Hepatic resection for post-cholecystectomy bile duct injuries: a literature review

Abstract

Objectives:

This study seeks to identify factors for hepatectomy in the management of post-cholecystectomy bile duct injury (BDI) and outcome via a systematic review of the literature.

Methods:

Relevant literature was found by searching the PubMed database and the bibliographies of extracted articles. To avoid bias selection, factors for hepatectomy were analysed in series reporting both patients undergoing hepatectomy and patients undergoing biliary repair without hepatectomy (bimodal treatment). Relevant variables were the presence or absence of additional hepatic artery and/or portal vein injury, the level of BDI, and a previous biliary repair.

Results:

Among 460 potentially relevant publications, only 31 met the eligibility criteria. A total of 99 hepatectomies were reported among 1756 (5.6%) patients referred for post-cholecystectomy BDI. In eight series reporting bimodal treatment, including 232 patients, logistic regression multivariate analysis showed that hepatic arterial and Strasberg E4 and E5 injuries were independent factors associated with hepatectomy. Patients with combined arterial and Strasberg E4 or E5 injury were 43.3 times more likely to undergo hepatectomy (95% confidence interval 8.0–234.2) than patients without complex injury. Despite high postoperative morbidity, mortality rates were comparable with those of hepaticojejunostomy, except in urgent hepatectomies (within 2 weeks; four of nine patients died). Longterm outcome was satisfactory in 12 of 18 patients in the largest series.

Conclusions:

Hepatectomies were performed mainly in patients showing complex concurrent Strasberg E4 or E5 and hepatic arterial injury and provided satisfactory longterm outcomes despite high postoperative morbidity.

Introduction

Post-cholecystectomy bile duct injury (BDI) remains a major health concern as its incidence has increased two-fold since the advent of laparoscopic cholecystectomy compared with that in open procedures.^1,2 Despite the evident learning curve, injuries that occurred in laparoscopic surgery were shown to be more complex than those occurring during the open procedure because of the more proximal location of injury in the biliary tree and its frequent association with vascular injury.³ Although minor BDI, such as leakage from the cystic duct or common bile duct, can often be managed endoscopically, surgical reconstruction is needed for major BDI.^4,5 In most patients, repair of BDI consists of immediate or delayed Roux-en-Y hepaticojejunostomy.⁶ In rare situations, management of these post-cholecystectomy BDIs requires hepatectomy.⁷ The aim of the present study was to specifically analyse, via a literature review, the factors and circumstances that led to the decision for hepatectomy in the management of post-cholecystectomy BDI. The results of such hepatectomies were further analysed in terms of postoperative morbidity and mortality and longterm outcome.

Materials and methods

A literature search was conducted for the period 1990 to July 2008 using the PubMed database, employing the search terms ‘biliary leakage and hepatectomy’, ‘bile duct injury and hepatectomy’, ‘biliary stenosis and hepatectomy’ and ‘cholecystectomy and hepatectomy’. The bibliographies of extracted articles were further cross-referenced. Both analyses by studies and by patients were performed to identify factors for hepatectomy. The aim of the analysis per study was to compare the study populations in their entirety and to search for differences in BDI features that might explain the differences between teams in the management of these BDIs. By contrast, in order to avoid selection bias, the analysis per patient focused on studies that reported on same-team treatment outcomes for patients who underwent biliary reconstruction without hepatic resection as well as patients in whom hepatic resection was required with the biliary reconstruction (bimodal treatment). Only studies providing information on the factors of interest, namely, the presence or absence of additional hepatic artery injury, portal vein injury and the level of BDI (Bismuth⁸ and/or Strasberg² classification), were included. Complex BDIs were defined as proximal BDIs with disruption of the hepatic ductal confluence (Bismuth level 4 or 5; Strasberg type E4 or E5) or any BDI level with concurrent arterial injuries. When available, other data, such as the period of inclusion of patients or previous biliary reconstruction, were collected. Hepatectomies were regarded as urgent when performed within the first 2 weeks. Patients who underwent liver transplantation for post-cholecystectomy BDI or hepatectomy for BDI that was not related to cholecystectomy were excluded from this study. Likewise, partial resection of segments IV and V to allow adequate exposure of the bile duct⁹ was not considered as hepatectomy in this study.

