- Email: [email protected]
Portal Vein Resection in Management of Hilar Cholangiocarcinoma
Portal Vein Resection in Management of Hilar Cholangiocarcinoma Alan W Hemming, MD, MSC, FACS, Kristin Mekeel, MD, Ajai Khanna, MD, FACS, Angeles Baquerizo, MD, Robin D Kim, MD, FACS Vascular reconstruction along with major liver resection in the setting of liver dysfunction caused by biliary obstruction can be associated with increased risk. The purpose of this report is to assess the role of portal vein resection and reconstruction in the surgical management of hilar cholangiocarcinoma. STUDY DESIGN: Ninety-five patients with hilar cholangiocarcinoma who underwent resection between 1999 and 2010 were reviewed. Liver resections performed along with biliary resection included 84 trisegmentectomies (63 right, 21 left) and 11 lobectomies (8 left, 3 right). Thirteen patients also had simultaneous pancreaticoduodenectomy performed. Forty-two patients underwent portal vein resection and reconstruction. Five patients required reconstruction of the hepatic artery. Preoperative portal vein embolization was used in 38 patients. RESULTS: Patients undergoing resection had a 5% mortality rate, with an overall morbidity rate of 36%. Patients who underwent portal vein resection had perioperative mortality and morbidity similar to those who did not have portal vein resection. Median survival was 38 months (95% CI, 29–51 months), with a 5-year survival rate of 43%. There was no difference in long-term survival between those patients who had portal vein resection and those that did not. Negative margins were achieved in 84% of cases and were associated with improved survival (p ⬍ 0.01). Five-year survival rate in patients undergoing R0 resection was 50%. Patients with positive lymph nodes appeared to have a worse 5-year survival rate than patients with node-negative status (23% versus 49%); however, only negative margin status was associated with improved survival by multivariate analysis. CONCLUSIONS: Surgical resection of hilar cholangiocarcinoma that requires resection of the portal vein can be performed safely and should not be a contraindication to resection. (J Am Coll Surg 2011;212: 604–616. © 2011 by the American College of Surgeons) BACKGROUND:
Cholangiocarcinoma is a relatively rare tumor, with an estimated incidence of 3,000 cases per year in the United States.1 A review of close to 300 patients with cholangiocarcinoma demonstrated that two-thirds of patients had hilar tumors, 27% of tumors occurred in the distal bile duct, and 6% were intrahepatic.2 An estimate of potentially
resectable patients who present with hilar cholangiocarcinoma (HCCA) was reported in a series by Jarnagin and colleagues;3 approximately 30% of patients with HCCA seen at their multidisciplinary hepatobiliary cancer center presented with unresectable disease, and 70% went on to some attempt at surgical management. Of those patients managed surgically, only 50% had an attempted curative resection. Initial reports of resection of HCCA consisted mainly of resections of the biliary tree and bilioenteric anastomosis to the intrahepatic ducts.4,5 Resection can be complicated by anatomic issues that make resection with negative margins for cure problematic with bile duct resection alone. Cholangiocarcinoma frequently extends along the hepatic ducts and along the hilar lymphatics and vessels, making the addition of liver resection essentially mandatory for cure in this disease. Blumgart and Launois, pioneers of modern hepatobiliary surgery, were early proponents of the addi-
Disclosure Information: Nothing to disclose. Presented at Southern Surgical Association 122nd Annual Meeting, Palm Beach, FL, December 2010. Received December 10, 2010; Accepted December 15, 2010. From the Department of Surgery, Center for Hepatobiliary Disease and Abdominal Transplantation, University of California San Diego, San Diego, CA (Hemming, Mekeel, Khanna, Baquerizo) and the Department of Surgery, University of Florida, Gainesville, FL (Kim). Correspondence address: Alan W Hemming, MD, MSc, FACS, Center for Hepatobiliary Diseases, University of California San Diego, Suite 2-280, 200 West Arbor Dr, #8401, San Diego, CA 92013-8401. email: ahemming@ ucsd.edu
© 2011 by the American College of Surgeons Published by Elsevier Inc.
