Biliary Complications After Liver Transplantation: Old Problems and New Challenges

Authors

  • D. Seehofer,

    Corresponding author
    1. Endoscopy Unit, Department of Gastroenterology and Hepatology, Charité Campus Virchow, Berlin, Germany
    • Department of General-, Visceral and Transplantation Surgery, Charité Campus Virchow, Berlin, Germany
    Search for more papers by this author
  • D. Eurich,

    1. Department of General-, Visceral and Transplantation Surgery, Charité Campus Virchow, Berlin, Germany
    2. Endoscopy Unit, Department of Gastroenterology and Hepatology, Charité Campus Virchow, Berlin, Germany
    Search for more papers by this author
  • W. Veltzke-Schlieker,

    1. Department of General-, Visceral and Transplantation Surgery, Charité Campus Virchow, Berlin, Germany
    2. Endoscopy Unit, Department of Gastroenterology and Hepatology, Charité Campus Virchow, Berlin, Germany
    Search for more papers by this author
  • P. Neuhaus

    1. Department of General-, Visceral and Transplantation Surgery, Charité Campus Virchow, Berlin, Germany
    2. Endoscopy Unit, Department of Gastroenterology and Hepatology, Charité Campus Virchow, Berlin, Germany
    Search for more papers by this author

Corresponding author: Daniel Seehofer

daniel.seehofer@charite.de

Abstract

Due to a vulnerable blood supply of the bile ducts, biliary complications are a major source of morbidity after liver transplantation (LT). Manifestation is either seen at the anastomotic region or at multiple locations of the donor biliary system, termed as nonanastomotic biliary strictures. Major risk factors include old donor age, marginal grafts and prolonged ischemia time. Moreover, partial LT or living donor liver transplantation (LDLT) and donation after cardiac death (DCD) bear a markedly higher risk of biliary complications. Especially accumulation of several risk factors is critical and should be avoided. Prophylaxis is still a major issue; however no gold standard is established so far, since many risk factors cannot be influenced directly. The diagnostic workup is mostly started with noninvasive imaging studies namely MRI and MRCP, but direct cholangiography still remains the gold standard. Especially nonanastomotic strictures require a multidisciplinary treatment approach. The primary management of anastomotic strictures is mainly interventional. However, surgical revision is finally indicated in a significant number of cases. Using adequate treatment algorithms, a very high success rate can be achieved in anastomotic complications, but in nonanastomotic strictures a relevant number of graft failures are still inevitable.

Abbreviations
AS

anastomotic stenosis

CIT

cold ischemia time

DCD

donation after cardiac death

HAT

hepatic artery thrombosis

ITBL

ischemic-type biliary lesions

LD-LT

living donor liver transplantation

LT

liver transplantation

MRCP

magnetic resonance cholangiopancreaticography

NAS

nonanastomotic (biliary) strictures

PSC

primary sclerosing cholangitis

TPA

tissue plasminogen activator

WIT

warm ischemia time

Clinical Classification of Biliary Complications

Biliary complications are frequently observed after liver transplantation (LT). Different classification systems for biliary complications are based on the time of occurrence, localization or etiology. Specific complications have a predominant manifestation period(Figure 1), but from the therapeutic aspect the clinical phenotype is the most suitable classification. Thereby anastomotic complications (stenosis or leak) are distinguished from nonanastomotic complications of the donor biliary system (Figure 2).

Figure 1.

Typical manifestation period of different biliary complications.

Figure 2.

Blood supply of the bile ducts after liver transplantation, classification of biliary complications and etiology of nonanastomotic strictures (NAS).

Vascular Supply of the Biliary Tract and Clinical Implications for LT

The vascular supply is the most vulnerable point of the biliary system. Whereas the liver parenchyma is nourished via a dual vascular supply via portal vein and hepatic artery, the bile ducts are supplied only arterially. It is known, that the biliary epithelium is more liable to ischemic injury than hepatocytes [1-3]. Severe hypotension eventually leads to an ‘ischemic cholangiopathy’ [4] with biliary necrosis, cast formation, subsequent scarring and multifocal stenosis. Likewise, severe hypotension in organ donors causes microcirculatory disturbances and an additional ischemic injury. This hinders optimal preservation of the biliary plexus during organ procurement, but optimal preservation is essential for a low biliary morbidity [5].

The common bile duct is supplied via two main arteries running at the right and left border of the bile duct, the “3 o'clock“and “9 o'clock“arteries, which variably arise form the retroportal, retroduodenal or gastroduodenal arteries and communicate with the right or less often with the left hepatic artery [6] (Figure 2). Approximately 60% of the arterial perfusion comes from the gastroduodenal, and only 30–40% downward from the hepatic artery. The nonaxial supply is sparse, contributing to less than 5% of blood supply [6]. Therefore after LT blood supply of the distal donor bile duct is crucial, since upstream arterial perfusion is lacking. The hilar and intrahepatic ducts are nourished by the peribiliary vascular plexus, a network of capillaries arising form the terminal arterial branches. Thereby the hilar region of the bile ducts is supplied mostly via the communicating arcade [7] (Figure 2), although substantial anatomical variability exists [7].

