Liver resection is the treatment of choice for patients with hilar cholangiocarcinomas. Depending on the tumor stage and surgical radicality, 5-year survival rates up to 65% are achievable by hilar en bloc resection in selected patients.1 However, in the case of contraindications due to parenchymal damage or local irresectability, liver transplantation (LT) has also been proposed. Initial results of LT for hilar cholangiocarcinoma were disappointing,2, 3 predominantly because of improper patient selection from a present-day perspective.
After either surgical resection or LT, locoregional recurrence in the region of the head of the pancreas is a frequently observed pattern. Therefore, our group has proposed a combination of hepatectomy and pancreatic head resection [extended bile duct resection (EBDR)] to increase the surgical radicality.4 This approach eradicates the entire biliary tree with its lymphatic drainage. It complies with the basic rule of oncological surgery of achieving wide safety margins while avoiding dissection in tumor-bearing areas. A similar approach was also reported by Anthuber et al.5 and Cherqui et al.6
Transplantation is currently being reassessed for inoperable hilar cholangiocarcinoma, especially because of promising data for neoadjuvant radiochemotherapy reported by the Mayo Clinic group.7 However, it is still unclear if EBDR might be useful for improving the long-term outcome of these patients. The long-term results (> 10 years) of our initial cohort of patients were analyzed to clarify the oncological value of this procedure.
CA 19-9, carbohydrate antigen 19-9; EBDR, extended bile duct resection; ELTR, European Liver Transplant Registry; IORT, intraoperative radiation therapy; LT, liver transplantation; ND, no data available; NS, not significant; PHR, pancreatic head resection; PTCD, percutaneous transhepatic cholangiodrainage; TNM, tumor-node-metastasis; UICC, Union Internationale Contre le Cancer.
PATIENTS AND METHODS
The long-term outcome of 16 consecutive patients (7 females and 9 males with a mean age of 50 ± 2 years) with hilar cholangiocarcinoma who underwent hepatectomy plus Whipple's procedure followed by LT was analyzed with a prospective database. These patients were transplanted between March 1992 and November 1998 and had, therefore, a minimum follow-up of 10 years. None of these patients received any neoadjuvant or adjuvant treatment during the study period. No patient had a history of primary sclerosing cholangitis.
The following patients with hilar cholangiocarcinoma transplanted during the same period were excluded from the present analysis because their tumor characteristics were not comparable to those of the present series: 4 patients with incidental cholangiocarcinoma detected only in the explanted liver by the pathologist, 2 patients who underwent transplantation in the early phase because of recurrent cholangiocarcinoma after previous liver resection, and 1 patient with hilar cholangiocarcinoma and primary sclerosing cholangitis.
The preoperative diagnostic workup included a chest X-ray, endoscopic retrograde cholangiography or percutaneous transhepatic cholangiodrainage with brush cytology, angiography, abdominal computed tomography and/or magnetic resonance imaging, and extensive laboratory investigations including carbohydrate antigen 19-9 (CA 19-9) and carcinoembryonic antigen levels. In patients with suspicion for disseminated intra-abdominal disease, diagnostic laparoscopy was performed prior to transplantation.
EBDR was considered when the procedure was likely to be curative. Regional lymph node involvement and intrahepatic metastases were not considered contraindications during this early period as long as the overall surgical strategy had a curative intention. The surgical technique of EBDR has been described in detail previously.4 All patients underwent regular follow-up investigations at our department at least every 6 months.
Matched Pair Analysis
To compare the results of EBDR (group I), a matched pair analysis was performed with patients transplanted for hilar cholangiocarcinoma without simultaneous partial pancreatoduodenectomy (group II). These patients (n = 8) were transplanted in a later period between June 1995 and September 2007. All of these patients underwent transplantation with the preoperative diagnosis of hilar cholangiocarcinoma. Patients with cholangiocarcinoma found incidentally in the explanted liver were excluded. Patients of both groups were matched 1:1 according to tumor stage, gender, and age. On account of the low number of patients available, 1 gender mismatch and 2 age mismatches > 10 years were tolerated because priority was given to tumor stage matching.
