WIT); total operating time; Rhesus status; presence of bile leak; MELD era; patient and donor CMV status; and CMV mismatch (defined as different patient and donor CMV status). A Roux-en-Y hepaticojejunostomy was used at the discretion of the surgeon, mainly in patients with primary sclerosing cholangitis. Otherwise, patients underwent a duct-to-duct anastomosis. An internal stent alone was used at the discretion of the surgeon, although these were used in <1% of the cases. Donor organs were preserved in University of Wisconsin solution. Almost all patients, in both study groups, were placed on tacrolimus-based immunosuppression.
Endoscopic retrograde cholangiopancreatogram (ERCP) or percutaneous transhepatic cholangiogram (PTC) was performed for clinical suspicion of a stricture. Indications for ERCP or PTC included abdominal pain, fever, jaundice, and abnormal liver function tests. An anastomotic biliary stricture was defined as any obstructive lesion at the anastomosis detected postoperatively by cholangiography that required a therapeutic intervention, such as stent placement/exchange or balloon dilation. A size mismatch was not considered evidence of a stricture. Nonanastomotic strictures were not evaluated as a primary endpoint in this study, due to uncertainty regarding their definition as well as potential overlap of this condition with recurrent primary sclerosing cholangitis.
A bile leak was defined as evidence of a leak seen on postoperative imaging (confirmed as bilious based on drain output or percutaneous drainage) or seen at the time of ERCP or PTC. The study was approved by the University of Pittsburgh Institutional Review Board.
Categorical variables were expressed as a percentage, and continuous variables were expressed as a mean ± standard deviation (SD). A chi-square or Student t test were used for comparison of categorical and continuous variables, respectively.
A multivariate logistic regression analysis for the outcomes of bile leak and stricture was performed on variables found to be statistically significant using Student t test or chi-square analysis (P < 0.05) and on variables that had other potential biological or clinical relevance, including donor and recipient age, CIT, WIT, and hepatic arterial thrombosis (HAT).
Survival analysis regarding the liver allograft was performed by creating a Kaplan-Meier curve. The log-rank test was used to compare differences among survival curves. All statistical analyses were performed with Stata software, version 10 (Stata Corp., College Station, TX). Variables in all analyses with a P value of less than 0.05 were considered significant.
Demographics and Clinical Outcomes
The baseline recipient, donor, and technical characteristics of the 1798 transplants for the study are listed in Tables 1 and 2. Prior to the introduction of the MELD score there were 705 transplants and 1093 transplants in the post-MELD era.
|Patient Characteristics||Pre-MELD (n = 705)||Post-MELD (n = 1093)||P Value|
|Recipient age (years), mean (±SD)||51.4 ± 10.2||53.6 ± 10.8||NS|
|Recipient sex (male), n (%)||444 (62.9%)||716 (65.5%)||NS|
|Recipient ethnicity (Caucasian), n (%)||649 (92%)||1008 (92.2%)||NS|
|Recipient BMI (kg/m2), mean (±SD)||27.6 ± 6.2||28.3 ± 5.8||NS|
|Recipient CMV-positive, n (%)||171 (24.3%)||751 (68.7%)||<0.001|
|Hepatitis C, n (%)||211 (29.9%)||380 (34.7%)||0.04|
|Alcohol, n (%)||130 (18.4%)||234 (21.4%)||NS|
|Nonalcoholic steatohepatitis, n (%)||12 (1.7%)||104 (9.5%)||<0.001|
|Cryptogenic, n (%)||45 (6.4%)||54 (4.9%)||NS|
|Primary sclerosing cholangitis, n (%)||48 (6.8%)||57 (5.2%)||NS|
|Primary biliary cirrhosis, n (%)||42 (5.9%)||40 (3.6%)||NS|
|Hepatitis B, n (%)||29 (4.1%)||26 (2.4%)||NS|
|Autoimmune hepatitis, n (%)||43 (6.1%)||42 (3.8%)||NS|
|Acute liver failure, n (%)||16 (2.3%)||39 (3.6%)||NS|
|Other, n (%)||124 (18%)||117 (10.7%)||0.02|
