Introduction

Surgical resection is considered the only treatment method that can provide cure for malignant biliary tumors, including gallbladder cancer, due to the lack of efficacy of adjuvant therapies.1–6 Bile duct cancer and gallbladder cancer still tend to be diagnosed at an advanced stage despite recent advancements in diagnostic imaging modalities, and R0 resection rates and long-term survivals remain low in patients with advanced stage disease.2,6–9 Therefore, aggressive surgical approaches have been tried to increase resectability.

During the past few decades, many advances have been made in perioperative patient management and our understanding of the anatomies of the hepatobiliary system and of the pancreatoduodenal region have been much improved, to the extent that synchronous hepatic resection and pancreatoduodenectomy can now be performed.

Hepatopancreatoduodenectomy (HPD) has been performed to achieve radical resection in advanced biliary malignancies. However, it remains controversial whether HPD is necessary for the treatment of locally advanced bile duct or gallbladder cancer. Despite promising reports on the long-term results of HPD, many surgeons still question its clinical usefulness because of unsatisfactory long-term benefits, and high morbidity and mortality rates.6,10,11 Although HPD theoretically offers the opportunity to cure patients with an advanced biliary tract cancer, surgical candidates that might benefit from HPD have not been characterized. Accordingly, we investigated the clinical outcome of HPD in patients with biliary malignancies to identify those factors that affect long-term outcome.

Materials and Methods

Between January 1995 and December 2007, 576 patients with gallbladder cancer or extrahepatic bile duct cancer underwent hepatic resection and/or pancreatoduodenectomy at Seoul National University Hospital, Seoul, South Korea. Of these 576 patients, 285 underwent only hepatic resection, 254 underwent only pancreatoduodenectomy, and 37 underwent synchronous hepatic resection with pancreatoduodenectomy. Among the 37 patients, 23 patients underwent synchronous major hepatic resection with pancreatoduodenectomy (HPD). There were 14 males and nine females of mean age 58.4 ± 10.4 years (range 27–72). Of these 23 patients, 10 (43.5%) had gallbladder cancer and 13 (56.5%) had extrahepatic bile duct cancer. Two bile duct cancer patients underwent HPD as second operation for recurrent bile duct cancer.

All 23 patients underwent biliary–pancreas protocol triphasic computed tomography (CT), as described in our previous study.12 In addition, abdominal ultrasonography, abdominal magnetic resonance imaging (MRI) with MR cholangiography, direct cholangiography, and choledochoscopy were also performed selectively to determine tumor extents, and the presence of lymph node and distant metastasis.

Indications for HPD were as follows: (1) gallbladder cancer with direct invasion to an adjacent organ, such as the duodenum, pancreas, or liver; (2) papillary bile duct cancer with multiple primary lesions, or diffuse extrahepatic bile duct and hilar cancer, except for that with bi-lobar intrahepatic bile duct invasion; (3) gallbladder cancer and hilar bile duct cancer with peripancreatic lymph node metastasis which was diagnosed at intraoperative frozen biopsy; and (4) Karnofsky performance score 70 or more. Contraindications for HPD were distant metastasis, invasions on both sides of the portal vein or common hepatic artery, contra-lateral major vessel involvement with opposite side bile duct invasion, severe comorbidities, or chronic liver disease.

Operative procedures for HPD are shown in Table 1. Regarding pancreatoduodenectomy, 10 patients (43.5%) underwent a conventional Whipple’s operation and 13 (56.5%) underwent pylorus-preserving pancreatoduodenectomy (PPPD). In all patients, major hepatectomies more than hemihepatectomy were performed as operative procedures. Regional lymph nodes were removed down to the right side of the celiac and superior mesenteric arteries, and all tissues in the hepatoduodenal ligament, except the portal vein and the hepatic artery, were removed.2

Table 1 Operative procedures of HPD
Full size table

With the exception of the patients with a poor performance status including hospital mortalities (n = 13), or chemoradiotherapy refusal (n = 3) or an early TNM stage (n = 2), the others underwent chemoradiation therapy. Four (all bile duct cancer) of the 23 patients underwent chemoradiotherapy based on 5-FU, and one gallbladder cancer patient was treated with gemcitabine. Chemoradiotherapy was administered according to a standardized schedule and dosage, as described in our previous report.2