It is generally recommended to avoid denudation of the bile duct to preserve the biliary arterial supply. The proper hepatic artery should be dissected only at its origin. Further preparation of the right and left hepatic arteries increases the risk of injury to biliary arteries, which eventually arise very proximally. The communicating arcade arises at a mean of 2.5 ± 0.8 cm (range 1.0 to 4.0 cm) from the origin of the right hepatic artery and 1.8 ± 0.8 cm (range 0.5 to 4.0 cm) from the origin of the left hepatic artery [7], which has to be respected especially in partial LT.

Diagnosis of Biliary Complications

Early symptoms of biliary complications are often unspecific or missing. Biliary leaks typically occur early and are diagnosed by routine cholangiography or bilious secretion. Increased inflammatory parameters or fever might occur in the case of undrained bilious collections. Anastomotic or nonanastomotic stenoses are often affiliated with jaundice, increased cholestatic enzymes and fever. Also recurrent cholangitis is a common symptom and should entail additional diagnostic measures.

A cholangiography is easily performed if an external biliary drainage is present. Otherwise, the diagnostic workup is mostly started with noninvasive imaging studies, bearing in mind that these sometimes fail to detect relevant stenoses after LT. Especially ultrasound is less sensitive after LT, since severe dilatation of the intrahepatic bile ducts is absent in >60% of patients with anastomotic stenosis (AS) [8]. Even 1 week before ERC diagnosis of AS, 96% of patients revealed a normal ultrasound [9]. However, in a newer series 92% of patients with biliary complications revealed ultrasound abnormalities [10]. Hepatic artery thrombosis (HAT)—or stenosis is excluded by additional Doppler examination. Contrast-enhanced ultrasound has been used recently to investigate the perfusion of the hilar bile ducts, since detection of severely impaired perfusion may facilitate the early diagnosis of biliary complications [11].

Normal ultrasound findings should not preclude further diagnostic measures An ERC is able to detect the cause of biliary obstruction in 95% and the site of bile leaks in 90% of cases [12]. However, a prospective study of MRCP and ERC revealed comparable sensitivities for detection of biliary obstruction [13]. Other studies confirmed a ≥90% sensitivity and specificity and positive and negative predictive values of 90% for MRCP [14, 15]. A normal MRCP therefore might avoid further invasive measures. In the case of ongoing clinical suspicion, cholangiography remains the gold standard. The route of access to the biliary system is among others based on local experience. In our own practice ERC is the method of choice in patients with duct-to-duct anastomosis and PTC is used as a first-line method only in patients with bilioenteric anastomosis.

To exclude other causes of graft dysfunction (e.g. rejection, CMV-hepatitis) a liver biopsy might be useful. Additional investigations like hepatobiliary scintigraphy (HIDA-scan) have been described with controversial results [16] and are rarely used in the routine workup nowadays [17].

Anastomotic Complications

Definition

Anastomotic complications include anastomotic leakage and segmental narrowing around the anastomosis (anastomotic stricture). For practical reasons this category also encompasses the site of T-tube insertion, since time of manifestation and therapy is similar.

Risk factors for anastomotic complications

Risk factors for anastomotic-leaks and -strictures are closely associated. Major risk factors are inadequate surgical technique, arterial complications or local ischemia of the donor bile duct, the type of biliary reconstruction and the type of liver graft (partial vs. whole LT). Moreover, a preceding bile leak is associated with later AS [18-20]. Other risk factors under discussion include the usage of an external or internal drainage, donor factors and different surgical techniques. Verdonk et al. have shown that the incidence of AS significantly increased from 5.3% before 1995 to 16.7% after 1995 [18], possibly related to an increased use of organs with extended donor criteria. Similarly, Sundaram found more biliary strictures in the post-MELD than in the pre-MELD era (6.4% vs. 15.4%) [19]. This is supposed to be multifactorial, since donor age was older and the incidence of hepatic artery thrombosis (HAT) higher in the post-MELD era. Nevertheless, transplantation in the post-MELD era was an independent predictor of stricture development (OR = 2.30). Other risk factors were donor age (OR = 1.01), a prior bile leak (OR = 2.24) and a choledocho-choledochostomy (OR = 2.22) [19]. However, Serran found no differences in the rate of AS in donors > = 60 but a markedly higher rate of nonanastomotic strictures (NAS) in older donors [21]. Foley detected a donor BMI of more than 25 kg/m2 or a donor weight > 82 kg as risk factors for AS [22]. Likewise, a multivariate analysis revealed a >25% macrovacuolar steatosis of the graft as only risk factor for biliary complications [23]. In HCV patients, early HCV recurrence was shown to additionally increase the risk of AS to 16% compared to 6% in patients with late HCV-recurrence [24], possibly based on inflammatory reactions of the hilar region.

