Transition from a low- to a high-volume centre for bile duct repair: changes in technique and improved outcome

Abstract

Background

Improvements in bile duct injury repairs have been shown in centres with specialized surgeons. The aim of the present study was to demonstrate the temporal change in the pattern of referral, technical variation associated with repair and long-term outcome of bile duct injuries at a tertiary referral centre in Mexico City.

Methods

A retrospective case note review was performed. Patients were divided into two groups: group I (GI) 1990 to 2004 and group II (GII) 2005–2008, and appropriate statistical analysis undertaken.

Results

Over a 20-year period, 312 patients with iatrogenic bile duct injuries required surgical treatment (GI = 169, GII = 140 patients). All injuries were reconstructed using a Roux-en-Y hepaticojejunostomy. The proportion of patients who had undergone a laparoscopic cholecystectomy increased from 24% to 36% (P = 0.017) over the two time periods. In the second time period there was an increase in segment IV and V partial resections (P = 0.020), a reduction in the use of transanastomotic stents (42% to 2%, P = 0.001) and an increase in the proportion of patients requiring a neoconfluence (2% to 11%, P = 0.003). In the second time period, the number of patients requiring a hepatectomy during repair (2% to 1%, P = 0.001), a portoenterostomy (16% to 9%, P = 0.060) or a double-barrel hepatico-jejunostomy (5% to 1%, P = 0.045) significantly decreased. During follow-up, patients in the second time period had a reduction in the incidence of post-operative cholangitis (11% to 6%, P = 0.310) and the frequency of post-operative anastomotic stenoses (13% to 5%, P = 0.010). Mortality remained low throughout the series but was absent in the second group.

Conclusions

Changes in technique and growing experience of the multidisciplinary team improved operative and long-term results of bile duct injury repair.

Background

The incidence of bile duct injury (BDI) associated with a laparoscopic cholecystectomy has remained constant at 0.3% to 0.6%.^1,2 Several studies have shown better long-term outcomes for patients with bile duct injury if subsequent surgery is performed at centres experienced in the repair of such injuries.^3–6 Knowledge of the mechanism of injury, correct classification and decision making with respect to the timing and method of repair (endoscopic, radiological and/or surgical) are all important in optimizing outcome. This is more likely to occur in a centre specialized in hepato-pancreato-biliary (HPB) surgery^6–8 and the best post-operative results are reached with an individualized therapeutic approach.^6,8–12

Previous studies have reported complete rehabilitation after injury in 75–98% of patients.^{3–6,8–10,13–15} Such a wide range in rehabilitation may be explained by the difference in the mechanism of injury and variability in the final anatomical status of the ducts. The individual characteristics of a given patient including the loss of biliary confluence and the state of the vascular blood supply of the ducts are important factors in determining long-term outcome.^16,17 A Roux-en-Y hepaticojejunostomy remains the best surgical alternative in patients where complete transection of the duct has occurred.^14,18

The aim of the present study was to analyse the temporal change in referral pattern, surgical technique and long-term outcomes in patients who suffered a bile duct injury.

Methods

With prior approval of the local Institutional Committee for Human Investigation, a retrospective review of the hospital database from January 1990 to December 2008 was conducted.

All patients referred with a BDI were evaluated by a multidisciplinary group and the best available treatment option decided. For the purposes of the present study, only patients who underwent surgical reconstruction were included for further analysis. Patients were divided in two groups according to the year of repair: GI 1990–2004 (172 patients) and GII 2005–2008 (140 patients). All were treated by means of a Roux-en-Y hepaticojejunostomy. Modifications^8,14,19 that occurred over time were recorded. This division of time periods was chosen as major changes (reduction of stent use, liver segment IV resections and the use of a neoconfluence as part of the reconstruction) had all occurred. All repairs were been performed by one surgeon (M.A.M.).

Variables assessed were divided according to pre-operative, operative and post-operative periods. A minimum follow-up of 6 months was obtained in the second group to ensure adequate assessment of long-term results.

The index operation was defined as the first reconstructive procedure after referral to the HPB surgeon. A hepatectomy was recorded either as part of the index operation or as a reintervention. A partial segment IV and V resection is defined as a small wedge resection performed just above the bifurcation of the bile duct at the hilar plate. A neoconfluence is defined as the surgical approximation of separated right and left hepatic ducts in order to perform a single hepatico-jejunostomy (partial resection of liver segment IV is always performed in combination to allow a tension free approximation of the separated ducts). Operative mortality was considered within 60 days of the index operation or during the same hospitalization as the index operation.