Statistical analysis

Continuous variables were expressed as means (or medians when indicated) and compared using the Mann–Whitney test. Categorical variables, expressed as frequencies and percentages, were compared using chi-squared tests or Fisher exact tests, as appropriate. To identify independent factors of hepatectomy, variables identified as significant in univariate analysis were subsequently included in stepwise logistic regression analysis. The variables of interest were then compared among four groups of patients with: combined arterial and Strasberg E4 or E5 injuries; isolated hepatic arterial injury; isolated Strasberg E4 or E5 injury, or none of these injuries. For logistic regression models, odds ratios and 95% confidence intervals (CIs) were reported relative to a chosen reference group. Statistical significance was considered for P < 0.05. All analyses were performed using spss Version 13.0 (SPSS, Inc., Chicago, IL, USA).

Results

The search strategy identified 460 publications reporting various studies of which 31 met the inclusion criteria (Fig. 1). Of these 31 studies, 10 were case reports and 21 were retrospective series. For three teams, patient(s) undergoing hepatectomy were reported on twice;^9–14 those patients were pooled and considered as one study for each team. Among other extracted articles that reported on only biliary repair with no hepatectomy, 31 studies that provided enough information on the factors of interest described above were used as a control group in the analysis per study.

Figure 1.

Flow chart showing details of articles selected from the PubMed database, subsequent exclusions and cross-referenced bibliographies of extracted articles. BDI, bile duct injury

Analysis per study

Study population

Altogether, 99 patients who underwent hepatectomy for post-cholecystectomy BDI (12 in case reports, 87 in the series reporting bimodal treatment) were identified in a total of 1756 patients (5.6%) and are summarized in Table 1. Overall, 91.5% (65/71 patients in 22 studies) of hepatectomies were performed after laparoscopic cholecystectomy. For anatomical reasons, right lobectomy was the most frequently performed hepatectomy (n = 79/99, 79.8%) (Table 1). In the 21 series reporting bimodal treatment, the median (range) incidence of hepatectomy was 6.1% (1–20%) of all patients referred for BDI, with a median number of three (one to 18) hepatectomies per series (Table 1). Previous biliary reconstruction had been performed in 71.2% (47/66) of patients.

Table 1.

Reported cases of hepatectomy performed for the management of bile duct injury

Author	Year	Patients, n/totala (%)	Vascular injury		Complex injuriesb, n	Time of hepatectomy	Definitive treatment
			RHAI	PVI
Branum et al.16	1993	2/50 (4)	NS	NS	2	NS	Right lobectomy
Madariaga et al.34c	1994	2/15	2	1	2	5/21 days	Right lobectomy
Ohtsuka et al.20	1999	10	0	0	1	4 months	Right lobectomy
Nishio et al.15	1999	10	1	1	1	135 days	Right lobectomy
Topal et al.24	1999	1/16	0	0	1	NS	Left lateral sectionectomy
Uenishi et al.23	1999	10	1	0	1	13 months	Right lobectomy
Lichtenstein et al.33c	2000	2/16	0	0	2	4/28 days	Segment V–VI resection
Nishiguchi et al.51	2000	10	0	0	1	13 months	Right lobectomy
Gazzaniga et al.22	2001	1/60 (1.7)	NS	NS	NS	NS	Left lateral sectionectomy
Kayaalp et al.48c	2001	20	2	0	2	4–15 days	Right lobectomy
Wong et al.52	2001	1	1	1	1	5 weeks	Right lobectomy
Buell et al.32	2002	3/49	3	0	3	NS	Right lobectomy
Heinrich et al.53c	2003	10	0	1	1	2 weeks	Right lobectomy
Schmidt et al.11,12	2002–2004	3/54 (5.5)	3	0	3	NS, 4, 8 months	Right lobectomy
Frilling et al.26	2004	5/40	4	0	4	NS	Right lobectomy
Ota et al.17	2004	10	0	0	1	144 days	Right lobectomy
Sekido et al.19	2004	2/10	2	0	2	2.5/5 months	Right lobectomy
Stewart et al.43	2004	4/261 (1.5)	4	0	4	NS	Partial lobectomy (3), right lobectomy (1)
Mercado et al.9	2006	3/285 (1.1)	NS	NS	3	NS	Right lobectomy
De Santibanes et al.27	2006	9/169 (5.3)	NS	NS	9	NS	Right lobectomy (7) or left hepatectomy (2)
Felekouras et al.54c	2007	10	1	1	1	1 day	Right lobectomy
Yan et al.35	2007	1/8	1	0	1	NS	Right lobectomy
Wu et al.55	2007	2/210 (1)	NS	NS	NS	NS	Right lobectomy
Li et al.56	2007	3/60 (5)	3	0	3	NS	Right lobectomy
Ragozzino et al.57c	2007	20	2	2	2	2–3 days	Right lobectomy
Ortega-Deballon et al.31c	2007	1/18	1	0	1	NS	Right lobectomy
Bektas et al.28	2007	12/74 (16.2)	6	3	7	NS	Right lobectomy (10), segmentectomy (2)
Thomson et al.21	2007	8/119 (6.7)	4	1	7	6–95 months	Right lobectomy (6), left lateral sectionectomy (2)
Alves et al.10Laurent et al.14	2003–2008	18/120 (15)	11	4	18	7 days to 21 years	Right (14) or left (3) lobectomy or left trisectionectomy (1)
Nuzzo et al.58	2008	2/77 (2.6)	NS	NS	NS	NS	Right hepatectomy
Truant et al.30	2009	3/45	1	0	3	8, 18, 30 years	Right lobectomy (2) or left hepatectomy (1)
Total		99/1756 (5.6)	53/80d (66.3%)	15/80d (18.8%)	87/94d (92.6%)		Right lobectomy