Interestingly, donation after cardiac death (DCD, see below) seems not to be a risk faktor for anastomotic complications. In most studies the incidence is not increased [22, 25]. However, anastomotic complications might be obscured by a high number of diffuse NAS [26].

Anastomotic leaks

Bile leaks may occur at the anastomosis, the T-tube insertion, the cystic duct or the cut surface of partial liver grafts. In a recent literature review enclosing more than 11 000 LT, an incidence of 8.2% was calculated. Thereby, the incidence after Living Donor Liver Transplantation (LDLT) was higher than after full-size LT (9.5% vs. 7.8%). The majority of bile leaks are either seen in the first month after LT or after T-tube removal [17]. Late forms until 6 months are occasionally reported [20]. A recent analysis [19] suggested, that the incidence of bile leaks has decreased from 7.5% in the pre-MELD to 4.9% in the MELD era (p = 0.02). However, the use of a T-tube was the strongest risk factor (OR 3.38) for bile leaks and T-tubes were used less often in the MELD era. However, less than 10% of bile leaks occurred within 2 weeks after surgery (mean 102 days), therefore most of the leaks presumably occurred after T-tube removal [19].

T-tube: risk or benefit?

There is still an ongoing debate about the use of a T-tube. The incidence of T-tube-related complications often exceeds the incidence of anastomotic leaks. Pfau et al. found 74% of leaks to be related to the T-tube [12]. In our own experience T-tube-related complications are rarely observed. In a prospective randomized trial no complications were observed after T-tube removal [27]. This was true even with routine removal after 6 weeks—compared to other reports with removal after 3 months or even later [19]. Possible explanations for these divergent results might be differences in the surgical technique as well as in T-tube material. In our own practice the intraluminal part of a 2.5 mm T-tube is bisected to relief folding during removal. For insertion, a special bile duct probe is connected to the T-tube and pushed through a small hole in the recipient bile duct. After final placement, the opening is further narrowed by PDS 5/0 stitches. Two metaanalyzes found an increased risk for bile leaks if a T-tube is used [28, 29]. A newer metaanalysis found no overall differences in biliary morbidity but a lower rate of biliary strictures in patients with T-tubes [30].

We have shown in a randomized trial using a side-to-side anastomosis [27], that complications with a T-tube were significantly less and especially less severe. In this trial, only one anastomotic leak occurred in 99 patients with T-tube. However, four additional early and clinically irrelevant leaks at the T-tube insertion site were detected only by routine cholangiography. All these leaks disappeared after prolonged unclamping of the T-tube. These clinically irrelevant leaks are diagnosed more often if a T-tube is present. Thus, comparison of the pure rate of bile leaks is misleading, but one has to consider their clinical severity. However this is not the case in most (meta-)analyses [29, 30]. Moreover, leaks after T-tube removal lack severity and are easy to treat, since they do not appear in the critical phase early after LT. Moreover, in retrospective studies a selection bias toward more complicated cases in the T-tube group has to be assumed. The subjective nature of the discussion on T-tubes is underlined by the fact, that even centers which identified a T-tube as significant risk factor for bile leaks in an earlier analysis [31], still used it in 58% of recent transplantations [19]. Also the ASTS guidelines [32] suggest the use of a T-tube after DCD donation due to a high risk of biliary complications.

Anastomotic stenosis

The reported incidence of AS is 13% after full size and 19% after LDLT [20]. AS in the early course are often a result of surgical failure, whereas late AS may develop gradually as consequence of local inflammation due to ischemia, bile leaks and other factors. Clinical manifestation takes place within 6 months [9, 17], but occasionally also after many years [17, 20]. Verdonk reported a cumulative incidence of 6.6%, 10.6%, and 12.3% after 1, 5 and 10 years, respectively [18].

Other rare causes of biliary obstruction like obstructing mucoceles of the cystic duct or bile duct concrements are not discussed in detail.

Prophylaxis of anastomotic complications

Many risk factors like donor age or steatosis cannot be influenced. The optimal surgical reconstruction technique including biliary drainage is still under debate. Choledocho-choledochostmy is the standard procedure in full size and LDLT. Some analyses report a similar [33] or even lower [19] rate of biliary complications after bilioenteric anastomosis, others a higher rate [34].