604
ISSN 1072-7515/11/$36.00 doi:10.1016/j.jamcollsurg.2010.12.028
Vol. 212, No. 4, April 2011
Hemming et al
Abbreviations and Acronyms
FLR ⫽ future liver remnant HCCA ⫽ hilar cholangiocarcinoma PVE ⫽ portal vein embolization
tion of liver resection to bile duct resection for this disease6,7 to manage those tumors with direct hepatic invasion as well as to obtain oncologic margins on tumors that frequently extend longitudinally out the hepatic ducts.3,8-10 Initially, the concept was greeted with little enthusiasm by surgeons who, at the time, had a relatively nihilistic view of any surgical treatment for HCCA. Recent advances in perioperative management, such as preoperative portal vein embolization (PVE) and, although controversial, preoperative biliary drainage of the planned remnant liver, have to some degree ameliorated many of the concerns about major liver resection in the setting of biliary obstruction. The involvement of major hilar vascular structures (ie, portal vein and hepatic artery) by tumor makes resection with negative margins possible only with concomitant resection of the liver and biliary tree along with vascular resection and reconstruction. The extent of disease suggested by this involvement, coupled with the additional risk of vascular resection in this setting, leads to questions about the use of the procedure. This study examines the results of surgical management of HCCA during a time when liver resection in addition to bile duct resection was considered the standard management for curative resection of HCCA. The role of portal vein resection for HCCA is assessed.
METHODS Patients
Ninety-five consecutive patients with HCCA undergoing surgical resection with curative intent between July 1999 and July 2010 at the University of Florida and University of California San Diego by the hepatobiliary surgery/ transplant service were retrospectively reviewed with IRB approval. Patients were classified as HCCA if there was an adenocarcinoma originating from the biliary confluence or the left or right hepatic duct. Tumors that arose from the proximal hepatic duct were included only if they extended to involve the right or left hepatic duct. All patients had HCCA confirmed by pathology after resection. Fifty-nine patients were male (62%; male-to-female ratio of 1.6:1). Mean patient age was 64 ⫾ 9 years (range 24 to 85 years). Preoperative assessment
The standard preoperative workup consisted of triphasic CT to assess biliary, portal venous, and hepatic arterial
Portal Vein Resection in Hilar Cholangiocarcinoma
605
Table 1. Patients as Classified by the Bismuth-Corlette Staging System5 and the Staging System of Jarnagin and Blumgart3 Stage
Bismuth-Corlette 2 3a 3b 4 Blumgart T stage T1 T2 T3
Patients, n
10 57 18 10 59 36 0
involvement. Ideally, imaging was performed before any interventions on the biliary tree. Stents had been placed before referral to our center in 84 of 95 patients. Chest and abdominal CTs were also performed to assess for the presence of extrahepatic disease. After an assessment of the CT scan, resectability was determined. Imaging with magnetic resonance cholangiopancreatography was also obtained in 64 patients, either as part of the outside workup or to more precisely define the degree of biliary tree involvement. Before 2005, if patients were already stented but the biliary tree on the side of the liver to be left in was not drained, then percutaneous transhepatic stents were placed to decompress the biliary tree of the future liver remnant (FLR), regardless of the patient’s bilirubin level. Patients who presented with bilirubin ⬎5.0 mg/dL and had not been stented previously underwent percutaneous transhepatic stenting of the biliary tree of the FLR. Only 1 patient presented with a bilirubin level of ⬍5 mg/dL and had not been stented before referral and this patient subsequently underwent left trisegmentectomy without preoperative biliary intervention. After publication of our initial series in 2005,11 we adjusted our preoperative management such that all patients who did not show hypertrophy of the FLR because of contralateral tumor involvement undergo PVE of the side to be resected, and biliary drainage of the FLR was performed regardless of preoperative bilirubin level. Patients were staged preoperatively (Table 1) according to both the Bismuth-Corlette staging system5 and the T staging system of Jarnagin and Blumgart.3 In 38 cases where there was no pre-existing lobar hypertrophy and it could be definitively determined which side of the liver was to be resected, preoperative PVE was performed on the side of the liver that was to be resected 4 to 6 weeks before surgery.12 Thirty-six patients had embolization of the right portal vein for subsequent right trisectionectomy and 2 patients had embolization of the left portal vein for subsequent left trisectionectomy.