All data are given as means and standard errors of the mean. Patient survival was calculated with the Kaplan-Meier method, and survival data were compared by the log rank test. Comparisons of categorical and continuous variables were performed with the χ2 test and the Mann-Whitney U test, respectively. Differences were considered statistically significant if the P value was less than 0.05. All statistical analyses were performed with SPSS 16.0 (SPSS, Inc., Chicago, IL).
According to the Bismuth-Corlette classification,8 the patients were classified as type IIIa (7 patients), type IIIb (3 patients), and type IV (6 patients). The pathological analysis revealed the following Union Internationale Contre le Cancer (UICC) stages: stage I in 6 patients (38%), stage IIA in 5 patients (31%), stage IIB in 3 patients (19%), and stage IV in 2 patients (13%). One of the latter 2 patients had an ovarian metastasis that was detected not during the primary operation but during an operative revision a few days later. In the other stage IV patient, the operation was too advanced when a positive para-aortal lymph node was detected at the left renal vein. Both distant metastases were resected, but the procedure was still not considered to be curative. Detailed patient and tumor characteristics are given in Table 1. In the other 14 patients (88%), the resection was formally curative.
Table 1. Clinical and Pathological Data for the 16 Consecutive Patients Undergoing LT with a Simultaneous Whipple Procedure
Date of LT
CA 19-9 (Pre-LT)
Pathological TNM Stage
Manifestation of Tumor Recurrence
Abbreviations: CA 19-9, carbohydrate antigen 19-9; LT, liver transplantation; ND, no data available; PTCD, percutaneous transhepatic cholangiodrainage; TNM, tumor-node-metastasis; UICC, Union Internationale Contre le Cancer.
All except 2 patients after EBDR experienced 1 or more postoperative complications. Most severe complications were results of the simultaneous pancreatic head resection. As described previously, 2 patients died postoperatively (90-day mortality = 12.5%).9 One patient died because of multiorgan failure 1 month after LT (after previous total pancreatectomy due to a pancreatic leak), and 1 patient died 12 days after transplantation because of hemorrhaging from the hepatic artery; both were caused by pancreatic leakage. In another 6 patients, a complication as a result of simultaneous pancreatic surgery evolved. Three of these patients had pancreatic leakage, which in 2 cases required total pancreatectomy 7 and 11 days after LT and in 1 case was treated conservatively. The other 3 patients had the following pancreas-related complications: 1 patient required total pancreatectomy because of necrotizing pancreatitis, 1 required surgical revision because of bleeding from the pancreas cutting edge, and 1 experienced pancreatitis and was treated conservatively. In all, the surviving patients with pancreas-related complications required significantly longer hospitalization (65 ± 6 days) than those without pancreas-related complications (36 ± 3 days, P = 0.03 by the Mann-Whitney U test). Other complications treated without operative revision were acute rejection episodes (n = 4), severe diarrhea (n = 6), cholangitis (n = 3), wound infection (n = 2), postoperative bleeding (n = 2), cytomegalovirus infection (n = 1), and pulmonary legionellosis (n = 1).
The overall 1-, 5-, and 10-year patient survival after EBDR was 63%, 38%, and 38%, respectively (Fig. 1). During the follow-up period, 12 patients died. Earlier deaths were due to postoperative complications in 2 patients (discussed previously) and tumor recurrence in 8 patients. Two other patients died while recurrence-free more than 10 years after transplantation for reasons not related to the cholangiocarcinoma (de novo urothel carcinoma and ruptured aortic aneurysm). Among the 6 patients with UICC stage I tumors only, 1 patient with a preoperative CA 19-9 level of 8249 U/mL developed tumor recurrence, but 2 patients died because of early postoperative complications (discussed previously). Therefore, 5- and 10-year survival was 50% in stage I patients and not significantly higher than that in UICC stage IIA (40%) patients. No patient with UICC stage IV survived longer than 14 months (Fig. 2). When all lymph node–negative patients (UICC stages I and IIA) were considered after the exclusion of perioperative deaths, 10-year survival was 56% (Fig. 2). In all, 6 patients survived recurrence-free more than 10 years after transplantation.