Classification of the injury

The Bismuth classification was used before 1997²⁰ and the Strasberg classification has been applied since.²¹ Other classifications systems have been proposed but are not currently used.^22,23 Since other imaging techniques were not readily available, ultrasound was used as the main imaging method in the 1990s to document intrahepatic dilation of the bile ducts and bilomas while providing very limited information on the type of injury. Percutaneous cholangiography was used and is still used selectively as a result of its invasive nature. A fistulography before the index operation through post-operative drains was a useful diagnostic tool in some patients. Magnetic resonance cholangio-pancreatography (MRCP) and endoscopic retrograde cholangio-pancreatography (ERCP) have been used in the last decade as the most effective tools to visualize of the biliary tree pre-operatively.

In recent years, patients were scheduled for endoscopic or radiological treatment, mainly when there is continuity of the bile ducts and in some situations where stenotic bilio-enteric anastomoses can be dilated percutaneously (either before or after the index operation).

In terms of timing, patients underwent reconstruction as soon as adequate anatomical detail had been obtained. Patients who presented with severe sepsis and multiple organ failure were excluded from such an approach. In this situation, repair was delayed until the patient's condition had stabilized. Thus the time of repair in this series was dictated mostly by the timing of referral.

Surgical technique

The standard approach consisted of selectively dissecting the porta hepatis, preserving all the identified arterial branches and freeing the anterior aspect of the proximal bile ducts. The repair was undertaken to the anterior surface of the proximal bile ducts ducts to preserve the circulation and to permit a more proximal anastomosis. In some patients lowering the hilar plate was perfomed to access more proximal ducts for the anastomosis. To obtain a wide, tension-free anastomosis and to give room for the intestinal loop, partial resection of segments IV and V was performed frequently.²⁴ The extension of the anterior opening of the common bile duct towards the left duct was common practice, particularly for thin ducts. The jejunal limb is anastomozed in a side-to-side fashion, with interrupted everted sutures of 5–0 hydrolyzable monofilament material (Figs 1,2).¹⁶ A major hepatectomy was performed at an index operation in patients in whom a duct was identified as having irreversible damage and/or the intrahepatic biliary tree was affected because of major arterial injury, mainly the right hepatic artery.^14,25 A closed suction drain was used routinely.

Figure 1.

Left sagital view of the latero-lateral hepatojejunoanastomosis at the level of the confluence. Anterior suture line with everted single layer 5–0 absorbable sutures is visible (black arrow head). Black arrow shows liver segment IV wedge resection. Long arrows show segmental left hepatic arteries. White arrow shows the portal vein

Figure 2.

Anterior suture line of the hepaticojejunal anastomosis

After discharge, routine follow-up was maintained via outpatient clinic appointments at 1, 3, 6, and 12 months and yearly thereafter. A clinical history and biochemical analysis²⁶ for cholangitis were undertaken at each visit. If there was any suspicion of cholangitis, further imaging of biliary tree by percutaneous cholangiography (PTC) or magnetic resonance imaging (MRI) were undertaken. Depending on the results, management ranged from ambulatory antibiotic use to hospitalization and, in rare situations, reoperation.

Statistical analysis

Continuous variables are presented as medians (range). GraphPad software was used, calculating Fisher's exact test to compare all dichotomic variables.

For analysis, groups I was compared with group II and the differences were considered significant when the P-value was <0.05.

Results

Pre-operative data are shown in Table 1. Of the 158 patients who had undergone previous repairs, 98 (62%), 22 (14%), 14 (9%), 5 (3%) had an hepatico-jejunal, choledococholedocal, choledoco-duodenal anastomosis or a primary repair, respectively. In 19 (12%) patients, the details of the previous surgical procedure were not available.

Table 1.

Demographics and pre-operative variables stratified in three time periods

	Group I 1990 to 2004		Group II 2005 to 2008		P (0.001)
	No.	%	No.	%
Total	172	100	140	100	–
Male	45	26	31	22	–
Female	127	74	109	78	–
Age (range)	16–81		18–85		–
Cholecystectomy
Open	131	76	89	63	0.017
Laparoscopic/converted	41	24	51	36
Previous repair
0	86	50	68	49
1	71	41	60	42	NS
2	10	6	10	7
3	4	2	1	1
>4	1	1	1	1
Type of lesion21
E1–E3	129	75	109	78	0.594
E4–E5	43	25	31	22
Diagnosis
ERCP	89	52	53	38	–
US	34	20	37	26	–
PTC	30	17	15	11	–
MRCP	37	22	32	29	–
CT	24	14	26	18	–
Fistulography	9	5	8	6	–
No pre-operatory image	18	10	17	12	–
Cholecystectomy – index operation (months)
Median (range)	9 (0–294)		6 (0–327)

ERCP, endoscopic retrograde cholangio-pancreatography; US, ultrasound; PTC, percutaneous cholangiography; MRCP, magnetic resonance cholangio-pancreatography; CT, computed tomography.

