All consecutive patients who underwent LT with duct-to-duct anastomosis between January 2010 and December 2019 were considered for inclusion. All patients who underwent a bilio-enteric reconstruction and those lost in follow-up or requiring early retransplantation after LT were excluded. Additionally, to avoid competing risks of early mortality due to causes other than ABS, patients who died within 90 days post-LT owing to ABS-unrelated causes were excluded. Recipients were divided into two groups based on the occurrence of ABS or not.

Data were retrieved from electronic medical records and from the prospectively maintained CRISTAL on-line data base from the Biomedicine Agency, approved by the French Data Protection Authority (Commission Nationale de l’Informatique et des Libertés) (Decision n◦96-025 of March 19 1996). The following data were collected: recipients’ characteristics at the time of transplantation, donors’ characteristics, intraoperative data, and postoperative outcomes. Extended criteria donors were those older than 75 years, or with confirmed steatosis >30%. All postoperative complications occurring within 90 days after surgery were collected and graded according to the Dindo–Clavien classification. In patients with multiple complications, the highest grade was retained for analysis. Regarding anastomotic biliary complications other than ABS, anastomotic biliary fistula was defined according to the ISGLS definition (12). Early allograft dysfunction was defined according to the current definition (13). Early rejection corresponded to a histological diagnosis defined upon Banff criteria within the first 3 months after LT (14). Medical complications including extra-abdominal infection, CMV infection, or reactivation and acute kidney injury were also collected.

All grafts were procured from a brain-dead donor using the en bloc technique (15) unless the pancreas was simultaneously procured for organ transplantation. Full dissection of the graft hepatic pedicle was next carried out on a hypothermic back table using an ice basin filled with cold preservation solution. Care was taken not to dissect above the gastroduodenal artery to prevent any proper hepatic artery lesion. Similarly, cholecystectomy was performed at this stage and the common hepatic duct was bluntly dissected and divided as low as possible after pancreatic head removal to avoid any injury or devascularization.

Regarding biliary reconstruction after vascular implantation, duct-to-duct anastomosis was performed using a 6/0 or 7/0 polyglyconate long-term monofilament absorbable sutures (Maxon™, Covidien, Medtronic, Watford, United Kingdom). Both graft and recipient bile duct were trimmed to length to ensure a tension-free anastomosis between two appropriately vascularized ducts. End-to-end anastomosis was then fashioned using a posterior running suture and anterior interrupted suture or two posterior and anterior running sutures, systematically knotted on the outside. In case of anticipated anastomotic difficulty, an anterior ductoplasty technique or a T-tube insertion could be used at the discretion of the transplant surgeon. Anastomotic difficulty was anticipated when the graft and/or the recipient bile duct diameter was smaller than 5 mm or when a donor-recipient duct size mismatch larger than 4 mm was present.

Systematic Doppler ultrasounds were performed daily from postoperative day one to five, then once a week to detect any vascular complication. Pre-transplantation screening of donors and recipients for CMV serological status (IgG) defined the strategy employed for the prevention of CMV reactivation or primary infection. A 6-month CMV prophylaxis was routinely given to “high-risk” recipients defined as seronegative recipients receiving a graft from a seropositive donor. The pre-emptive strategy was applied to other patients with routine determination of CMV viremia by sensitive assay (molecular diagnosis). Of note, in this situation, CMV antigenemia assay and qPCR were weekly checked from LT to patient discharge, then once a month for the first year. In case of positive CMV viremia, whether symptomatic or not, CMV therapy was systematically initiated to prevent progression to clinical disease.

The diagnosis of ABS was suspected on the presence of a size discrepancy at the site of the bilio-biliary anastomosis with or without upstream bile duct dilatation on imaging (ultrasound, cholangiography, CT scan, or MR-cholangiography). This had to be associated with a concomitant cholestasis and/or an elevated serum bilirubin after excluding other cholestasis causes such as graft rejection and viral reactivation. Each patient with suspected ABS underwent an endoscopic retro-grade cholangiopancreatography (ERCP) to confirm ABS. In case of ABS confirmation, a plastic or self-expandable metallic stent was placed at the discretion of the endoscopic team. Endoscopic stent replacement was repeated each 4–6 months until ABS clearance.

The χ² test or Fischer exact test was used for analysis of categorical variables, as appropriate. Continuous variables with a normal distribution were presented as mean (standard deviation) and non-normally distributed variables as median (range); t test and Mann–Whitney U test were used for statistical analysis. All perioperative variables associated with the occurrence of ABS in univariable analysis (p < 0.100) were included in a binary logistic regression model to identify independent risk factors of ABS. Backward selection was used, with a 0⋅1 cut-off for entry into the model. In case of collinearity between variables, only the most relevant variable was included in the model. Regarding postoperative variables, given the potential time-dependent relationship between the occurrences of postoperative events, only those occurring before the occurrence of ABS were deemed of interest and were retained in multivariable analysis. Performance of the multivariable model was assessed in terms of discrimination, expressed as the area under the curve (AUC) ± standard error (SE). Additionally, overall and graft survival estimates were calculated using the Kaplan–Meier method. OS and graft survival corresponded to the interval between LT and the date of last follow-up or death and between LT and date of graft failure. Survival differences between patients who did and did not experience ABS were compared using the log rank test. All p values were based on two-tailed statistical analysis and p < 0⋅050 was considered to indicate statistical significance. Analyses were performed with SPSS^® software, version 27.0 for Windows^® (IBM, Armonk, New York, United States). The present study complied with the STROBE Guidelines (16).