The following factors showed a trend toward inferior survival: distant metastases, positive lymph nodes, CA 19-9 levels > 1000, and preoperative percutaneous transhepatic cholangiodrainage (instead of endoscopic retrograde cholangiography). However, because of the low patient numbers, these differences were not statistically significant, and a multivariate analysis was not considered useful.
Case Control Analysis
Both groups revealed comparable patient and tumor characteristics (Table 3); however, in group I (EBDR), significantly more patients had a poorly differentiated carcinoma, and the mean CA 19-9 level was markedly higher, indicating unfavorable tumor characteristics (Table 2). After LT alone (group II), markedly lower perioperative mortality was observed, but the long-term outcome was similar (Figure 3). However, the follow-up in group II was significantly shorter. The only long-term survivor with positive lymph nodes (lymph node ratio = 2/27) was seen in group I. One patient with positive lymph nodes in group II is also alive and recurrence-free, but the follow-up period is only 11 months at present.
Table 2. Patients Included in the Matched Pair Analysis
Group I (LT + Whipple)
Group II (LT)
NOTE: Eight patients transplanted in the later period without partial pancreatoduodenectomy were matched to respective patients undergoing extended bile duct resection by the Union Internationale Contre le Cancer tumor stage, age, and gender.
Table 3. Patient Characteristics in the Two Subgroups of the Matched Pair Analysis
Group I (LT + Whipple; n = 8)
Group II (LT; n = 8)
NOTE: Differences (P < 0.05) by univariate analysis are indicated.
Abbreviations: CA 19-9, carbohydrate antigen 19-9; LT, liver transplantation; NS, not significant; UICC, Union Internationale Contre le Cancer.
Mean age (years)
52 ± 3
47 ± 3
Tumor stage (n)
Tumor grading (n)
Not determined (n)
1530 ± 619
117 ± 89
Tumor recurrence (n)
170 ± 9
66 ± 16
Surgical resection and LT are the only potentially curative treatment options for patients with hilar cholangiocarcinoma. This study presents the long-term results of a cohort of patients undergoing EBDR for inoperable central bile duct tumors without underlying primary sclerosing cholangitis. Currently, no effective medical treatment for hilar cholangiocarcinoma is available, and median survival with palliative chemotherapy or radiochemotherapy is 9 to 12 months with no long-term survivors.10
The overall results in this series were encouraging in lymph node–negative patients. The 10-year survival of 14 patients without lymph node metastasis and without distant metastasis (UICC stages I and IIA) undergoing either EBDR (n = 11) or LT alone (n = 3) was 56%, including perioperative mortality. The overall survival was still worse than that observed, for example, in patients transplanted for hepatocellular carcinoma within the Milan criteria,11 for whom 5-year survival rates of 70% or more have been confirmed in many studies. However, also in patients transplanted for benign diseases, inferior results are accepted under some circumstances. For example, patients transplanted with decompensated liver cirrhosis and a Model for End-Stage Liver Disease score of more than 25 have significantly impaired survival of 75% in the first year.12, 13 This group represents a significantly larger number of patients and, therefore, demand for organs in comparison with the small number of patients with cholangiocarcinomas appropriate for LT.
The results of the present series compare well to 5-year survival rates after curative resection of hilar cholangiocarcinoma, which lie between 20% and 65% in selected cases. Because of the increasing organ shortage, cholangiocarcinoma is not an accepted standard indication for LT. However, from an oncological point of view, long-term survival close to 60% is still an excellent result for these otherwise irresectable tumors. Notably, none of the patients in the present study received any kind of neoadjuvant or adjuvant therapy (discussed later).
