Analysis of Factors Associated With Biliary Complications in Children After Liver Transplantation

Abstract

Background 

Biliary complications (BCs) remain an important cause of morbidity after pediatric liver transplantation. Technical factors have already been implicated in the development of BCs. Previous reports have associated the use of partial grafts, particularly living donor grafts, with a higher incidence of BCs. Our aim is to study the factors associated with the development of BCs in a large cohort of pediatric liver transplant recipients.

Methods 

Retrospective cohort study of 670 children (<18 years of age) who underwent a primary liver transplant between March 2000 and January 2015. Patients who did and did not develop BCs were compared with identify associated factors. Univariate and multivariate analyses were performed.

Results 

A total of 115 patients (17.2%) developed BCs (83 strictures and 44 leaks). Of the study participants, 594 had living donor liver transplants. Multiple arterial anastomoses was a protective factor for BCs, and a ductoplasty was a risk factor. Living donor grafts and multiple biliary anastomoses were more frequently associated with leaks. Patients with BCs had a higher reoperation rate and longer hospital stays. There was no difference in patient or graft survival.

Conclusions 

Technical factors play a major role in the development of BCs, particularly leaks. Strictures are more frequently associated with an inadequate arterial supply to the bile duct, and multiple arterial anastomoses may protect children from this complication. The use of partial grafts was not an independent factor for BCs in high-volume centers that are experienced with this technique.

In Brief

This experience from Brazil concerns 670 pediatric liver transplants and represents a vast clinical experience. Interestingly multiple arterial reconstructions are protective for biliary problems consistent with the need for optimal perfusion of the bile duct.

Biliary complications (BCs) are a significant cause of morbidity after pediatric liver transplantation (LT) and significantly increase the liver transplant costs.^1,2 Less-invasive treatments, such as percutaneous transhepatic procedures, are successful in approximately 75% to 90% of patients and prevent graft loss. However, the duration of these procedures is long, and they are not without complications, both of which lead to higher costs.^3-5 Additional surgery may be necessary and is also associated with recipient morbidity.^3,6-8

The incidence of BCs after LT is higher in children than in adult recipients.^1,6,9-11 The small size of the structures and the more common use of technical variant grafts^6,10 have been implicated as factors that contribute to this higher rate of complications. Other factors associated with the development of BCs include hepatic artery thrombosis (HAT), the presence of multiple bile ducts in the liver graft, and the type of biliary reconstruction performed.^6,9,12

The restricted data that were available for a large, single-center cohort in the field of pediatric LT led us to study the factors associated with the development of BCs in our primary transplanted pediatric patients.

MATERIALS AND METHODS

A retrospective cohort study based on a prospectively collected database was conducted. All children (<18 years old) who underwent primary LT between March 2000 and January 2015 at AC Camargo Cancer Center and Hospital Sirio Libanes were included. The primary outcome was the development of BCs (biliary stricture [BS] or bile leak [BL]) after LT. Preoperative and intraoperative variables that were studied for their association with the development of BCs in the univariate analysis included recipient sex, age, weight, and z-scores; pediatric end-stage liver disease/model for end-stage liver disease scores; presence of ascites (yes/no); underlying liver disease; type of LT performed; graft-to-recipient weight ratio; intraoperative blood transfusion (mL/kg); number of arteries in the liver graft and number of arterial anastomoses performed; number of bile ducts in the liver graft and number of biliary anastomoses performed; type of biliary reconstruction; ductoplasty; and cold ischemia time and warm ischemia time. Postoperative events were also analyzed, including the development of HAT, portal vein thrombosis, the need for reoperation, and length of intensive care unit (ICU) and hospital stays.

Technical variant allografts included split livers, livers reduced from deceased donors, and living donor grafts.