Operative variables are shown on Table 2. Injury to the right hepatic artery was assessed intra-operatively and was present in 22% of patients. No other vascular injuries were documented. Operative mortality over the whole time period was 3/312 (1%). One patient died of acute liver failure in the first period shortly after a right hepatectomy was performed at the index operation for an atrophied right lobe secondary to a stenosed right hepatic duct. The second and third patients died 2 and 3 months after surgery as a result of septic complications directly related to their injury; both had E4-E5 bile duct injuries and one of them required reconstruction of the separated hepatic ducts with a neoconfluence.

Table 2.

Intra-operative variables stratified in two time periods

	Group I 1990 to 2004		Group II 2005 to 2008		P (0.001)
	No. (n = 172)	%	No. (n = 140)	%
Partial IV-V segment resection	72	42	99	71	0.020
Transanastomotic stent	72	42	3	2	0.001
Hepatectomy	4	2	1	1	0.001
Right	2	–	0	–	–
Left	2	–	1	–	–
Vascular injury	3	–	0	–	–
Intrahepatic biliary stenosis	1	–	0	–	–
Atrophy	1	–	1	–	–
Abscess	2	–	1	–	–
Portoenterostomy	27	16	12	9	0.060
Neoconfluence	4	2	15	11	0.003
Double barrel hepaticojejunostomy	8	5	1	1	0.045
Operative mortality	3	2	0	0	0.255

Post-operative variables are shown in Table 3. The proportion of patients lost to follow-up that were asymptomatic and with no documented complications on their last visit was GI 78% and GII 87%. Twelve patients presented with secondary biliary cirrhosis, of which six died: one remaining patient had a successful liver transplant, and the other five are currently being followed and treated for their liver disease (two are on the transplant list). Mortality during follow-up was considered and is included in the losses. Six and one patient died in GI and GII, respectively, during follow-up. All except one of these patients died from complications directly related to their injury, either secondary to end-stage liver disease, cholangitis and one patient of cholangiocarcinoma (a year after the index operation where cancer was not documented). Five of seven of the patients that died during follow-up had complex lesions (E4, E5): four had a portoenterostomy and one had a post-operative hepatectomy.

Table 3.

Post-operative data stratified in the two time periods

	Group I 1990–2004		Group II 2005–2008		P (0.001)
	No. (n = 172)	%	No. (n = 140)	%
Post-operative complications
Cholangitis	53	31	19	13	0.001
Cholangitis/year (2 year follow-up) %	11		6		0.310
Stenosis	23	13	7	5	0.010
Abscesses	12	7	8	6	0.479
Fistula	9	5	4	3	0.396
Biloma	14	8	18	13	0.192
Reintervention after index operation	16	9	9	6	0.400
Redo hepaticojejunostomy	6		4
Portoenterostomy	2		0
Abscess drainage	2		1
Liver resection	4		0
Hernioplasty	2		2
Bleeding	1		2
Intestinal occlusion	1		0
Median (range) first repair-reintervention in months	24 (0.1–174)		0.3 (0.1–11)
Other treatments
Percutaneous drainage	8	5	8	6	0.797
ERCP (stents)	0	0	3	2	0.893
Transplant	1	1	0	0	1.000
Follow-up (months)
Median (range)	52 (12–224)		25 (6–70)
Lost to follow-up	24	14	8	6	0.020

ERCP, endoscopic retrograde cholangio-pancreatography.

Discussion

The present study demonstrates a learning curve at a referral centre for bile duct injury repair. As it has evolved from a low to high volume the outcomes have improved. Both the growing number of patients and the accumulated experience of the multidisciplinary team have impacted favourably on the outcomes in this series.

Winslow and Strasberg have recently stated that the technical aspects of repair are essential for early and long-term success,¹⁶ these include: well-vascularized ducts, no tension, an anastomosis with the largest possible diameter and complete drainage of the biliary tree. Proper epithelium to mucosa apposition using sutures that produce minimum reaction are also recommended to achieve good long-term results (Fig. 2).

No differences were found in the general characteristics of the patients when comparing groups. Although the percentage of laparoscopic cholecystectomies has increased through time, to this date the majority of patients referred to our centre for repair still had open surgery. This is partially explained by a lack of resources and equipment in smaller cities and rural areas as well as a tendency to perform open surgery in patients with complex disease.