Over the study period, 288 LT were performed, of which 48 LT were excluded (ABS-unrelated 90-day mortality in patients with bilio-biliary reconstruction, n = 25; bilio-enteric reconstruction, n = 18; early retransplantation, n = 5) and 240 LT were included (Figure 1). Recipients’ characteristics are listed in Table 1. Mean recipient age and mean MELD score were respectively 55.7 years old and 21.2 at the time of transplantation. Donors’ characteristics are listed in Table 2. Mean donor age was 57.6 years old. Most grafts (90.4%) were allocated according to the standard national liver score system. ECD was used in 141 LT (58.8%). Main causes for brain death were stroke (58.2%) and trauma (24.6%) and nearly one third (28.7%) of donors presented cardiac arrest. Pre-transplantation donors’ and recipients’ serological CMV status are described in Table 2.

Intraoperative data are reported in Table 3. Mean cold ischemia duration was 533.2 min. Anastomotic technical difficulty as defined above was encountered in 81 LT (33.8%). Both biliary ductoplasty (n = 45; 18.8%) and T-tube placement (n = 46; 19.2%) were not performed routinely. Owing to significant bleeding (n = 4) or failure to fascial closure (n = 3), an open abdomen approach with negative wound therapy was adopted at the end of LT in seven patients (2.9%), of which one had a delayed biliary reconstruction.

Among all transplanted patients with bilio-biliary reconstruction over the study period (n = 288), the 90-day mortality rate was 10.4% (n = 25). Causes of 90-day mortality are listed in Supplementary Table S1. No patient died due to an anastomotic biliary fistula or stricture after a bilio-biliary reconstruction. Among excluded patients, one patient with bilio-enteric reconstruction died of multiorgan failure due to an anastomotic biliary fistula.

Surgical reintervention within 90 days after LT was needed in 34 patients (14.2%), mostly for hemorrhage (n = 24), anastomotic biliary fistula (n = 8), arterial complication (n = 3), large for size syndrome (n = 2), and portal vein thrombosis (n = 1). Postoperative outcomes are displayed in Table 4.

Median follow-up was 49 months (range, 7–126). Overall ABS rate was 27.1% (n = 65), of which 20 (33.8%) developed within 90 days after LT. Median time to ABS diagnosis was 142 days (range, 13–1,265).

Upon univariable analysis, ABS risk factors related to donor, recipient, intraoperative characteristics, and postoperative outcomes are listed in Tables 1–4. Recipient age (p = 0.090), donor BMI (p = 0.015), and donors’ and recipients’ serological CMV status (p = 0.030) were statistically associated with ABS. Intraoperatively, the existence of anastomotic technical difficulty was associated with ABS (p = 0.005). Regarding postoperative events, both the occurrence of anastomotic biliary fistula (p < 0.001) and a CMV infection (p < 0.001, Figure 2) were associated with ABS.

Upon multivariable analysis, elevated donor BMI (OR = 0.509, CI95% 0.270–0.959; p = 0.037), postoperative CMV primoinfection (OR = 5.244, CI95% 2.281–12.054; p < 0.001) or CMV reactivation (OR = 2.421, CI95% 1.192–4.920; p = 0.015) and the occurrence of anastomotic biliary fistula (OR = 2.691, CI95% 1.162–6.233; p = 0.021) were independently associated with ABS, although anastomotic technical difficulty did not reach a statistically significant association (Table 5). Discrimination of the estimated risks from the multivariable analysis was deemed acceptable (AUC, 0.740; SE, 0.035, p < 0.001).

Among 65 patients who developed ABS, 46 (70%) experienced postoperative CMV infection, of which eight (17%) were under CMV prophylaxis. Among these 46 patients, 34 patients (75%) developed first postoperative CMV infection before experiencing ABS diagnosed after a median of 172 days after LT (range, 18–1,222). In contrast, 12 patients (25%) developed first ABS with a median interval after LT of 131 days (range, 30–447) before presenting CMV infection.

Similarly, 31 (12.9%) patients experienced anastomotic biliary fistula including three grade A (10%) and 22 grade B (71%) all managed endoscopically, either alone (n = 15) or combined with a percutaneous drainage (n = 7). The remaining six patients (19%) were deemed grade C as they required reoperation. Of them, 15 (48%) developed subsequent ABS after a median interval between anastomotic biliary fistula and stricture of 41 days (range, 5–239).

First-line ABS treatment was systematically endoscopic (100%), requiring a median treatment duration of 252 days (range, 133–912) for a median number of two (range, 1–11) procedures per patient. Twelve patients (18.5%) eventually developed ABS recurrence, of which nine were managed endoscopically, two percutaneously, and one surgically. Repeat LT was not required due to ABS but was performed in 11 patients owing to ischemic cholangitis (n = 5), acute hepatic artery thrombosis (n = 3), and chronic rejection (n = 3).

Regarding long-term outcomes, 1-year and 5-year overall survival and graft survival rates were respectively of 93%, 72% and 92%, 71%. The occurrence of ABS was not associated with OS (p = 0.912) and graft survival (p = 0.521) (Figure 3).