Surgical Techniques

The operative techniques for pediatric LT performed by our group have already been described.¹³ Briefly, grafts recovered from deceased donors are preserved with histidine-tryptophan-ketoglutarate solution and prepared in the back table. Split and reduced livers are prepared ex situ. Living donor liver resection has been described previously,¹⁴ and these grafts are also preserved with histidine-tryptophan-ketoglutarate solution. Grafts are implanted using the piggyback technique; in cases in which an anastomosis with the hepatic veins cannot be performed, the classic technique is used. Biliary reconstruction is performed by a Roux-en-Y bilioenteric (BE) anastomosis in cases with an absent extra-hepatic biliary tree or when the common bile duct is not considered suitable for a duct-to-duct (D-D) anastomosis (size mismatch or tension/twisting at the anastomotic site). The parachute technique (posterior wall with prolene 6.0 or 7.0 running sutures and interrupted sutures in the anterior wall with poliodaxone 6.0 or 7.0, depending on the size of the bile duct) is used to perform a BE anastomosis (Figure 1). Duct-to-duct anastomosis is performed in cases with a suitable common bile duct, a size match and no tension in the anastomosis with prolene 6.0 or 7.0 running sutures with the parachute technique already described. All biliary anastomoses are performed under loupe magnification (2.5× and 3.5×).

FIGURE 1: Single-duct anastomosis with the parachute technique. A, illustration. B, trough (D) anastomosis confection.

If more than 1 bile duct is present in the liver graft, it is considered a multiple duct (2 or 3 ducts). Previously, ductoplasty was performed when 2 ducts were close (<5 mm) and included in a single BE anastomosis (Figure 2). However, ductoplasty is no longer performed, and when 2 ducts are close, they are simply included in a single BE anastomosis (Figure 3). When multiple separated ducts are present, multiple anastomoses are performed (Figure 4). Biliary reconstruction is usually performed by a senior surgeon. When other surgeons in the group perform anastomosis, it is under the supervision of a senior surgeon.

FIGURE 2: Ductoplasty.

FIGURE 3: Multiple ducts close together (<5 mm) with a single anastomosis. A, illustration. B, 2 bile ducts close together (arrowheads). C and D, anastomosis confection.

FIGURE 4: Multiple ducts with multiple anastomoses. A, illustration. B, 2 separated bile ducts (arrowheads). C, 2 separate jejunal openings (arrowheads). D, final aspect of the anastomoses.

Arterial reconstruction is performed with microsurgical techniques in living donor LT (LDLT) and reduced and split grafts. In recipients of whole grafts, the decision to use microsurgical techniques for arterial reconstruction is based on the size of the recipient. Arterial anastomosis for all partial grafts is performed with a microscope. For adolescent recipients of whole livers from deceased donors, arterial anastomoses are performed under loupe magnification. Immunosuppression is based on cyclosporine or tacrolimus associated with steroids.

The preoperative living donor imaging evaluation for recipients of left lateral segment (LLS) grafts includes a Doppler ultrasound, and additional imaging exams are only performed if an abnormality is identified. In those cases, no imaging of the biliary tree is performed before donor surgery. In left- and right-lobe donations, preoperative cholangio magnetic resonance imaging is performed. Donation is contraindicated only if an anatomy that will risk the donor's liver remnant is identified. The number of arteries is not a contraindication. If more than 1 artery is present in the liver graft, a backflow test is performed. If there is no backflow, all arterial anastomoses are performed if possible. Even in cases with a good backflow, all efforts are made to reconstruct at least 2 arteries. However, if 1 of the anastomoses jeopardizes the other, the vessels with a better caliber and positioning are selected for the arterial anastomosis.

Diagnosis of BCs

The diagnostic criteria for BCs have been described previously.³ The diagnosis of BLs is mostly clinical and is based on the aspect of the abdominal drain and confirmed by the measurement of bilirubin from the contents of the drain. The diagnosis of BSs is confirmed with percutaneous transhepatic cholangiography/endoscopic retrograde cholangiopancreatography findings.

Statistical Analysis

The mean (±SD) and median (interquartile range [IQR]) were used to summarize continuous data that were compared using t tests or appropriate nonparametric tests when distributional assumptions were in doubt. Categorical variables were expressed as counts and percentages. Differences between groups were assessed by the χ² or Fisher exact tests when needed.

A survival analysis was conducted according to the Kaplan-Meier product-limit estimates, and patient subgroups were compared using a 2-sided log-rank test. A univariate Cox proportional-hazard model of the predictor variables was performed, and those found to be significant at P less than 0.20 were selected for the multivariate analysis. The multivariate analysis was conducted using a forward stepwise selection.