Pathologic staging was performed according to the sixth edition of the AJCC TNM staging system, and patients who had developed postoperative complications were classified from grade I to grade V according to the Clavien classification.13 Clinico-pathologic data were collected prospectively in a standardized form of electronic medical records with the approval of our Institutional Review Board, and retrospective analysis was performed to determine the short- and long-term outcomes of HPD. Statistical analysis was performed using SPSS® 17.0 (SPSS, Chicago, IL, USA) for Windows, and p values of <0.05 were considered statistically significant.

Results

Detailed Indication of HPD and Resectability

Detailed indications of HPD by diagnosis are shown in Table 2. In gallbladder cancer, peripancreatic lymph node metastasis (n = 4) and diffuse bile duct infiltration along the mucosa or submucosa (n = 4) were the common indications. For the bile duct cancers, HPD was performed because of diffuse bile duct infiltration (n = 11) and bile duct cancer recurrence (n = 2). Two patients with recurrent bile duct cancer had underwent hilar bile duct resection with stage IA diseases, and recurrences at resection margins were detected at 26 and 66 months after first operations, respectively. Both patients underwent synchronous hepatic resection and pancreatoduodenectomy with curative intent.

Table 2 Indications of HPD and R0 resection rates by diagnoses
Full size table

R0 resection was possible in 17 of 23 patients (73.9%), and six patients (26.1%) underwent surgical resection with microscopic residual disease (R1). According to diagnoses, R0 resection was possible in nine of 10 gallbladder cancer patients (90.0%) and in eight of 13 bile duct cancer patients (61.5%) (p = 0.179). One patient with R1 resection also had para-aortic lymph node metastasis, which was negative by frozen biopsy during operation. R1 resection was performed in one gallbladder cancer patient and in five bile duct cancer patients. All underwent HPD because of diffuse bile duct infiltration.

Pathologic Staging

Among 10 gallbladder cancer patients, nine patients had stage IIB or more diseases. One patient with stage IB underwent HPD because of diffuse bile duct infiltration along the mucosal surface. Six patients with stage IIB gallbladder cancer underwent HPD because of retropancreatic lymph node metastasis (n = 4) and diffuse bile duct infiltration (n = 2). Stages were evenly distributed among the 13 bile duct cancer patients. All patients with stage IA underwent HPD because of recurrence at bile duct resection margins, and all with stage IIA or IIB patients underwent HPD because of diffuse bile duct infiltration. There were two stage IV bile duct cancer patients, one had para-aortic lymph node metastasis and the other had intrahepatic metastasis to a resected liver, which was diagnosed by postoperative permanent biopsy.

Morbidity and Mortality

A postoperative complication occurred in 21 of the 23 patients (91.3%). The common problems were abdominal fluid collection, pleural effusion, and pancreatojejunostomy leakages, but most complications responded to conservative management. Two patients with a duodenojejunal anastomosis leakage and a postoperative bleeding required re-exploration. Liver failure, which was diagnosed by the criteria suggested by van den Broek et al.14, occurred in two patients (8.7%); both underwent right trisectionectomy with PPPD. One of these recovered with conservative care, but the other, who had undergone preoperative right portal vein embolization, succumbed to progressive liver failure. Another two patients with pneumonia and a liver abscess died at 96 days and 38 days after operation, respectively. Therefore, the in-hospital mortality rate was 13.0%. According to the complication grading system devised by Dindo et al.13, there were 10 (43.5%) Grade IIIa patients who need radiologic intervention or wound repair or nasogastric tube insertion, followed by five (21.7%) grade I patients who recovered without any pharmacologic treatment or intervention, and three (13.0%) grade V patients who died in hospital. There were also one grade II patient who required antibiotic therapy, one grade IIIb patient who required relaparotomy due to postoperative bleeding, and one grade IVa patient who had hepatic failure and duodenojejunostomy leakage.