^aTotal number of patients referred for bile duct injury

^bComplex injuries were defined as proximal biliary injuries with disruption of the hepatic ductal confluence (Bismuth level 4 or 5; Strasberg type E4 or E5) or any biliary level injury with concurrent arterial injuries

^cUrgent hepatectomy

^dTotal including only patients for whom information was specified

RHAI, right hepatic artery injury; PVI, portal vein injury; NS, not specified

The frequency of vascular injury, severity of biliary injury and timing of hepatectomy are shown in Table 1. In nine patients, urgent hepatectomy (within 2 weeks) was needed in cases of parenchymal necrosis caused by combined biliary and vascular injury following laparoscopic cholecystectomy (Table 1). Substantial variations existed in the indications mentioned for postponed hepatectomy, as follows: recurrent biliary sepsis;^12,15 biliary strictures caused by continuous cholangitis;^12,16,17 development of intrahepatic abscesses;^18,19 non-visualization and/or unsuitability of the proximal stump of the injured bile duct(s) for anastomosis;^9,20,21 intrahepatic injuries of an aberrant right hepatic duct;²⁰ anastomotic strictures and intrahepatic lithiasis;^21,22 right hepatic lobe atrophy with^23,24 or without^10,21 recurrent cholangitis; secondary biliary cirrhosis;^11,12 hepatic confluence stented with metallic stents,^14,25 or primary non-diagnosed Klatskin tumour.²⁶ Such indications were related to symptoms in all patients.

Postoperative morbidity, mortality and outcome of hepatectomies

Rates of postoperative morbidity were high, mainly secondary to infectious complications^14,21,27 (Table 2). In one series of 18 patients, despite the use of intermittent portal trial clamping in 14 patients (78%), 12 patients required a blood transfusion for a mean of 6 ± 3.6 units (range 2–12 units).¹⁴ Biliary fistula occurred in 25% (3/12)²⁸ to 39% (7/18)¹⁴ of patients. One study also reported a longer hospital stay than that found among patients who underwent hepatectomy for other reasons.²⁹ Postoperative death following hepatectomy occurred in 11.1% (9/81) of patients for whom data were available. Nevertheless, mortality was significantly lower in those undergoing non-urgent hepatectomy compared with urgent (within 2 weeks) hepatectomy (2/53 vs. 4/9 patients; P = 0.003). With regard to longterm outcome, the largest series¹⁴ reported that 13 of 18 patients who underwent hepatectomy for complex BDI had no symptoms after a median follow-up of 8 years (range 3–12 years) (Table 2).

Table 2.