606
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
J Am Coll Surg
Surgical technique and procedures
An initial exploration to detect disseminated intraabdominal disease was performed. Laparoscopy was used selectively in patients with bulky disease or obvious nodal involvement on preoperative imaging. Patients with clearly disseminated disease had no additional surgical intervention and were managed with stents, chemotherapy, and/or radiotherapy, depending on patient-specific factors. Patients without disseminated disease underwent a standardized assessment of resectability. Intraoperative ultrasound was used in all cases. If the tumor was predominantly rightsided, then dissection of the left hepatic duct and left portal vein at the base of the falciform ligament was performed. The presence of an uninvolved hepatic duct at the segment 2 to 3 junction along with a patent left portal vein would allow us to proceed with resection, although this did not preclude the necessity of reconstructing the left portal vein due to more proximal involvement by the tumor. If the tumor was predominantly left-sided and a left trisectionectomy was contemplated, then ultrasound examination of the right liver became more important. In most cases, it could be determined that the tumor clearly did not extend to the posterior division of the right hepatic duct by using a combination of dissection along the posterior aspect of the right portal pedicle, manual palpation, and intraoperative ultrasound. The division into segment 6 and 7 ducts occurs relatively intrahepatically, however, and without lowering the hilar plate a definite identification of tumor at the segment 6 to 7 takeoffs without dividing the liver is difficult. We did not lower the hilar plate because of concerns about broaching tumor planes and the segment 6 to 7 confluence was approached from within the hepatic parenchyma during the phase of liver transection later in the case. Patients with localized disease who were believed to be unresectable had either a segment 3 or segment 6 bypass performed. Only patients who underwent resection are included in the current study. Once a decision about resectability within the liver was made, dissection of the hepatic artery was performed, ensuring arterial supply to the planned liver remnant. The common bile duct was divided at the level of the pancreas and reflected superiorly. Lymph nodes of the celiac axis, common hepatic artery, and all lymphatic structures in the hepatoduodenal ligament were resected. Caudate lobectomy was performed routinely. Frozen-section analysis of margins was used to guide resection; additional resection was performed, if possible, when margins were initially positive. In suitable candidates who had initial positive margins of the distal bile duct or those with disease that clearly extended into the intrapancreatic portion of the bile duct, pancreaticoduodenectomy was also performed.
Figure 1. The bifurcation of the portal vein has been resected and the left portal vein brought down in end to end fashion. LH artery, left hepatic artery; LH duct, left hepatic duct; PV anastomosis, portal vein anastomosis.
Roux-en-Y biliary enteric reconstruction was performed using a 60-cm-long segment of jejunum. Patients received postoperative chemoradiotherapy if clear margins (R0 resection) were not obtained. Patients with positive lymph nodes also received adjuvant chemoradiotherapy. In the 95 patients who underwent resection, liver resections performed with biliary resection included 84 trisectionectomies (63 right, 21 left) and 11 lobectomies (8 left, 3 right). Thirteen patients also had simultaneous pancreaticoduodenectomy performed for extension of disease into the intrapancreatic portion of the bile duct. Forty-two patients required portal vein resection to achieve negative margins. Portal vein reconstructions were from main portal vein to left portal vein in 36 cases (Fig. 1), main portal vein to right portal vein in 3 cases, and from main portal vein to the posterior branch of the right portal vein in 3 cases. Five patients required arterial resection and reconstruction; 4 reconstructions of the right hepatic artery (main right hepatic artery to main right hepatic artery in 2 patients and main right hepatic artery to posterior branch right hepatic artery in 2 patients) and 1 of the main hepatic artery. Portal vein resection
Portal vein resection was performed in end-to-end fashion. No interposition grafts of either autologous or prosthetic material were required. The portal vein resection was performed at 2 different time points during the procedure, depending on the extent of mobility of the portal vein, extent of tumor, and accessibility in the patient. In the majority of patients undergoing right-sided trisectionectomies in which access was reasonable and the amount of portal vein expected to be resected relatively short, the portal vein resection and reconstruction were performed before hepatic transection. After dissection of the left hepatic
Vol. 212, No. 4, April 2011
Hemming et al
Figure 2. The transverse portion of the left portal vein is dissected such that vascular clamps can be applied proximally and distally to the bifurcation and tumor (solid white lines). The tumor is left en bloc with the bifurcation of the portal vein. CBD, distal common bile duct reflected superiorly; LHA, left hepatic artery; LPV, left portal vein; PV, portal vein.