The significance level was set at α = 0.05. All analyses were performed using the SPSS 22.0 statistical package (IBM, Inc., Chicago, IL).

RESULTS

A total of 670 pediatric LT recipients were included. Of these patients, 375 (55.9%) were girls, the median age at LT was 13.7 months (IQR, 8.9-34.3 months), and the median weight was 8.3 kg (IQR, 6.8-12.9 kg). The diagnoses that indicated LT included biliary atresia (413 patients, 61.6%), cryptogenic cirrhosis (38, 5.7%), choledocal cysts (30, 4.5%), Alagille syndrome (25, 3.7%), α-1-antitrypsin deficiency (19, 2.8%), fulminant hepatitis (18, 2.7%), hepatoblastoma (18, 2.7%), primary familial intrahepatic cholestasis (16, 2.4%), tyrosinemia (16, 2.4%), primary sclerosing cholangitis (13, 1.9%), Budd-Chiari syndrome (13, 1.9%), autoimmune hepatitis (11, 1.6%), cholestasis (8, 1.2%), other metabolic diseases (15, 2.2%), other liver tumors (7, 1.0%), and other diagnosis (10, 1.4%).

Of the transplants, 594 (88.7%) were LDLTs. In 668 patients, information about the liver segments used was available. There were 534 (79.9%) LLS, 88 (13.1%) left lobes (LL), 7 (1%) right lobes (RL), and 39 whole liver grafts (5.8%). A total of 76 (11.3%) children were transplanted with deceased donor grafts, including 39 (51.3%) whole liver grafts, 15 (19.7%) reduced liver grafts, and 22 (29%) split liver grafts. Technical variant grafts (reduced, split liver and living donor) comprised 94.2% of the sample.

Bilioenteric anastomosis was the biliary reconstruction performed in 594 of 632 patients (93.9%). Recipients of deceased donor grafts had proportionally more D-D anastomoses performed than did recipients of living donor grafts (21.6% vs 4.3%, P <0.001). Multiple bile ducts in the liver graft were found in 14.9% of the LLSs and 60% of the RLs. Multiple bile ducts were found in 2 recipients (3.4%) of deceased donor grafts and in 85 recipients (15%) of living donor grafts. Multiple arteries were identified in 52.9% of patients with multiple ducts compared with 27.8% of patients with single ducts (P < 0.001). In addition, multiple arterial anastomoses were performed in 31.7% of patients who underwent multiple biliary anastomoses compared with 9.3% of patients who underwent a single biliary anastomosis (P < 0.001). Multiple arterial anastomoses were only performed in recipients of living donor grafts (68 patients).

Of the 670 patients included in the study, 115 (17.2%) developed BCs after LT, with 83 (12.4%) developing BSs (82 anastomotic and 1 intrahepatic) and 44 (6.5%) developing BLs. Twelve patients developed a BL followed by a BS. The median follow-up was 72.5 months.

The incidences of BCs were 17.3% in recipients of living donor grafts and 15.7% in recipients of deceased donor grafts. Our HAT rate decreased overtime: from 2000 to 2005: 4.22%, from 2006 to 2010: 5.43% and from 2011 to 2015: 1.1%. Hepatic artery thrombosis rates did not differ significantly between recipients of deceased donor grafts and living donor grafts (5.3% vs 3.4%, P = 0.33).

Biliary Complications

Table 1 compares the preoperative, intraoperative and postoperative variables of patients who did and did not develop BCs. Pretransplant recipient characteristics did not differ between groups. Among the technical variables studied in the univariate analysis, only multiple arterial anastomoses showed a statistically significant difference between groups, with 12.2% of the patients in the no-BC group undergoing multiple arterial anastomoses compared to 4.4% of the patients in the BC group (P = 0.018). The number of arteries in the liver graft and the type of biliary reconstruction performed showed a trend toward statistical significance, with 32.9% of the patients in the no-BC group undergoing multiple anastomoses versus 23.9% in the BC group (P = 0.07) and 5.2% of the patients in the no-BC group undergoing D-D anastomosis versus 9.6% in the BC group (P = 0.08). An aberrant artery was ligated in 19.5% of the patients in the BC group and in 20.5% in the no-BC group (P = 0.89).