Postoperative morbidity, mortality and outcome of hepatectomy in the management of post-cholecystectomy bile duct injury

Author	Year	Patients, n	Complex injuriesa, n	Morbidityb, n	Mortalityb, n	Longterm outcome
Madariaga et al.34c	1994	2 (1c)	2	–	0	Good at 14–16 months
Uenishi et al.23	1999	1	1	–	0	Good at 22 months
Lichtenstein et al.33c	2000	2 (1c)	2	–	0	Good at 12–13 months
Kayaalp et al.48c	2001	2 (2c)	2	–	1	Good at 3 years (one postoperative death)
Buell et al.32	2002	3	3	NS	2	NS
Schmidt et al.11,12	2002–2004	3	3	NS	1	Good at 39 months/SBC at 83 months (one postoperative death)
Heinrich et al.53c	2003	1	1	–	0	–
Ota et al.17	2004	1	1	–	0	Good at 5 years
Sekido et al.19	2004	2	2	–	0	No symptoms at 16–36 months
Frilling et al.26	2004	5	4	NS	1	NS
De Santibanes et al.27	2006	9	9	3	0	NS
Yan et al.35	2007	1	1	–	0	Normal postoperative liver function
Li et al.56	2007	3	3	NS	1	No symptoms at >12 months (one postoperative death)
Felekouras et al.54c	2007	1	1	–	1	–
Ragozzino et al.57c	2007	2 (2c)	2	–	1	–
Thomson et al.21	2007	8	7	5	0	Good in 8/8 (time of follow-up not specified)
Ortega-Deballon et al.31c	2007	1	1	–	1	–
Alves et al.10Laurent et al.14	2008	18	18	11	0	No symptoms in 13/18 patients; median (range) follow-up: 8 (3–12) years
Truant et al.30	2009	3	3	2	0	Good at 13, 17, 63 months

^aComplex injuries were defined as proximal biliary injuries with disruption of the hepatic ductal confluence (Bismuth level 4 or 5; Strasberg type E4 or E5) or any biliary level injury with concurrent arterial injuries

^bNumber of complication(s) and death(s) are given for studies reporting at least three cases of hepatectomy

^cUrgent hepatectomy (when more than one hepatectomy were performed, the number of urgent hepatectomies is given in parentheses in the Patients, n column)

NS, not specified; SBC, secondary biliary cirrhosis

Comparison of patient features between series reporting bimodal treatment and control studies

When the study populations were considered in their entirety, the 21 series reporting bimodal treatment were found to be comparable with the 31 control studies in terms of laparoscopic approach (75.4% [range 23–100%] vs. 76.3% [range 0–100%], respectively; P = 0.60) and periods of inclusion (median time of inclusion before the year 2000 in 90.5% vs. 90.6%; P = 1.00). By contrast, when the 21 series reporting bimodal treatment were compared with the 31 control studies, the incidences of hepatic arterial injury (22.3% [range 0–100%] vs. 13.0% [range 0–61.1%], respectively; P < 0.001], portal vein injury (1.0% [range 0–6.7%] vs. 0.1% [range 0–3.4%], respectively; P = 0.047), complex BDI (26.5% [range 17.8–100%] vs. 21.7% [range 0–100%], respectively; P = 0.01) and previous biliary reconstruction (46.0% [range 0–100%] vs. 41.4% [range 0–100%], respectively; P = 0.042) were significantly higher.

Analysis per patient

Among the 21 studies reporting bimodal treatment, eight^{19,28,30–35} provided enough detail per patient on the variable of interest, which included: injury of the hepatic artery and/or portal vein; proximal pattern of BDI (Strasberg E4 or E5 injury), and, to a lesser extent, previous biliary reconstruction. Other potential factors, such as referral pattern, and presence or absence of sepsis or lobar atrophy, could not be assessed, as insufficient data were provided. Data were collected from these eight studies for a total of 233 patients and results are shown in Table 3. One patient was excluded from analysis as he underwent liver transplantation. Logistic regression models were performed for 182 patients for whom data on all the variables of interest were available; results are shown in Table 4.

Table 3.

Univariate analysis: comparison of patients who underwent hepatectomy with patients who underwent biliary repair in series reporting bimodal treatment

	n (%)	No hepatectomy (n = 206), n	Hepatectomy (n = 26), n	P-value
Artery injury	45 (19.3)	29	16	<0.001
Portal vein injury	5 (2.1)	1	4	0.001
Strasberg E4 or E5 injury	25/182a (13.7)	15/158a	10/24a	<0.001
Previous biliary reconstruction	65/158a (41.1)	56/144a	9/14a	0.06

^aMissing data for remaining patients

Table 4.