Secondary repair is the most frequent scenario in this series. In many centres, patients with a history of bilioenteric anastomosis are treated radiologically with percutaneous dilation and/or transhepatic transanastomotic stent placement for subsequent dilation.²⁷ However, in this series, most patients with a strictured bilioenteric anastomosis performed before referral required surgical repair. By following the technical characteristics described above, the frequency of stricture and anastomotic dysfunction decreases significantly after the index operation.

Acute repair has remained constant since the beginning of this series. Occasionally the authors are called during the course of a cholecystectomy once or twice a year to perform a repair at other hospitals that are close enough to make this option plausible. These patients are not included in this series as their repair was not performed at this institution and follow-up data are not available. Most patients are referred after conversion and drain placement or after an attempt of repair. The time from a cholecystectomy to the index operation was significantly lower in the last period probably because of a quicker referral and the tendency to repair lesions without delay. Variability of these numbers is evident in different series and change depending on the centre, the city and the country analysed.^6,28,29

Although a Roux-en-Y hepaticojejunostomy has been used as the preferred method of repair since the beginning of this series, the operative technique has evolved significantly. At first, end-to-side anastomoses with transanastomotic stents were used for E1–E3 injuries. In patients with E4 or E5 injuries, a portoenterostomy was performed with transhepatic stents placed through the intestinal lumen. As the authors realized that the end-to-side anastomosis was sometimes jeopardized because of impaired biliary vascularity,³⁰ an adjustment to a more proximal anastomosis was constructed at the confluence of the biliary tree where vascular intergrity was preserved.

Early in the this series, transhepatic and transanastomotic stents were placed.³¹ The rational being that leaks of small bilioenteric anastomoses promote stricture and that both the lowering of the intraductal pressure and an adequate flow through the anastomosis were warranted by the stents. The opportunity to obtain an anastomosis at the hilar level, as described previously, meant stents were no longer used. A wide anastomosis permits free flow of bile, produces low pressure and decreases the chance of a leak. It also minimizes the risk of a stricture and the need of subsequent instrumentation. Nevertheless, some patients with loss of confluence and complete destruction of the isolated right and left hepatic duct need a portoenterostomy. As there is no chance of obtaining a high-quality anastomosis in these patients, ducts are stented to obtain long-term patency. The experience from this series is that the long-term prognosis in these patients is poor.

As technical aspects of the repair progressed and the aproximation of separated right and left ducts to create a single duct to anastomose to the jejunal limb (neoconfluence) became more frequent, the need for other alternatives for repair in E4-E5 such as a portoenterostomy, double barrel anastomoses or a hepatectomy was reduced.

The wedge resection of liver segment IV increased considerably during the last time period and its incidence is probably higher than any other series. The authors need for a high-quality latero-lateral anastomosis in the hilum is thus accomplished while also permitting the lowering of the hilar plate. The resection is easy to achieve, allowing an anteroposterior view of the confluence (instead of a cephalocaudal view) that permits a wide opening of a healthy left hepatic duct with good vascular supply. In very high lesions where the right and left hepatic ducts are isolated, a wedge resection may allow the formation of a neoconfluence for a single hepatico-jejunal anastomosis. The amount of tissue resected is usually quite small but depends on the shape and size of each liver. When the base of segment IV is overdeveloped it may obstruct the adequate dissection of the left duct and prevent the jejunal limb from fitting comfortably for the anastomosis (Fig. 1). For these reasons partial resection of liver segment IV has now become a routine part of the reconstruction for proximal anastomoses. Hepatectomies were necessary in a few patients either during the primary repair or after the index operation. The latter stuation would arise commonly when the patient presented with chronic bile obstruction, hepatic atrophy or persistent cholangitis. Only one of the nine patients submitted to liver resection either during or after the index operation died in the post-operative period.

Selection and timing of the operation has also improved. Operating on patients without optimizing their condition in situations of severe sepsis and multi-organ failure probably explains the mortality in the first period of review. During the second period, resolution of these issues became mandatory before surgical repair. Better outcomes over this period, with a reduction of stenosis, and lower rates of reintervention and mortality, can be partially explained by the refinement of the surgical technique. Other aspects of patient care have also become standardized and more advanced. The multidisciplinary group of surgeons, internists, endoscopists, radiologists and anesthesiologists experienced at managing such complex cases play a major role in improving results.

Conclusions

The present study shows a learning curve at a referral centre for bile duct injury and supports the fact that patient volume translates into better outcomes. Better outcomes are mainly as a result of multidisciplinary care and changes in technical aspects of repair which have changed considerably through time and portray the authors growing experience and personal preferences. Referral to high-volume centres specialized in bile duct injury repairs should be encouraged.

Conflicts of interest

None declared.