Technical modifications of choledocho-choledochostomy include an end-to-end versus end-to-side or side-to-side reconstruction as well as running versus interrupted suture. Theoretical disadvantages of an end-to-end anastomosis include the fragility and an impaired perfusion at the distal end of the bile duct. Moreover, incongruence in diameter might be difficult to adjust. For small bile ducts the risk of stenosis is considerable, particularly if a running suture is used. This might be avoided by a side-to-side anastomosis, which allows a long anastomosis irrespective of the bile duct diameter. In addition, a good perfusion of the bile duct is warranted, since the 3 and 9 o'clock vessels are completely preserved by incision of the bile duct at 12 and 6 o'clock (Figure 2). However, a randomized trial comparing these technical modifications showed no significant differences [35]. Overall, no differences between running and interrupted sutures have been detected [19, 36], but there is some indication for less AS with interrupted sutures (5.1% vs. 9.8%) [36]. Moreover, it is generally accepted that the best prerequisite for low biliary morbidity is good perfusion with active bleeding at the biliary ends, preservation of periductal tissue and avoidance of vascular injury to the bile duct vessels, e.g. by extensive cautery.

Treatment of bile leaks

If a T-tube is in situ small leaks may be managed by leaving the T-tube open. The therapeutic algorithm is center specific and depends on the personal experience. In the case of significant leaks most centers regard an ERC as treatment of choice after choledoch-choledochostomy. Reported success rates of 80–90% reinforce this strategy [20]. Small leaks may be managed by sphincterotomy alone. Leaks after T-tube removal can be managed by ERC in approximately 100% of cases, whereas anastomotic leaks respond only in about 50% of cases [12, 37]. Rarely a PTC is performed directly, whereas it is used as second line therapy and in patients with bilioenteric anastomosis [20]. Few centers generally prefer nasobiliary catheters for the treatment of bile leaks [38]. In case of abscess- or bilioma-formation an additional CT- or ultrasound-guided drainage is indicated.

In selected cases a primary surgical revision might be indicated, especially in early leaks (<1–2 weeks after LT), large defects or if bile duct necrosis is suspected. Surgical intervention is also indicated after failure of interventional therapy. Local repair is rarely feasible; therefore a bilioenteric anastomosis is often required, especially in the case of periductal infection and/or bile duct necrosis. Overall, bile leaks require surgical revision in up to 30% depending on time of occurrence and severitiy [36, 39]. Using all these options, success rates of 85–100% are reported. Although single cases of graft loss have been reported, there is no significant impact on graft survival [19].

Therapy of anastomotic stenosis

The majority of centers use ERC plus ballon dilatation with [40] or without [9] stent placement for the treatment of AS. In case of very early stenosis (<4 weeks after LT), dilatation can provoke anastomotic leakage and might therefore be postponed. It is supposed, that dilatation plus insertion of one or multiple stents is associated with a higher success rate, than dilatation alone, where a 50% treatment failure has been reported [41]. However, even with dilatation alone, good long-term results have been reported with a lower risk of bacterial cholangitis [9]. The standard procedure in our center as well as in many other centers [17, 42] is repetitive balloon dilatation (every 2–3 months) mostly with parallel stenting (two or three 10 to 12F stents). In case of an adequate treatment response the stents are removed after 3 [18] to 12 months followed by early control ERC.

Although severe complications after ERC are rare, a large prospective series from Birmingham revealed a complication rate of 6.6% per procedure. Since in most patients more than one investigation was necessary, the cumulative complication rate per patient was 21% [10].

Overall 60–90% of AS can be treated interventionally [18, 43, 44]. However, AS with a narrow diameter and AS manifesting later than 6 months after LT required endoscopic interventions more often, for a longer interval and required more often surgical revision [18]. Prolonged biliary obstruction and cholangitis during interventional treatment might be associated with progressing graft fibrosis. Early conversion to surgical therapy is indicated, since delay might be accompanied with persistent allograft dysfunction [45]. In the long term, about 10–20% of patients with AS require surgical revision [10, 18], mainly in the form of bilioenteric anastomosis [46, 47]. In most centers surgical treatment is regarded as second-line therapy and it has been shown, that previous failure of interventional therapy does no negatively affect the outcome after ‘second-line’ surgical reconstruction [48]. This algorithm is also used in our own practice. However, in selected cases with very narrow AS occurring later than 6 months conversion to bilioenteric anastomosis is also used as first-line therapy.

In patients with bilioenteric anastomosis balloon dilatation is performed percutaneously [17, 43]. Additionally, Yamakawa drains [49, 50] with increasing size are often used in the own practice for 6–12 months. Afterward the catheter is retracted intrahepatically and removed after a subsequent control. However, the overall interventional success rate after hepaticojejunostomy is slightly lower than after duct to duct anastomosis [18]. Also after LDLT balloon dilatation plus single or multiple stenting has been reported with good success rates [51]. However, partial LT and DCD remain negative prognostic factors [40]. Although the overall success rate of AS therapy is very high (98%) a close follow-up is indicated, since the risk of recurrence is relevant.