Multivariate analysis of 182 patients (hepatectomy: n = 23; biliary repair: n = 159) for whom data on all the variables of interest (hepatic artery injury, portal vein injury and Strasberg E4 or E5 injury) were available

Factors		P-value	Hepatectomy
			OR	95% CI for OR
Hepatic artery injury	Yes vs. no	<0.001	7.7	2.6–22.4
Portal vein injury	Yes vs. no	0.1	6.9	0.6–75.7
Strasberg E4 or E5 injury	Yes vs. no	0.003	5.7	1.8–17.9
Complex injury
No hepatic artery, no Strasberg E4 or E5 injury			1.0	–
Isolated artery injury		<0.001	14.9	4.3–51.6
Isolated Strasberg E4 or E5 injury		<0.001	11.8	2.9–47.2
Artery + Strasberg E4 or E5 injury		<0.001	43.3	8.0–234.2

A separate logistic regression model was used to evaluate the independent effect of complex injury, (hepatic artery injury or Strasberg E4 or E5 injury, or both), in comparison with the group with neither hepatic artery nor Strasberg E4 or E5 injury

OR, odds ratio; 95% CI, 95% confidence interval

Discussion

Overall, 99 hepatectomies have been reported thus far in the management of post-cholecystectomy BDI. This review is the first to show that proximal biliary injuries with disruption of hepatic ductal confluence (Strasberg types E4 or E5) and/or concurrent arterial injuries are independent factors for hepatectomy in the management of post-cholecystectomy BDI. Among 31 studies, 92.6% of 99 patients undergoing hepatectomy had an underlying complex BDI. Patients with combined arterial and Strasberg E4 or E5 injuries were at particular risk of hepatectomy, with an odds ratio of 43.3 (95% CI 8.0–234.2) (Table 4), compared with patients with none of these complex injuries.

Overall, 91% (91/99) of hepatectomies performed for the management of post-cholecystectomy BDI were reported after the year 2000 – one decade after the advent of laparoscopic cholecystectomy – and were carried out mostly by teams specializing in hepatobiliary surgery (Table 1). This strongly contrasted with older series reported by specialist hepatobiliary surgery teams,³⁶ but also with recent series^3,37–39 reporting no hepatectomy despite concomitant vasculo-biliary injuries with or without lobar atrophy. This new indication for hepatectomy may reflect the specific complexity of laparoscopic BDI, which is considered difficult to repair with uncertain outcome. Overall, there were significant differences in the rates of concomitant vascular injuries and complex injuries between teams reporting hepatectomies performed in the management of BDI patients and those performing biliary repair exclusively by hepaticojejunostomy. The enhanced complexity of BDI has certainly impacted the decision to perform hepatectomy in this setting.

Indeed, in recent years, certain teams have deliberately chosen to perform hepatectomy in patients with complex laparoscopic BDI for reasons related to the excellent longterm outcome of hepatectomy.¹⁴ In the present review, rates of good or excellent (no symptoms) results of hepatectomy series, which ranged from 72% (13/18) to 100% (8/8) in the two largest series,^14,21 were comparable with those obtained in most series of BDI patients treated by hepaticojejunostomy.^3,36,40,41 By contrast, given the expected difficulties in dissection of such hepatectomies and their high morbidity rates, which are higher than those reported in the literature for biliary repair, it is likely that additional factors were taken into account in the choice of this treatment.

In fact, most indications for performing hepatectomy were encountered in patients displaying concurrent arterial and proximal injuries with disruption of the hilar confluence, with an odds ratio for hepatectomy of 43.3 (95% CI 8.0–234.2) (Table 4), in comparison with patients with neither arterial nor proximal Strasberg E4 or E5 injury. Indeed, over the longterm, such complex concomitant vasculo-biliary injuries were shown to compromise the viability of biliary ducts and/or parenchyma of the hemi-liver because confluence disruption impaired the development, within the hilar plate, of omega-shaped collateral arterial circulation originating from the preserved branch of the hepatic artery.¹⁰ In this setting, the justification for partial hepatectomy was based on the danger of irreversible fibrotic and atrophic parenchyma, with high risk for secondary complications such as persistent cholangitis as a result of vascular or septic lesions. In accordance with the findings of this study, one recent study of 74 BDI patients showed that the type of vascular involvement and the location of the lesion, at or above the bifurcation of the hepatic duct, had a major impact on the extent of surgical intervention for iatrogenic BDI and especially the need for hepatectomy.²⁸