|Child-Turcotte-Pugh score||8.9 ± 1.8||8.9 ± 1.8||NS|
|MELD score||18.6 ± 8.9||20.3 ± 8.8||0.04|
|Donor age (years), mean (±SD)||40.6 ± 17.4||46.9 ± 18.8||<0.001|
|Donor sex (male), n (%)||385 (54.7%)||626 (57.3%)||NS|
|Donor ethnicity (Caucasian), n (%)||623 (88.3%)||889 (81.3%)||0.003|
|Donor BMI (kg/m2), mean (±SD)||25.8 ± 8.1||27.2 ± 9.7||0.002|
|Donor CMV-positive, n (%)||425 (60.3%)||669 (61.2%)||NS|
|CMV mismatch, n (%)||376 (53.3%)||495 (45.3%)||0.001|
|Characteristic||Pre-MELD (n = 705)||Post-MELD (n = 1093)||P Value|
|Duct-to-duct anastomosis, n (%)*||547 (84.4%)||974 (97.3%)||<0.001|
|T-tube, n (%)||500 (70.9%)||636 (58.2%)||<0.001|
|Running suture, n (%)†||611 (90.5%)||314 (90.5%)||NS|
|Interrupted suture, n (%)†||70 (10.3%)||33 (9.51%)||NS|
|DCD, n (%)||24 (3.4%)||107 (9.8%)||<0.001|
|CIT (minutes)||664.6 ± 202||641.59 ± 176||0.02|
|WIT (minutes)||45.1 ± 14.6||30.3 ± 20.8||0.001|
Recipient characteristics were similar between the pre- and post-MELD era except for etiology of liver disease. The mean age of recipients was approximately 52 years, with two-thirds being male and the vast majority Caucasian. The mean BMI among recipients was 28 kg/m2. Child-Turcotte-Pugh score was similar, but MELD score was significantly higher in the post-MELD group. There were more CMV-positive recipients in the post-MELD group. Hepatitis C was slightly more common in the post-MELD era, making up 34.7% of the recipients, and nonalcoholic steatohepatitis was much more common in the post MELD era (9.5% versus 1.7% pre-MELD). However, there were fewer “other” diagnoses in the post-MELD era, suggesting at least some of this difference was due to increased recognition and diagnosis of nonalcoholic steatohepatitis in the latter period.
Donor characteristics differed significantly. In the post-MELD era, donors were older (46.9 versus 40.6 years), heavier (BMI = 27.2 versus 25.8 kg/m2), and more likely to be nonwhite in terms of ethnicity. The proportion of males in each group was similar, as was donor CMV status. There was significantly less CMV mismatch in the post-MELD group.
Surgical characteristics at the time of surgery also differed depending on MELD era. In the post-MELD era, duct-to-duct anastomosis was more commonly performed. In addition, there was more use of DCD and less use of a T-tube. Ischemia times were significantly shorter in the post-MELD era. There was no difference regarding use of running or interrupted sutures between the pre- and post-MELD groups.
Table 3 shows the clinical outcomes after transplant. HAT was significantly more common in the post-MELD era, but the rate of primary nonfunction did not differ. Bile leak was more common in the pre-MELD era (7.5% versus 4.9%). The incidence of anastomotic biliary stricture, however, was significantly greater in the post-MELD era (15.4% versus 6.4%). Mean time to development of bile leak was 102.4 days but only 7.5% of leaks occurred within 2 weeks of transplantation. Mean time to diagnosis of stricture development was 244.2 days but was significantly shorter in the post-MELD era (223.7 days) compared to the pre-MELD era (374.6 days) (P < 0.05).
|Outcome||Pre-MELD (n = 705)||Post-MELD (n = 1093)||P Value|
|Mortality, n (%)*||323 (45.8%)||379 (34.7%)|
|Hepatic artery thrombosis, n (%)||18 (2.6%)||104 (9.5%)||<0.001|
|Biliary stricture, n (%)||45 (6.4%)||168 (15.4%)||<0.001|
|Bile leak, n (%)||53 (7.5%)||54 (4.9%)||0.024|
|Primary nonfunction, n (%)||23 (3.3%)||22 (2.01%)||NS|
Predictors of Anastomotic Biliary Stricture
Table 4 compares the risk factors for the development of anastomotic biliary strictures. Donor age, duct-to-duct anastomosis, prior bile leak, and transplant in the post-MELD era were all significantly associated with stricture development. The etiology of liver disease was not associated with anastomotic stricture development.