Survival

All patients, except three who died in hospital, were followed up for a median 15.0 months (3–117 months). Overall cumulative 1-, 3-, and 5-year survival rates were 65.2%, 33.5%, and 22.4%, respectively. Cumulative 1-, 3-, and 5-year survival rates were 76.9%, 43.1%, and 32.3% (median = 27.0 months) in bile duct cancer, and 50.0%, 20.0%, and 10.0% (median = 9.0 months) in gallbladder cancer, respectively. Bile duct cancer showed better survival than gallbladder cancer, but this was not significant (p = 0.088). The 3- and 5-year survival rates of all patients were 39.2% and 31.4% in R0 resection group, and 16.7% and 0% in the R1 resection group (p = 0.196) (Fig. 1). Adjuvant chemoradiotherapy did not significantly influence survivals of those with gallbladder cancer (median; chemoradiation vs. no chemoradiation—15 vs. 9 months, p = 0.942) or bile duct cancer (median—16 vs. 27 months, p = 0.627). Two recurrent bile duct cancer patients who were diagnosed at 26 and 66 months after first operations survived 80 and 117 months after HPD with no evidence of recurrences, respectively, which were 108 and 183 months after first operations, respectively.

Fig. 1
figure 1

Survival curves according to the R-status. The 3- and 5-year survival rates of all patients were 39.2% and 31.4% in the R0 resection group and 16.7% and 0% in the R1 resection group. No statistical difference was evident between resection groups (p = 0.196)

Full size image

Gallbladder Cancer

One gallbladder cancer patient with stage IB disease survived 104 months. For stage IIB, 3- and 5-year survival rates were 16.7% and 0%, respectively. All patients with stage III recurred and expired within 15 months. Patients with early stage disease survived longer than those with advanced stage disease, but this was not statistically significant (p = 0.233). In addition to two in-hospital deaths due to hepatic failure and liver abscess, one patient died of a varix bleeding without tumor recurrence 8 months after operation. All the remaining seven patients had recurrences, and most of the recurrences were due to distant metastases (n = 6) rather than loco-regional recurrence (n = 1).

Bile Duct Cancer

All patients with early stage disease of stage IIA or less survived more than 24 months. By contrast, all patients with advanced stage disease of stage IIB or more died within 2 years (p = 0.006). In bile duct cancer, 3- and 5-year survival rates were 56.3% and 56.3% in the R0 resection group, and 20.0% and 0% in the R1 resection group, respectively (p = 0.038).

Two patients with stage IA had papillary type cancers, and all survived more than 5 years. Four out of five stage IB and IIA patients with R0 resection survived without recurrence; however, all patients with lymph node metastases or R1 resection had recurrences at 4 to 31 months after operation, and all died at 0 to 11 months after diagnosis of recurrences.

Survival analysis by the log-rank test showed no significant survival difference according to T-stage for gallbladder or bile duct cancer (p = 0.405 and 0.172, respectively). However, we did find a significant survival difference with respect to lymph node metastasis status (Fig. 2); that is, bile cancer patients without lymph node metastasis survived longer than those with lymph node metastasis (p < 0.001), but no significant survival difference was found for gallbladder cancer (p = 0.237). For gallbladder cancer, 3- and 5-year survival rates were 50.0% and 50.0% in the pN0 group, and 12.5% and 0% in the pN1 group, respectively, whereas for bile duct cancer, 3- and 5- year survival rates were 80.0% and 60.0% in pN0 group; all patients with pN1 died within 2 years of HPD.

Fig. 2
figure 2

Survival curves for all patients by lymph node status. Three- and 5-year survival rates were 74.1% and 59.3% in the pN0 group and 7.1% and 0% in the pN1 group (p < 0.001)

Full size image

Discussion

HPD was introduced in order to achieve curative surgical resection in patients with advanced bile duct cancer or gallbladder cancer. Many studies, although the majority are case reports, have described long-term survival after HPD in selected patients.8,15–21 However, many surgeons have not actively performed HPD because of its high morbidity and mortality. These were particularly high during the early pioneering days of HPD, when morbidity rates of 30.8–91.0% and mortality rates of 0–41.0% were reported.11,17,18,20,22–25 In the present series, in-hospital death occurred in three patients (13.0%), and the causes of these deaths were hepatic failure, pneumonia, and liver abscess.