Sphincter Oddi Dysfunction/Papillary Stenosis

Outflow obstruction at the papillary region occurs in 2–7% of patients after LT [52]. It is thought to be associated with denervation of the recipient bile duct—or due to inflammation or scarring with resulting stenosis of the sphincter Oddi. Typical clinical manifestation includes increased cholestatic enzymes and eventually serum bilirubin together with dilatation of the extrahepatic donor and recipient biliary system. Endoscopic sphincterotomy is the treatment of choice and effective in the vast majority of patients [12].

Nonanastomotic Biliary Strictures (NAS)

A common picture of alterations of the donor biliary system caused by different etiologic mechanisms is summarized under the clinical phenotype ‘nonanastomotic (biliary) strictures’ (NAS, Figure 3).

Figure 3.

Classification of the anatomic regions of the biliary tree affected by nonanastomotic biliary strictures (according to Buis et al. [54]): hilar bifurcation (zone A), ducts between the first- and second-order branches (B), between second- and third-order branches (C) and in the periphery of the liver (D). Especially the extent of intrahepatic affection predetermines treatment success, whereby severe involvement of zone C is most critical.

Etiology and clinical manifestation

The most severe forms of NAS evolve in the case of early hepatic artery thrombosis (HAT). The absence of early arterial collateral perfusion results in partial or complete biliary necrosis. Desquamated epithelial cells forms together with bile products typical biliary casts (Figure 4), which lead to the corresponding ERC finding of multiple filling defects (Figures 5 and 6). In the case of late HAT or arterial stenosis attenuated forms are observed owing to collateral perfusion via the intrahepatic-, the biliary- or the capsular route. A PSC like manifestation with multiple segmental stenoses, subsequent dilatations, sludge and concrement formation is observed. Later a rarefication of intrahepatic bile ducts evolves. The severity of NAS correlates with the time of manifestation, being most severe in the first year [53], Figure 5), whereas late HAT may even be clinically inapparent. Biliary complications after HAT are referred to as macroangiopathic form of NAS; in contrast to microangiopathic forms, which are characterized by injury to the peribiliary vascular plexus as a result of ischemia-reperfusion or immunological injury. Other known causes of NAS include ABO-incompatiblity, chronic rejection and recurrent PSC. The common macroscopic picture is often termed ischemic-type biliary lesions (ITBL).

Figure 4.

Typical biliary cast from a patient with early NAS removed during operative revision with creation of a bilioenteric anastomosis due to bile duct necrosis.

Figure 5.

Progressive form of NAS with diffuse involvement of intrahepatic biliary tree (according to zones A to D after Buis et al. [54]) in a patient with hepatic artery thrombosis with almost complete disappearance of the functional bile ducts despite intense interventional treatment by the endoscopic and percutaneous route and requirement of liver retransplantation.

Figure 6.

Successful long-term therapy of NAS in a patient with (A) dominant involvement of the extrahepatic and proximal intrahepatic involvement of the bile ducts according to zones A and B after Buis et al. (54). (B) Placement of one stent and two PTCD. (C) Insertion of Yamakawa drains with increasing diameter. (D) Repeated ballon dilatations for consolidation of the treatment succes after removal of the Yamakawa drains (E) Control ERC two years after primary manifestation.

In a detailed retrospective analysis by Buis, several risk factors for NAS were identified. These include the use of UW-solution (vs. low-viscosity HTK solution), Roux-en-Y reconstruction, postoperative CMV infection and PSC as underlying liver disease [54]. In contrast to several other analyses, ischemia time was not a significant risk factor [55]. Possibly, not the pure ischemic time is crucial for the development of NAS, but other risk factors like donor age [21, 56] and organ quality [19] may potentiate the effect of prolonged ischemia. Other suspected risk factors include repeated rejection episodes, positive lymphocyte cross-match, poor HLA-match and a high peak of liver enzymes as indicator of preservation injury and/or organ quality [57]. Especially severe macrovascular steatosis (>25%) has been described as risk factor [58] since the hepatic microcirculation is generally impaired after transplantation of steatotic livers [59]. Moreover, several immunological factors may increase the risk for NAS. For example chronic rejection has been identified to cause NAS, presumably not via direct injury to the biliary epethelium, but via injury to the biliary arteries. Also a genetic predetermination, e.g. the CCR5-Δ32 mutation in recipients is supposed to be associated with an increased risk for NAS [60, 61].

Classification of NAS

Owing to prognostic and therapeutic implications, extrahepatic lesions (type I) are distinguished from intrahepatic (type II) or a combination of intra- and extrahepatic abnormalities [62, 63]. Since mainly the extent of intrahepatic alterations is relevant for the treatment success (Figure 5), Buis et al. have suggested a classification of the involved intrahepatic zones A to D [54], Figure 3). Involvement of zone C which represents the second-order ducts, seems to predict a more severe clinical course, because of therapeutical difficulties. Zone D, representing the small peripheral ducts, is mainly involved in late forms of NAS with immunological pathogenesis.