TABLE 1: Recipients' intraoperative and postoperative variables in patients without and with biliary complications

Patients in the BC group underwent more reoperations than did patients in the no-BC group (39.5% vs 13.9%, P < 0.001). The BC group also had longer ICU and hospital stays (Table 1).

Table 2 shows the variables that were entered into the multivariate analysis for the occurrence of BCs. Multiple arterial anastomoses had a protective effect on the development of BCs (odds ratio [OR], 0.38; 95% confidence interval [95% CI], 0.13-1.06; P = 0.06), and a ductoplasty increased the risk of BC development (OR, 2.14; 95% CI, 0.9-5.11; P = 0.08); in both cases, the difference bordered on statistical significance.

TABLE 2: Multivariate analysis for the development of post-LT biliary complications

The 1-year and 5-year patient survival rates in the BC and no-BC groups were 97.4% versus 90.8% and 86.6% versus 85.3%, respectively (log-rank P = 0.75) (Figure 5A). The 1-year and 5-year graft survival rates in the BC and no-BC groups were 91.3% versus 88.6% and 77.8% versus 82.5%, respectively (log-rank P = 0.59) (Figure 5B).

FIGURE 5: Patient (A) and graft (B) survival rates for patients with and without BCs.

Biliary Strictures

There was no difference in pretransplant characteristics between patients who did and did not develop BSs. The number of arteries in the liver graft and the number of arterial anastomoses performed differed between groups in the univariate analysis. Patients in the no-BS group were more likely to have multiple arteries in the liver graft (P = 0.03) and to undergo 2 arterial anastomoses than were those in the BS group (12% vs 2.5%, P = 0.006) (Table 3). Patients in the BS group had more reoperations than did patients in the no-BS group (25.6% vs 17.3%, P = 0.09), and they also had a longer hospital stay (P < 0.001). Patients in the BS group were more likely to have been diagnosed with a previous BL (14.5% vs 5.5%, P = 0.007).

TABLE 3: Recipients’ intraoperative and postoperative variables in patients without and with biliary strictures

Table 4 shows the variables that were entered into the multivariate analysis for the development of BS. Only the presence of a previous BL had statistical significance as a risk factor (OR, 2.25; 95% CI, 1.01-5.02; P = 0.046). The number of arterial anastomoses performed had a protective effect, but the significance was borderline (OR, 0.26; 95% CI, 0.05-1.22; P = 0.08).

TABLE 4: Multivariate analysis for the development of post-LT biliary strictures

The median time to develop a BS was 8.06 months (minimum, 9 days; maximum, 11.4 years).

The 1-year and 5-year patient survival rates in the BS and no-BS groups were 98.8% versus 90.9% and 84.1% versus 85.8%, respectively (log-rank P = 0.66) (Figure 6A). The 1-year and 5-year graft survival rates in the BS versus no-BS group were 93.9% versus 88.3% and 75.8% versus 82.6%, respectively (log-rank P = 0.47) (Figure 6B).

FIGURE 6: Patient (A) and graft (B) survival rates for patients with and without BSs.

Bile Leaks

Table 5 describes the characteristics of patients who did and did not develop BLs. Patients with BLs had lower preoperative z scores for height and age than did patients without BLs (P = 0.08). Regarding the type of LT performed, more patients in the BL group received an LDLT than did patients in the no-BL group (97.7% vs 88%, P = 0.04). In addition, the type of liver segment implanted differed between groups. There was a higher prevalence of RLs in the BL group and a higher prevalence of whole livers in the no-BL group (P = 0.042). The number of bile ducts in the liver graft and the number of biliary anastomoses performed also differed between groups. Patients in the BL group were more likely to have multiple bile ducts in the liver graft (P = 0.04) and to undergo multiple biliary anastomoses (P = 0.05). Patients in the BL group had more reoperations than did patients in the no-BL group (75% vs 14.3%, P < 0.001). They also had longer ICU (P < 0.001) and hospital stays (P = 0.002).

TABLE 5: Recipients' intraoperative and postoperative variables in patients without and with bile leaks

Table 6 shows the variables that were entered into the multivariate analysis for the development of BLs. No single variable retained enough statistical power as an independent risk or protective factor.