One of the most life-threatening complications of HPD is hepatic failure. According to previous reports, hepatic failure occurs in 10.0–46.8% of patients.4,5,11,26 In the present series, hepatic failure occurred in two of the 23 patients (8.7%); both patients underwent right trisectionectomy with PPPD. One recovered under conservative care, but the other patient who had undergone preoperative right portal vein embolization died from progressive liver failure.

To prevent this lethal complication, many authors have recommended preoperative biliary drainage and portal vein embolization. Preoperative biliary drainage is necessary to improve impaired mitochondrial function, which is vital for liver regeneration after hepatectomy, although it should be added that there is some debate about the benefits of preoperative biliary drainage. Preoperative biliary drainage also might restore intestinal barrier function, and remedy impaired cell-mediated immunity and the increased susceptibility to endotoxemia caused by obstructive jaundice.27–29 Some have found that a total bilirubin level of more than 3.0 mg/dl is significantly correlated with an increased risk of anastomotic leakage after pancreatojejunostomy and of intra-abdominal bleeding caused by anastomotic leakage.5,22 Furthermore, cholangiography, performed using a drainage catheter, with CT and MRI can determine the extent of bile duct infiltration. For these reasons, we performed preoperative biliary drainage in patients with obstructive jaundice. Several studies have been conducted on the efficacy of preoperative portal vein embolization in terms of reducing postoperative hepatic failure,5,21,26 but on the other hand, some have reported that HPD can be performed without preoperative portal vein embolization without inducing mortality.8,20,23 Ebata et al.26 reported that hepatic failure rates decreased from 56% to 14% after introducing preoperative portal vein embolization. However, Nimura et al.4 reported that hepatic failure occurred in 19.0% of patients despite preoperative portal vein embolization. We do not routinely perform preoperative portal vein embolization in all patients scheduled for major hepatic resection; rather, we adopt this course when the presumptive remnant volume after resection is suspected to be smaller than 30% of total liver volume. In the present study, only one patient who underwent right trisectionectomy had been performed preoperative portal vein embolization; nevertheless, he died from hepatic failure. However, extended hepatectomy more than right hemihepatectomy could be performed safely without hepatic failure in most patients who have not undergone preoperative portal vein embolization.

Even though diagnostic modalities, surgical techniques, and perioperative management have improved remarkably, the prognoses of bile duct cancer and gallbladder cancer remain unsatisfactory. This study also showed that survival during the recent 6 years (n = 12) has not improved compared with the previous 7 years (n = 11) for bile duct cancer (median—27 vs. 16 months, p = 0.747) or gallbladder cancer (median—9 vs. 1 months, p = 0.649). However, R0 resection rates have been increased reportedly, and recent studies have reported 61–80% for hilar bile duct cancer and 36–80% for advanced gallbladder cancer.2–4,9,19 In the present series, R0 resection achieved by HPD was possible in 17 of the 23 patients (73.9%). Unexpected microscopic tumor spread as compared with that suspected by preoperative imaging and difficult interpretation of resection margin status by frozen section are regarded the main causes of R1 resection. Frozen section diagnosis is often difficult even for experienced pathologists, and diagnoses of dysplasia grade may differ between pathologists.30 Endo et al.31 suggested that inflammatory stromal infiltration, the presence of atypical cells around a tumor, and a tendency toward submucosal extension contribute to the inaccuracy of intraoperative frozen section analysis. Furthermore, tumor extent often exceeds presumed preoperative extent because mucosal or submucosal microscopic spread is common in bile duct cancer.32 In the present series, permanent biopsy results of R1-resected patients were carcinoma in five and carcinoma in situ with high-grade dysplasia in one; however, frozen section results were carcinoma in one, low to moderate grade dysplasia in three, and high grade dysplasia in two. We were unable to find any statistical survival difference between R0 and R1 resection in overall 23 patients (p = 0.196); however, significant survival difference was observed in the bile duct cancer (p = 0.038).