|Risk Factor||No Stricture (n = 1585)||Stricture (n = 213)||P Value|
|Patient age (mean ± SD)||52.6 ± 10.6||53.3 ± 10.7||NS|
|Patient sex (male)||64.25%||68.27%||NS|
|Patient race (white)||91.7%||96.3%||NS|
|Patient BMI (mean ± SD)||28.16 +/- 5.9||27.01 +/- 5.6||NS|
|Child-Turcotte-Pugh score (mean ± SD)||8.93 ± 1.81||8.97 ± 1.82||NS|
|MELD score (mean ± SD)||19.62 ± 8.9||19.28 ± 8.05||NS|
|Donor age (mean ± SD)||44.04 ± 18.2||48.16 ± 17.2||0.004|
|Donor sex (male)||56.8%||56.9%||NS|
|Donor race (white)||83.3%||83.6%||NS|
|Donor BMI (mean ± SD)||26.62 ± 9.3||27.38 ± 7.7||NS|
|CIT (mean ± SD)||650.2 ± 188.5||653.4 ± 180.5||NS|
|WIT (mean ± SD)||36.2 ± 14.4||35.8 ± 39.4||NS|
|Hepatic arterial thrombosis||6.91%||6.10%||NS|
|Transplanted post-MELD era||58.4%||78.9%||<0.001|
Further multivariate logistic regression analysis was performed on those variables found to be significant in parametric testing for anastomotic stricture development (Table 5). We also included CIT and WIT, due to previous literature demonstrating their importance in posttransplant biliary complications. The factors that were most significantly associated with development of a stricture include use of duct-to-duct anastomosis (odds ratio [OR] = 2.22), occurrence of prior bile leak (OR = 2.24), donor age (OR = 1.01), and transplant in the post-MELD era (OR = 2.30). To determine whether the MELD era variable may have masked the impact of other factors, we repeated the analysis after removing MELD era (Table 6). The results were similar, with donor age, prior bile leak, and duct-to-duct anastomosis still significantly associated with anastomotic stricture development. However, the use of DCD now became just significant with an OR of 1.71 (P = 0.043). We also analyzed the data using 3 discrete MELD categories (MELD 6-20, 21-30, >30) rather than the MELD score as a continuous variable and found no association between stricture formation and MELD category.
|Risk Factor||Odds Ratio||P Value||CI|
|Risk Factor||Odds Ratio||P Value||CI|
Although it was not one of the primary endpoints, we looked at nonanastomotic stricture (NAS) formation, because the use of DCD was not significantly associated with overall stricture formation. NAS, defined as stricture(s) not at the anastomosis detected postoperatively by cholangiography, were noted in 1.83% of DDLT recipients and were more likely to occur in nonwhite recipients (P = 0.01), and if a DCD was used (P < 0.001). Pretransplant intensive care unit stay (P = 0.029), and use of a T-tube (P = 0.048) were also associated with NAS formation. A subsequent multiple logistic regression analysis demonstrated use of a DCD (OR = 7.25, P < 0.001) and incidence of pretransplant intensive care unit stay (OR = 3.31, P = 0.004) to have a significant association with occurrence of nonanastomotic strictures.
Predictors of Bile Leak
For bile leak, the only variables found to be a significantly associated with this outcome were use of a T-tube and transplantation in the pre-MELD era. However, on multivariate analysis adjusting for T-tube use and transplantation in the pre-MELD era, we found that only use of a T-tube remained a significant risk factor for bile leak development (OR = 3.38, confidence interval = 1.76-7.62).
Allograft Survival in Patients With Anastomotic Strictures
A Kaplan-Meier curve was created to estimate allograft survival among 4 patient groups: pre-MELD with and without stricture and post-MELD with and without stricture (Fig. 1). Log-rank testing demonstrated no significant difference among the 4 patient groups regarding graft survival.