TABLE 6: Multivariate analysis for the development of post-LT bile leaks

The 1-year and 5-year patient survival rates in the BL and no-BL groups were 95.5% versus 91.7% and 92.6% versus 85%, respectively (log-rank P = 0.17) (Figure 7A). The 1-year and 5-year graft survival rates in the BL and no-BL groups were 86.4% versus 89.2% and 83.4% versus 81.5%, respectively (log-rank P = 0.88) (Figure 7B).

FIGURE 7: Patient (A) and graft (B) survival rates for patients with and without BLs.

DISCUSSION

Biliary complications remain a cause of morbidity after pediatric LT. The overall incidence varies from 15% to 40%,^6,9,10 which is comparable to the incidence of 17.2% of this series. One of the factors associated with a higher incidence of BCs is the use of technical variant grafts.^15-17 A report from the SPLIT database comparing outcomes of different types of grafts, including 360 LDLTs, showed that this group had an increased incidence of BCs (early and late complications). Patients surviving for 24 months presented an incidence of BCs of 40%. In addition, a 2-fold increase in the incidence of BS has been reported.¹⁰ In a similar analysis in a single center in Belgium,Darius et al⁶ showed a higher incidence of anastomotic strictures in LDLT recipients (14%) compared to whole liver grafts (7%). This higher incidence was also reported by Reding et al.¹⁸ However, other studies did not find this difference regarding technical variant grafts,^6,9 and it was not evident in our analysis. A single-center analysis of 400 LDLTs from Japan reported a BC incidence of 17%.¹⁹ A learning curve effect may explain the initial difficulties with partial grafts.^20,21 The high prevalence of LDLTs in our series (594 patients) and the experience with this technique in our facility may have overcome the risks from the use of partial grafts that have been reported in smaller, single-center studies.

The biliary reconstruction technique and an adequate arterial supply to the bile duct are key to avoiding BCs. Approximately 60% of arterial perfusion to the bile duct comes from the gastroduodenal artery, with 40% downward from the hepatic artery.²² The hilar and intrahepatic ducts are nourished by the peribiliary vascular plexus that arises from the terminal arterial branches.²³ The arterial supply to the bile duct is compromised during LT. The 9 o'clock and 3 o'clock branches that arise from the gastroduodenal artery must be preserved in the recipient operation when performing a D-D anastomosis, and the branches from the right and left hepatic arteries should be preserved in the donor in LDLT and split grafts to ensure adequate supply to the sectioned bile duct. In LDLT, minimal dissection techniques in both the donor and recipient can lower the complications rates, probably due to the preservation of the arterial plexus.^24,25 Multiple arterial anastomoses in grafts with multiple arteries was a protective factor for the development of BCs and BSs in this study. Multiple arterial anastomoses was previously noted as a protective factor for the development of BCs.¹⁹ This finding is in accordance with previous theories about the preservation of the blood supply to the bile ducts because more arterial flow to the branches is ensured when more than 1 artery is reconstructed. It has been suggested that the ligation of aberrant arterial branches could lead to the development of BCs, but this was not confirmed in a study by Egawa et al.¹⁹ In our study, grafts with multiple ducts more frequently presented with multiple arteries. The ligation of aberrant branches, however, did not increase the development of BCs. The development of HAT was already associated with the development of BCs in children,^6,9,12 but it was not sustained as a risk factor in this study. This is likely due to the small incidence of HAT after a routine microsurgical technique was introduced for arterial reconstruction in our group. Flushing and retrograde perfusion of the hepatic artery has been suggested as a beneficial technique to decrease the rate of BCs, but larger studies are needed to confirm its benefits.^6,26

Multiple bile ducts or multiple biliary anastomoses increased the incidence of BCs, particularly BLs, in a study by Salvalaggio and colleagues⁹in pediatric LT recipients. This factor also appeared in our univariate analysis for BL. The attempt to unify 2 adjacent bile ducts by performing a ductoplasty increases the risk of bile duct ischemia. In agreement with previous studies, ductoplasty increased the risk of BCs in our study.^25,27,28 Currently, ductoplasty is avoided in our group. When 2 adjacent bile ducts are present in the partial liver graft, a single Roux-en-Y BE anastomosis is performed that includes the 2 ducts. Stitches are placed in the preserved periductal tissue to hold the suture. When the ducts are separated (>5 mm), 2 separate anastomoses are performed. Further analysis and longer follow-up is needed to better understand the effects of ductoplasty, and if the new adopted technique of performing a single bileoenteric anastomosis including the 2 ducts is superior to the ductoplasty in protecting patients from a BS.