Many authors have reported that TNM staging is well correlated with survival in bile duct and gallbladder cancer,2,7,9 and a similar tendency was observed in the present series. For gallbladder cancer, patients with an earlier stage showed better survival than those with an advanced stage, though this was not significant (p = 0.233). Some authors have reported that long-term survival can be achieved by HPD in some selected patients with far advanced gallbladder cancer,8,10 but in the present series, the long-term outcomes of stage III gallbladder cancer were disappointing. Debate continues as to whether pancreatoduodenectomy should be performed in patients with gallbladder cancer with peripancreatic lymph node metastasis.25,33 In the present study, it was found that most patients with peripancreatic lymph node metastasis suffered early recurrence even after HPD. Early recurrences within 6 months of surgery in some patients suggest that metastasis might have been present during surgery, but it was not detected by preoperative imaging studies. Some authors have reported that staging laparoscopy with ultrasonography is helpful for the identification of occult dissemination missed during imaging studies.34,35 We have not routinely performed staging laparoscopy before surgery. Although staging laparosopy is probably limited for the detection of vascular invasion, lymph node metastasis, and liver metastasis, it may be helpful in some patients with advanced gallbladder cancer to avoid unnecessary laparotomy.35 For bile duct cancer, our findings show that patients of stages IA, IB, and IIA can expect long-term survival of more than 3 years (p = 0.006). It is believed that surgical removal of potentially involved lymph nodes and surrounding tissue around the hepatoduodenal ligament may be the only chance of long-term survival and cure for those with an extrahepatic bile duct cancer, but it is unclear whether aggressive lymph node dissection for widespread lymph node metastases improves surgical outcome.16 In the present study, patients with advanced bile duct cancer achieve dismal long-term outcomes, and despite D2 plus alpha lymph node dissection, patients with lymph node metastasis in the present series achieved long-term survivals that were no better than those with distant metastasis. In the present study, lymph node metastasis was found to be a significant poor prognostic factor in bile duct cancer (p < 0.001), which concurs with other studies.2,4,17 Therefore, we suggest that the use of HPD in patients with bile duct cancer with lymph node metastasis should be reconsidered.

The role of adjuvant chemoradiotherapy has not been established in patients with resected extrahepatic biliary cancer, and no prospective study has supported the survival benefit of adjuvant chemoradiotherapy, which is in line with our findings.1,36 However, some retrospective studies, including ours, have found a survival benefit for patients who undergo adjuvant chemoradiotherapy after surgical resection, which suggests that adjuvant concurrent chemoradiotherapy has a beneficial role in patients with extrahepatic bile duct cancer.36–38 However, further well-designed prospective randomized trials are required to clarify the efficacy of adjuvant chemoradiotherapy.

We treated two bile duct cancer patients with stage IA; both had papillary type cancers and survived for more than 5 years. In the literature, papillary type or superficial spreading type bile duct cancer is reported to have better survival than the other types.3,15 Papillary carcinomas of the extrahepatic bile duct tend to grow initially in lumen and then invade the wall of the bile duct where they remain for a considerable time, and thus appear to be associated with a better prognosis than other types of adenocarcinoma.39 However, surgeons should be aware that papillary tumors may be multiple and spread long distances superficially, thus, the extent of cancer in the biliary tree should be defined preoperatively to prevent recurrence due to incomplete resection.40

In conclusion, past and current data confirm without doubt that HPD is a highly morbid operation with significant risk of death regardless of preoperative preparation of patients. Unfortunately, HPD did not offer patients with advanced disease any benefit of prolonged survival. Consequently, though a sound conceptual approach for these cancers, clinical experience does not support its use. In the cases requiring a longitudinal resection margin, such as in papillary type bile duct cancer, or in cases with bile duct infiltration from gallbladder cancer, HPD might be a good treatment choice that could attain long-term survival. However, for the patients who had obvious adjacent organ invasion or lymph node metastases which were diagnosed preoperatively or intraoperatively, the application of HPD might not be recommended.