The present study of 1798 patients who underwent DDLT in a period straddling the introduction of the MELD organ allocation system suggests there has been a significant increase in biliary anastomotic stricture formation from 6.4% to 15.4% in the post-MELD era. The incidence of stricture formation is similar to previous studies during the pre- and post-MELD era,2-6, 30, 31 which document an incidence ranging from 5%-16.3%. Of the independent variables analyzed, transplant in the post-MELD era was the most predictive factor for biliary stricture formation.
The biliary anastomosis has been considered a particularly problematic area of liver transplantation, and despite advances in surgical technique continues to cause significant morbidity. The present study is similar to other studies over the last 20 years in that several factors, surgical and nonsurgical, were associated with strictures including a duct-to-duct anastomosis, older donors, prior bile leak, and use of DCD in the case of NAS.7, 9
The importance of classifying strictures as anastomotic versus nonanastomotic is illustrated by the lack of an association between anastomotic strictures and DCD (Table 5). A recent study comparing DCD to heart-beating donors found a very high relative risk for NAS with DCD, but no difference in anastomotic strictures.32 The authors speculated that the increase in NAS with the DCD was due to ischemia/reperfusion injury, a hypothesis that has previously been described,33, 34 and is strengthened by animal studies showing biliary epithelium is more sensitive to ischemia than are hepatocytes.35, 36 The present study shows that anastomotic strictures are more common in the post-MELD era, coinciding with an increase in the use of DCD organs, although the initial regression analysis failed to show an independent association between DCD use and stricture formation (Table 5). However, removing the MELD era variable in the analysis (Table 6) demonstrated a significant association suggesting that some of the increase in stricture incidence is related to the use of these expanded criteria organs.
Other studies regarding outcomes using a DCD have confirmed the risk of biliary stricture formation.37-39 One recent study has disputed this, but it was limited due to a very small number of patients.40 As seen in the current study, neither the use of DCD nor the development of strictures equates with worse graft survival.41 However, it would seem prudent to judiciously use DCD.
In comparing the pre- and post-MELD eras, it was evident that there were a few minor differences in recipient characteristics, but these were unlikely to affect outcome. In contrast, donor characteristics were significantly different. Older donor age and higher donor BMI were apparent in the post-MELD era. Donor age may be related to biliary complications but the data is conflicting. In the study by Welling et al., a statistically significant older donor age was found in those with postoperative strictures (38.2 years versus 32.5 years),7 but other studies have failed to demonstrate such an association.3-6 In our study, we found neither donor age nor BMI to be associated with stricture development.
Surgical factors, including the type of biliary reconstruction and occurrence of bile leak, can also be predictive of stricture development. Shorter ischemia times, less T-tube use, and greater proportion of duct-to-duct anastomoses occurred in the post-MELD era. Among these variables, the type of reconstruction was associated with stricture rate, as 93% of recipients with a stricture had undergone a duct-to-duct anastomosis versus 82% in those without a stricture (P = 0.004). Some older studies have suggested an opposing view. The series described by Greif et al. from the University of Pittsburgh, from 1988 to 1991, demonstrated strictures to be more common after a hepaticojejunostomy than a duct-to-duct anastomosis (52.9% and 36.4%, respectively).3 Other reports published in the 1990s appear to support these findings.14, 31
Because of concern that biliary anastomotic suture technique could influence the incidence of stricture formation, we compared the use of running and interrupted sutures and their association with stricture development. Suture data were available in the majority of cases, and there was no difference in the use of running and interrupted sutures between the pre- and post-MELD era (Table 2). Furthermore, suture technique was not associated with anastomotic stricture formation (Table 4).
Other surgical risk factors that may influence stricture formation include surgeon experience and technique. Most of the biliary anastomoses were performed by surgical fellows under the direct supervision of an attending surgeon. On review of operative reports and questioning surgeons, it was apparent that each surgeon used the same technique for biliary reconstruction consistently. We reanalyzed the logistic regression including the attending surgeon as a variable and found no association between surgeon and biliary stricture formation, suggesting that surgical technique was not a relevant factor in the increasing stricture rate seen in the post-MELD era.