For similar reasons, the need for hepatectomy was associated, in the present study, with a higher rate of previous biliary reconstruction, with a trend towards significance when compared with patients who had biliary repair (P = 0.06) (Table 3). Each failed repair is indeed associated with some loss of bile duct length and requires more substantial dissection in the pedicle, with subsequent damage to the vascularization of the bile duct.^36,42 Stewart et al.⁴³ studied the impact of injury propagation in successive operations (i.e. the increase in the proximal extent of the BDI between initial injury and subsequent successful reconstruction) upon the incidence of hepatic arterial injury. For injuries located at or above the bifurcation of the common hepatic duct, 49% of those without injury propagation had right hepatic artery injury in comparison with 71% of those with injury propagation (P < 0.05). These authors further showed that the need for hepatectomy was more common among patients with arterial injury than among those without such injury, but only in patients in whom biliary repair had been performed by a surgeon who was inexperienced in hepatobiliary repair.⁴³ These data underline the importance of early referral to a tertiary centre in BDI patients, as previously reported.⁴⁴

By contrast, when the biliary confluence was preserved, patients with and without hepatic arterial injuries had comparable longterm outcomes, provided that high-level repair using the Hepp–Couinaud technique⁴⁵ was performed to guarantee non-ischaemic bilioenteric anastomosis.^10,36,38 Over one-third of patients with injuries for which Strasberg E4 or E5 classification was the only criterion of gravity underwent hepatectomy, which appears high despite the fact that the integrity of the hilar confluence was shown to be important in predicting postoperative biliary stricture.^4,9,38 With respect to the Hepp–Couinaud technique, hepatectomy for Strasberg E4 or E5 BDIs should not be indicated in patients without fibrosis or significant vascular compromise except in rare cases when high-quality biliary repair is difficult or even impossible, such as when segmental bile ducts are either constricted as a result of persistent cholangitis or are unsuitable for anastomosis.^9,17,20 As alternative approaches, side-to-side hepaticojejunostomy on separated right and left ducts,⁴⁶ or partial resection of segments IV and V to enable adequate exposure of the left and right ducts,⁴⁷ have been proposed.

Finally, indications for urgent hepatectomy may be questionable except in cases of concomitant portal vein and arterial injuries. When arterial injury is the only criterion of gravity, its hypothetical role as a factor for urgent hepatectomy should be subject to caution.⁴⁸ Indeed, a delayed repair can improve local and general conditions for potentially difficult hepaticojejunostomy, given the rapid development of collateral circulation within the hilar plate¹⁰ and spontaneous improvement in hepatic infarction and abscesses in most patients.⁴⁹ By contrast, a delayed repair may allow for the accurate assessment of the risk for progression of injury over several months,² which may lead, in turn, to the decision to perform hepatectomy rather than a risky hepaticojejunostomy when duct ischaemia persists. One argument supporting this policy in the current study was the high mortality rate (4/9 patients) in patients undergoing urgent hepatectomy (within 2 weeks), which is far greater than the current near-zero mortality of cancer patients undergoing hepatectomy.⁵⁰

Conclusions

In conclusion, indications for hepatectomy in the management of BDI are rare, given the high success rate of biliary repair, but have increased since 2000, possibly reflecting increases in the complexity of BDIs. This analysis of the literature suggests that hepatectomy has been required mainly in cases of concomitant vascular and proximal biliary injury. Indeed, performing hepatectomy could provide better results and longterm outcomes than repeat hepaticojejunostomy in extensive ischaemic and fibrotic biliary lesions. Mortality rates in hepatectomy seem to be comparable with those reported in the literature for hepaticojejunostomy. Except for concomitant portal vein and arterial injury, emergency hepatectomy should be avoided. Further comparative studies are needed to precisely determine the circumstances in which hepatectomy might be justified.

Conflict of interest

None declared.