The value of increasing surgical radicality in the context of LT seems to be limited, as suggested in the present analysis and several other studies. Even so-called upper abdominal exenteration, including resection of the liver, stomach, spleen, pancreas duodenum, and part of the colon, as described by Starzl's group,14 did not improve long-term results. In this series, the 90-day mortality was 18%, and long-term survival (defined as survival > 4 years after transplantation) was seen only in 3 lymph node–negative patients out of 20 cholangiocarcinoma patients. However, many of these patients had tumors of very advanced stages and eventually multiple intrahepatic metastases, and this hinders a direct comparison with the present series. Subsequently the procedure of upper abdominal exenteration was modified to reduce the surgical risk. This led to the introduction of techniques such as the pylorus-preserving cluster operation15 and EBDR.4 Although the initial experience of cluster exenteration was due to the surgical necessity of curatively resecting advanced tumors, the latter techniques are rather aimed at increasing the radicality by en bloc resection of the hepatoduodenal ligament, even in patients with tumors of early stages. Because of the low number of patients available and the increased perioperative mortality after extended procedures, the oncological benefit cannot be finally judged. Our case control analysis could not show a general benefit of EBDR in terms of long-term patient survival in comparison with LT alone. This was mainly based on the increased early postoperative mortality after EBDR. The long-term results were nevertheless equal, possibly pointing to higher oncological radicality. Likewise, the only long-term survivor with positive lymph nodes was in the group with EBDR. However, on the basis of past experiences confirmed by the present data, lymph node–positive patients clearly represent a contraindication for LT nowadays; therefore, this issue is of scientific interest only. Altogether, even if the perioperative mortality might be reducible with newer surgical techniques, there is no good evidence that more radical resections alone are able to markedly improve long-term results.
The general interest in LT for hilar cholangiocarcinoma is again increasing, as documented by several publications on that issue within the last year. The analysis of the published studies16-32 confirmed the strong correlation between the proportion of patients without metastasis (lymph node or distant) and patient survival (Table 4 and Fig. 4), emphasizing the fundamental importance of patient selection. Additionally, more recent data have shown a tendency toward improved overall survival. In parallel with our matched pair analysis, studies with surgical interventions extended beyond LT showed no tendency toward improved patient survival if nodal status was considered (Fig. 4). However, a comparison of the different results is difficult because many studies do not separately indicate survival rates for distinct tumor stages. From the cumulative experience with LT for hilar cholangiocarcinoma, it is clear that long-term survival in a significant proportion of patients is achievable only in nodal-negative patients. This was demonstrated first by groups from Pittsburgh14 and Hannover16 and thereafter confirmed by several experiences, including the present one. The Hannover experience revealed 3- and 5-year overall survival of 56% and 44%, respectively, in lymph node–negative patients, whereas no patient with positive lymph nodes survived longer than 16 months. The impaired survival in comparison with the present experience might be caused by several factors. First, the Hannover series is relatively inhomogeneous because, for example, patients with incidental cholangiocarcinoma and patients after previous incomplete tumor resection were included. In addition, the data were derived from a relatively early period of LT (the inclusion period is 1975-1993); because no detailed recurrence rates are given, the improved modern perioperative management might contribute to better results in more recent series such as the present one.
Table 4. Overview of Published Data for LT and Extended Transplantation Procedures for Hilar Cholangiocarcinoma
M+ or N+ Patients (%)
Postoperative Mortality (%)
1-Year Survival (%)
3-Year Survival (%)
5-Year Survival (%)
NOTE: Case reports, studies on tumors found incidentally after transplantation, and studies published before 1996 were excluded.
Abbreviations: ELTR, European Liver Transplant Registry; IORT, intraoperative radiation therapy; LT, liver transplantation; ND, no data available; PHR, pancreatic head resection.