Other technical modifications, such as using wide-interval interrupted sutures, have decreased the incidence of BCs to 4.0%.²⁹ Recently, microsurgical techniques were introduced in biliary reconstruction for LDLT. After surpassing the learning curve, the incidence of BCs went from 21.8% in the conventional technique group to 5.4% in the microsurgery group.³⁰ A later publication by the same group reported an incidence of BCs of 9.6% after routine microsurgery for biliary anastomosis.³¹ This might be the most recent technical refinement to lower BC rates after LT. Microsurgical techniques are only used for arterial reconstructions in our group to date, although a protocol is being established for biliary reconstructions. If other groups adopt this technique, further comparisons and validation of the technique will be available in the future.

Additional technical factors are associated with the development of BLs. The type of donor and the type of liver segment used had a greater impact in the development of BLs in the univariate analysis. There were more recipients of LDLTs in the BL group. Right lobe grafts were also more prevalent in the BL group. Right lobe grafts are more likely to have multiple bile ducts,^21,32 which are also more frequent in patients who develop BLs. Left lateral segments are frequently associated with multiple ducts,³³ and they comprise the majority of segments used in small children. No single factor remained as an independent factor associated with BLs, probably due to the small number of events. A previous BL appears to be an independent factor for the development of a BS, with a 2.25-fold increase. Other studies have already described this risk,^11,34,35 and patients with a previous BL should be followed closely to detect a possible BS in the future.

The type of biliary reconstruction performed is a matter of debate, and studies have shown conflicting results. In pediatric LT, the baseline disease often mandates a Roux-en-Y BE reconstruction. In adults undergoing LDLT, the D-D reconstruction is the current standard technique.^36,37 In children undergoing LDLT, a D-D anastomosis has been associated with the development of BCs,³⁸ particularly BSs.¹² Sakamoto et al³⁹ reported a BC incidence of 47.4% in children who underwent a D-D anastomosis. However, most studies had a small number of patients in whom a D-D was performed.^12,40,41 Other larger studies of LDLT did not find any difference between D-D and BE.^21,42,43 The use of biliary stents is also controversial,⁴² and the presence of a stent can lead to complications, such as leakage, cholangitis, and biliary peritonitis after stent removal.⁴³ Sakamoto et al³⁹ reported a higher incidence of BCs after D-D anastomosis if the stent was not left in place. In our group, biliary reconstruction is performed without a stent, and the type of reconstruction performed was not considered to be a factor that influences the development of BCs.

Biliary complications were previously implicated in decreases in both patient and graft survivals.^25,44,45 However, with adequate management, it has been shown that comparable survival to other recipients can be achieved.^3,6,12 These patients experience greater morbidity, as shown by a higher rate of reoperations and longer ICU and hospital stays. However, neither patient nor graft survival was affected in our cohort.

Because many surgeons in our group perform biliary anastomosis, the surgical factor could not be analyzed. It is clear that individual surgical skills affect the outcome after biliary reconstruction. In addition, long-term follow-up must be taken into account when analyzing 15 years of transplants. The technique and the individual surgical skills have evolved during this period. On the other hand, with a longer follow-up period, more patients with BSs could be identified. In our previous publication, there were patients who took up to 8 years to develop a clinically significant BS.³

In conclusion, BCs remain a significant cause of morbidity following pediatric LT, although they do not affect patient or graft survival. The importance of arterial supply preservation was reinforced with our findings that showed that multiple arterial anastomoses protected children from the development of BCs. Technical factors play a major role in the development of BCs, particularly BLs. Biliary stricture seems to be more related to an inadequate arterial supply to the bile duct and to a previous BL. Improvements in the reconstruction technique through routine microsurgery may help decrease BC rates in the future. The use of technical variant grafts is not a risk for the development of BCs in high-volume centers experienced with partial LT.

ACKNOWLEDGMENTS

The authors thank Rodrigo Tonan for the illustrations.