Incidence of NAS

In different studies the incidence of NAS ranges between 5% and 25%. However, this largely depends on the era of the analysis, inclusion criteria, patient selection, follow-up and other factors. Recently, the incidence of NAS increased in many centers due to a more liberal acceptance of older donors, donors with extended criteria (ECD) and donors after cardiac death (DCD) [19].

The risk of NAS is considerably high using DCD, which is associated with NAS in more than 40% of transplantations [64, 65]. Comparably high rates are reported in ABO-incompatible LT, since ABO-antigens are expressed on the biliary epithelial cells. Also the incidence in PSC patients is increased, presumably because of a summation of genetic components, PSC recurrence and ITBL [66]. The incidence of ITBL in non-PSC patients varies between 1.4% and 26%. After exclusion of PSC patients and patients with HAT in our own experience, an incidence of 3.9% was observed [55]. The variable incidences of ITBL in different publications may be partially due to different patient characteristics. However an incidence above 10–15% “true” ITBL in many series suggest, that there might be potential for improvement by adequate prophylactic measures.

Prophylaxis of NAS

The basis for a sufficient biliary microcirculation is established during organ procurement. The viscosity of the UW solution hinders adequate perfusion of the small biliary arteries [57], and therefore doubles the risk for NAS [53, 55, 67]. Moreover, arterial pressure perfusion significantly reduces NAS [55, 68] but the effective pressure in the hepatic artery is difficult to control. Thus, additional arterial back-table pressure perfusion has been suggested [56].

It has been shown that a cold ischemia time (CIT) of less than 11.5 h is associated with a 2% incidence of ITBL, whereas it increases to >50% after 13 h CIT [69]. This trend has been confirmed repeatedly [55]. Therefore, CIT should be kept ≤10 h, if other risk factors for NAS are present. Bile salt residues are a further factor, which may potentiate the ischemia-reperfusion injury [70, 71]. Thorough retrograde flushing of the bile ducts is therefore recommendable. Also recipient characteristics like a Child-Pugh status C have been reported to significantly increase the incidence of NAS [55, 72].

Importantly, the duration of the warm ischemia time (WIT) influences the NAS risk. Portal reperfusion followed by later arterial reconstruction is commonly used. This increases the WIT of the arterially perfused bile ducts. Simultaneous arterial and portal reperfusion reduced the biliary ischemic time and subsequently also NAS [58, 73]. However, others failed to prove this correlation [74]. Accordingly, retrograde reperfusion via the vena cava has been shown to increase the risk for ITBL, since arterial revascularization is delayed [75]. The overall risk of ITBL might be balanced by matching of donor and recipient characteristics and avoidance of accumulation of risk factors. In the case of very high risk, e.g. after DCD-LT, some authors recommend additional secondary prophylaxis of NAS by insertion of a T-tube for earlier diagnosis and intervention [32].

Clinical course of NAS

Clinical presentation of NAS includes a cholestatic picture often with episodes of cholangitis. Up to 50% manifest within the first year [54]. In two-third of cases progressive forms are observed [53], eventually leading to fibrosis and cirrhosis. Thus, 10-year graft failure rates of 20–50% are reported [53, 66]. The most unfavorable prognosis is seen in patients with manifestation <1 year after LT and/or recurrent episodes of cholangitis [53].

Therapy of NAS

Early diagnosis and therapy is important. Treatment of NAS is not standardized. A multidisciplinary, individualized treatment approach is required. Due to diffuse involvement of the graft's biliary system therapy is complex and success rates are lower than in anastomotic complications. Especially in severe forms with cast formation endoscopic therapy often fails [12], Figure 5). Several treatment options are available (Table 1). Although not evidence-based, ursodesoxycholic acid is commonly used to increase bile flow, and lower the lithogenicity. Additionally, antibiotic therapy and prophylaxis are often necessary. Recurrent episodes of cholangitis are observed in approximately every forth patient [53], requiring even antibiotic maintenance therapy in a relevant number.