Similarly, we evaluated whether the technique employed for liver implantation and biliary anastomosis may have affected the biliary complication rate. In the post-MELD era, there was a trend toward creating an anastomosis between the donor's celiac axis patch and the recipient's hepatic artery-gastroduodenal artery confluence. However, there was no association between the vascular anastomosis and stricture formation on chi-square or multivariate analysis, although we did not have information on individual donor arterial variation such as replaced hepatic artery from the superior mesenteric artery. Regarding bile leak, we found that this was more common in the pre-MELD era and attribute this finding to the greater use of T-tubes. The use of T-tubes has clearly been demonstrated to increase risk of postoperative bile leak. However, studies have shown that when used with a duct-to-duct anastomosis, the incidence of strictures tended to decrease, although without statistical significance.8-10 The fact that so few of the leaks were seen early after transplant (only 7.5% occurred within 2 weeks of transplant) and that the mean interval from transplant to leak was 102 days further indicates that most leaks occurred at the time of T-tube removal.
In the present study, we also found that a lower proportion of patients had T-tube placement in the post-MELD era. Although this was associated with a lower incidence of bile leak, stricture formation was more common in patients with a prior bile leak, which is in accordance with previous findings.7, 42, 43 It is reassuring that allograft survival was not affected despite the increased incidence of stricture formation, confirming the literature consensus that endoscopic treatment of biliary complications is very effective.44, 45
The explanation for the increased anastomotic stricture rate seen in the post-MELD era is unclear but donor organ quality has to be considered. The transplant community has accepted that marginal or expanded criteria organs are necessary to increase the pool of donor organs but outcome may be adversely affected. This has been recently shown in the case of transplantation because of hepatitis C infection.29
In the current study, there was a significant increase in donor age in the post-MELD era and also a significantly higher BMI, suggesting the possibility of increased allograft steatosis and also microvascular disease. A seminal article by Hoofnagle et al. that looked at donor age and liver transplantation in the 1990s demonstrated that older donors (and this was defined as age >50 years) were more likely to function poorly immediately after transplant.46 This was manifested by higher aminotransferases and poor bile production after transplant and equated with excess 3-month mortality and allograft loss. In the era where the 60- to 70-year-old organ is not uncommon and where significant donor steatosis is frequent, perhaps one of the consequences is biliary ischemia and hence stricture formation from problems with the biliary microcirculation, as a recent study has suggested.47
The HAT rate was higher in the post-MELD era and may have been partially explained by the increased donor age as has recently been demonstrated.48 HAT is associated with biliary strictures but typically the nonanastomotic type.49 However, on chi-square analysis there was no difference in HAT incidence between those with and without stricture. Furthermore, on multivariate logistic regression analysis, the post-MELD era was still significant even allowing for HAT. It is possible that despite no obvious HAT, some patients experienced a low flow state in the hepatic artery which contributed to stricture formation due to biliary ischemia.
There are several limitations to the current study. Although the data were prospectively collected, the analysis was retrospective in nature. In addition, despite the large numbers, the study was from a single center employing or omitting surgical techniques used at other centers. For instance, it essentially excluded some factors that may reduce stricture formation such as internal stents,7 and some data, such as the type of suture used, were incomplete. We did not have data on individual arterial and biliary anatomy that may have influenced stricture formation.
In conclusion, since the introduction of the MELD score there has been an increase in the rate of biliary complications after DDLT. The decreased incidence of bile leak was certainly due to less use of a T-tube. However, the incidence of anastomotic biliary strictures has increased in the post-MELD era, even when allowing for surgical factors that could influence stricture formation such as duct-to-duct anastomosis, HAT, prior bile leak, and surgical technique, suggesting that changes in nonsurgical factors are responsible. Donor age was associated with increased stricture formation but because this was included in the multivariate model, other changes in clinical practice in the post-MELD era that were unaccounted for in our analysis, such as greater use of extended criteria donor organs, may be responsible. Further studies are warranted.
- 26The difference in the fibrosis progression of recurrent hepatitis C after live donor liver transplantation versus deceased donor liver transplantation is attributable to the difference in donor age. Liver Transpl 2008; 14: 1778-1786., , , , , , et al.