In the present series, a high CA 19-9 level was an indicator of poor survival after transplantation. Of 5 patients with CA 19-9 units/mL > 1000, none survived longer than 13 months, and the median overall survival was only 6 months. In contrast, in patients in whom CA 19-9 was below 1000 or not determined before transplantation, the median overall survival was 94 months. Unfortunately, in most of the long-term survivors, no preoperative CA 19-9 was determined; therefore, no cutoff can be derived from the present data. The Mayo Clinic experience is the only one that provides explicit data on CA 19-9 levels. This group found a CA 19-9 units/mL level > 100 before transplantation but not before the start of neoadjuvant treatment to be a significant predictive factor for tumor recurrence (hazard ratio = 4.503).7
All in all, the increase in surgical radicality has failed to improve long-term results. Therefore, extended resections should be reserved for patients with a positive frozen section examination of the bile duct margin, as practiced in the Mayo Clinic protocol in about 20% of patients.27 The major issue of LT for cholangiocarcinomas remains thorough patient selection. As shown in the present analysis and several other studies, patients with positive lymph nodes or distant metastases are clearly no candidates for LT. However, above this, no generally accepted selection criteria exist for patients with cholangiocarcinoma. Analogously to hepatocellular carcinoma criteria, a maximal tumor diameter of 3 cm or less in computed tomography or magnetic resonance imaging has been suggested as a surrogate parameter. In addition, after a previous attempt at tumor resection or transperitoneal biopsy, LT is discouraged. In contrast, patients with early tumors (pT1N0 or pT2N0) have a low risk of tumor recurrence. The subgroup of T3N0 patients seems to have an intermediate risk of recurrence. This subgroup might especially benefit from tumor biological selection parameters or selection by neoadjuvant treatment.
There are 2 major experiences using the approach of neoadjuvant radiotherapy or radiochemotherapy for patient selection. This is reasonable because several studies have demonstrated the potential efficacy of radiotherapy with or without chemosensitization, mainly as palliative therapy.10 One LT experience originates from the University of Nebraska, where a combination of neoadjuvant biliary brachytherapy and 5-fluorouracil chemotherapy resulted in a long-term survival rate of 45%.22 The neoadjuvant therapy was accompanied by a relatively high rate of mainly septic complications preoperatively and postoperatively. However, the overall outcome in the reported 11 lymph node–negative patients was not better than that in the present series without any (neo)adjuvant therapy.
More favorable results have been reported from the Mayo Clinic, where a combination protocol of neoadjuvant brachytherapy (target dose = 2000-3000 cGy), external beam radiation (4500 cGy), and chemotherapy (oral capecitabine, 2 g/day for 2 of every 3 weeks until transplantation) was applied in combination with strict patient selection. This patient selection was ensured by a mandatory staging laparotomy after completion of neoadjuvant therapy with complete abdominal exploration and routine biopsy of regional lymph nodes, and it “is a major reason for the success of the Mayo Clinic … treatment protocol.”33 Therefore, about one-third of the patients were removed from the waiting list because of tumor progression, whereas the transplanted patients showed an excellent 5-year survival rate of 76%.7 These excellent results in highly selected patients with localized, lymph node–negative hilar cholangiocarcinoma could not be reached in the present series. Besides the fact that no neoadjuvant therapy was applied in our patients, the Mayo patient series included mainly primary sclerosing cholangitis patients (about 60%7, 27) and presumably a certain number of tumors of very early stages because in 16 of 38 patients (42%), no residual tumor was found in the pathological examination of the recipient liver27; therefore, the Mayo series is not directly comparable to the present series. Because in both studies using neoadjuvant therapy7, 22 no control group existed, it is still not finally clarified if the results are really improved by neoadjuvant therapy or are mainly improved by the strict patient selection. Alternatively, one could speculate that by the use of neoadjuvant therapy, patients with good tumor biology are selected, as also supposed for hepatocellular carcinoma patients after transarterial chemoembolization.34
In conclusion, because of the promising results of several single-center experiences, it seems not to be justified to categorically exclude patients with cholangiocarcinoma as candidates for LT. Extended surgical procedures in combination with LT, as shown in the present series and several other series, do not improve the overall survival because of significantly increased perioperative mortality. Protocols of either adjuvant or neoadjuvant (radio)chemotherapy need to be established to improve the outcome before LT might become a generally accepted therapy in properly selected patients with hilar cholangiocarcinoma. The Mayo protocol is an important step toward this goal, but it is accompanied by severe side effects. Therefore, effective, less toxic protocols would be desirable, but they have not been investigated so far. Further prospective trials are urgently needed and, because of acceptable long-term results, are clearly justified in lymph node–negative patients. However, the low number of patients in individual centers necessitates multi-institutional/multinational trials.