Table 1. Treatment options for nonanastomotic biliary strictures after LT
Interventional• Dilatation ± stent placement (usually ≥ 1year)
 • Extraction of necrotic tissue and biliary casts
Surgical (rarely indicated)• In selected cases (dominant hilar stenosis) resection of the bile duct bifurcation and hepatico– jejunostomy
 • Additional liver resection
Medical• Ursodeoxycholic acid
 • Antibiotic treatment for (recurrent) cholangitis
Liver retransplantationUp to 50% of patients suffer from graft failure

In severe cases with biliary necrosis and casts, repeated interventions with baskets and dilatations are necessary [40]. Placement of stents is not generally recommended in the early course, since they tend to be occluded by biliary debris. Treatment success in patients with biliary casts is lower than in patients without [12]. In the case of a dominant hilar stenosis, balloon dilatation might be consolidated by insertion of one or multiple stents. However, also insertion of transcutaneous drains (PTCD or Yamakawa-type prosthesis [49]) with increasing diameter is routinely used even in patients without bilioenteric anastomosis (Figure 6). In patients with dominant hilar involvement conversion to hepaticojejunostomy might be useful very rarely after failure of nonoperative treatment [76]. Using intense endoscopic and percutaneous therapy a sufficient graft function might be maintained in about half of the patients or at least the time until retransplantation might be delayed.

Special Issues of DCD

As a measure to increase the donor pool, DCD has been increasingly performed in many countries. During the additional period of WIT without organ preservation stasis and thrombus formation in small vessels is expedited [77]. Since also the peribiliary plexus is involved, the incidence of biliary complications is almost doubled [78]. An ‘ischemic cholangiopathy’ evolves in >20% of recipients [79]. The clinical picture is similar to NAS of other etiology, but it manifests regularly within 3 months after LT [78]. Its clinical course is often more severe and retransplantation required in a significant percentage [79, 80]. Analysis of the UNOS database of 2351 DCD-LT revealed a significantly decreased patient and graft survival with a 5-year graft survival rate of 56% [81]. In this analysis CIT >8 h, donor age >50 years and donor WIT ≥ 20 min were identified as most relevant graft related factors for graft failure, but biliary complications were not specifically analyzed. However, single center experiences report overall results after DCD-LT, which are not significantly different from LT with donation after brain death [79]. In a specialized experience with careful donor selection the incidence of ischemic cholangiopathy was only 12.6%. However, 70% of the affected patients either died or underwent retransplantation [82].

The time from asystole to cross-clamp has been identified as major risk factor for the development of ischemic cholangiopathy [82]. Thereby the authors have calculated that each minute of additional WIT increases the odds ratio for the development of ischemic cholangiopathy or hepatic necrosis by 16% [79]. Also other risk factors, known from donation after brain death, have been confirmed in a US cohort of 1567 DCD-LT including donor age, donor weight and CIT [80]. In view of the high biliary complication risk routine surveillance might be important in the early period after LT. This is relieved by insertion of a biliary drainage either via the transcystic route [79] or as T-tube [32] to facilitate early therapeutic intervention. Other, potentially preventive strategies recommended by the ASTS include limitation of the total WIT to less than 30–45 min, limitation of the cold ischmia time to less than 8–10 h and especially staying at the lower limit of both in older donors and fatty livers. Moreover simultaneous arterial and portal revascularization should be considered [32].

Innovative approaches for lowering the biliary mobidity of DCD-LT include thrombolytic agents [83] or machine perfusion [84]. First clinical results using backtable injection of tissue plasminogen activator (TPA) into the donor hepatic artery before DCD-LT revealed a low incidence of ischemic cholangiopathy of 9% and an overall biliary complication rate of 27%. However, excessive bleeding evolved in two-third of the recipients, mainly if other risk factors like poor graft quality or severe adhesions were present [83] In conclusion, DCD-LT is a suitable method to expand the donor pool, a careful donor selection (avoid accumulation of risk factors) and an optimum management of organ procurement (CIT, WIT, consider TPA) and transplantation (simultaneous revascularization, early surveillance for biliary complications) provided.

Special Issues in Split and Pediatric Liver Transplantation

Several additional problems contribute to an increased biliary morbidity after split LT [85]. These include cut-surface leaks due to unclosed or aberrant bile ducts and necrosis of liver tissue, mainly of segment 4 in the case of asymmetric splitting. Biliary complications are reported in 29% of adult and 40% of pediatric recipients of partial liver grafts [86]. Nevertheless, excellent long-term results are reported in the case of extended right/left lateral splitting, with 10% surgical revisions in adult and 28% in pediatric recipients. No significant differences between in situ and ex vivo splitting were observed [86]. The risk of bile leaks from the cut-surface is higher than after LDLT; e.g. in the Hanover experience 15% cut surface leaks were observed [87]. Most cut surface leaks have a favourable managed either by percutaneous drainage or surgical revision [87]. Although available evidence is sparse, some authors recommend routine use of fibrin glue or fibrin–collagen sponges for prophylaxis of cut-surface leaks [88]. Due to devascularization of the hilar plate in split grafts, bile duct necrosis is observed in up to 5% of cases, which requires surgical revision or even retransplantation [87].

However, the overall survival after split LT, e.g. of extended right lobes is not significantly different from whole LT; therefore the slightly increased biliary complication rate has to be disregarded, especially since split LT is unavoidable in pediatric LT due to the lack of an adequate number of full size organs. The biliary morbidity in pediatric recipients of partial liver grafts is also significantly higher than in recipients of whole liver grafts. The cumulative 2-year biliary complication rate was 17% in pediatric whole LT compared to 29% in pediatric split LT and 40% in pediatric LDLT [89]. Since about 75% of pediatric LT are performed with a bilioenteric anastomosis [89], most biliary complications are managed by percutaneous intervention. Additionally an overall surgical revision rate of 30% is reported [90]. Since the biliary morbidity in pediatric LT consists mainly of anastomotic complications, the treatment success is good and retransplantation is rarely required [89].

Special Issues in LDLT

Biliary anastomotic complications are more frequently observed after LDLT than after full-size LT (except DCD). Therefore several specific issues concerning the prophylaxis of biliary complications after LDLT are to be discussed.

Prophylaxis of biliary complications in LDLT

Bile leaks from the resection surface are occasionally observed after LDLT. Especially biliary branches from the caudate lobe are at considerable risk. These branches, 3–5 in number, mostly drain into the left hepatic duct, but sometimes into the right sectoral ducts or the right hepatic duct [91]. Due to these anatomical variations careful dissection is mandatory. Moreover some authors recommend continuous suturing of the portal plate or U-stitches in the caudate region for the prevention of bile leaks in the donor [92] as well as in the recipient [91].

Bile ducts are generally small and thin walled in LDLT. Moreover, the blood supply of the donor as well as the recipient bile duct is critical. Many technical refinements of the donor [93-95] as well as the recipient operation [96-98] are aiming at an optimal blood supply of the anastomotic region. In general, during the donor operation the arterial branches to the right respectively left hepatic duct are preserved by sparse dissection of the hepatic artery and especially by avoidance of dissection between the hilar plate and the artery to minimize injury to the communicating artery and the peribiliary plexus [5]. Accordingly, during recipient hepatectomy only minimal dissection of the hilar region is recommended and methods like the ‘high hilar dissection’ technique are used to optimize the blood supply of the recipient duct in case of duct-to-duct anastomosis.

Flushing and perfusion of the hepatic artery are controversial in LDLT. It is not used in many centers because of potential injury to the intima and alternative techniques, e.g. with retrograde flushing of the arterial system have been postulated [99], but their clinical benefit remains to be confirmed.

It is well known, that size and number of donor bile ducts are the major determinants of biliary complications after LDLT [100, 101]. One analysis revealed even a sixfold increased risk of biliary complications in the case of more than one biliary orifice [102]. However, since the donor biliary anatomy cannot be influenced, the increased risk can only be weight against the benefit of LDLT over potential alternatives.

In parallel to full-size LT, a preceding bile leak significantly increases the risk for a subsequent biliary stenosis [100]. Also cold ischemic time seems to be an important predictor. Park et al. showed that only 10% of patients developed biliary complications if the cold ischemic time was ≤ 71 min compared to 35% in the case of more than 71 min cold ischemic time [100].

Anastomotic complications after LDLT

In most centers a duct to duct anastomosis has evolved as the standard technique after LDLT, as the incidence of AS is comparable or only slightly higher than after Roux-Y reconstruction [101, 103, 104]. However, after duct-to-duct anastomosis complications are easier to manage using ERC. Only in the case of multiple or very small ducts, a Roux-Y reconstruction might be preferable [101]. Again, the placement of transanastomotic drains is discussed controversially. Whereas some authors report a lower rate of biliary complications using transanastomotic stents [103] others avoid stents due to the risk of infectious complications [105, 106]. However, for reconstruction of small bile ducts (<2 mm) it is used by most surgeons [106].

Nevertheless, the overall biliary morbidity in most series is still between 10% and 30% [100, 107-109]. A considerably lower rate has been reported in newer analyses; however long-term data remain to be awaited, since many stenoses manifest several years after LDLT.

Therapy of anastomotic complications is in general more difficult after LDLT, since often more than one orifice is involved [110]. Nevertheless, the majority can be managed nonsurgically, and endoscopy remains the first treatment option after duct-to-duct anastomosis [111-113]; PTC in the case of Roux-Y reconstruction or as reserve option after failure of endoscopic therapy [113]. The final success rate of endoscopic therapy is 60–75% [109] and thereby lower than after full-size LT.

Nonanastomotic complications after LDLT

In contrast, the rate of NAS is low after LDLT and those are mainly based on PSC recurrence [109], since major risk factors like long ischemic time, graft steatosis, old donor age, hemodynamic instability of the donor and especially the accumulation of several risk factors is absent in LDLT. In the case of failure of interventional therapy, surgical optinons are limited and technically demanding after LDLT. Conversion from duct-to-duct to Roux-Y might be useful in selected cases [100], in the case of severe graft damage retransplantation might be considered.

Disclosure

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

Ancillary