artery to its segmental branches and ligation of the right hepatic artery, the main portal vein below the tumor was isolated. The left portal vein above the tumor was dissected out past the transverse portion of the vein up to the ascending segment just before branching into main segmental branches. Multiple caudate branches require division to free up enough left portal vein to work with. The area of the main portal venous bifurcation was left en bloc with the tumor. Vascular clamps were placed on the main portal vein and left portal vein just before its branching to main segmental branches (Fig. 2). The vein was then divided and left attached to the tumor. The left portal vein was then brought down to the main portal vein with a primary endto-end anastomosis (Fig. 1). Arterial perfusion was maintained throughout the resection and reconstruction. The hepatic transection and specimen removal occurred after reconstruction with maintenance of both portal and hepatic arterial flow. In some patients, when access was difficult or the tumor extensive, the portal vein would be dissected as much as possible before hepatic transaction, but not divided until hepatic parenchymal transection was complete. Dividing the portal vein at this later stage removes the specimen and allows tremendous improvement in mobility of the hepatic side of the portal vein, which can be rotated down considerably from its original position to minimize tension on the venous anastomosis. Left-sided resections with anastomosis of the main portal vein to either right portal vein branch or posterior sectoral portal vein branch were all done after hepatic parenchymal transection. Parametric statistical analysis was performed using Student’s t-tests and nonparametric analysis of data was per-
Portal Vein Resection in Hilar Cholangiocarcinoma
607
formed using Kendall’s tau, chi-square, or Fisher’s exact test when appropriate. Survival was calculated using the Kaplan-Meier method, with differences in survival assessed using the log-rank test. Cox regression analysis was used on variables that were suggested by univariate analysis to be significant. Statistical analysis was performed using SPSS 10.0 software (SPSS Inc). Operative mortality was defined as death within 30 days or within the same hospital admission. Death within 90 days was also assessed as a separate variable. Operative blood use was defined as intraoperative transfusion or blood received within 48 hours of surgery. Liver failure was defined as postoperative bilirubin ⬎7.0 mg/dL or an international normalized ratio ⬎2.0 requiring fresh-frozen plasma longer than 48 hours after resection.13 Results are reported as mean ⫾ 1 SD unless otherwise specified.
RESULTS Median blood loss in patients undergoing resection was 675 ⫾ 329 mL (range 100 to 1,800 mL). Thirty percent of patients received no blood, but the median number of transfused units was 2 (range 0 to 8 U). Hospital stay was 13 ⫾ 5 days (range 6 to 32 days). Complications occurred in 32 of the 95 patients (36%) who underwent resection. Some patients had more than one complication. There were 9 bile leaks managed conservatively, 5 cases of pleural effusions large enough to require drainage, and 9 intra-abdominal abscesses that required drainage. One patient undergoing hepatopancreaticoduodenectomy developed a controlled pancreatic fistula that resolved with conservative management. Wound infections occurred in 14 patients. Two patients developed renal failure requiring temporary dialysis and 3 patients developed pneumonia. Liver failure not related to an infectious complication occurred in 3 patients, 2 of whom did not have the FLR’s biliary tree drained preoperatively. One of these patients had a preoperatively placed biliary drain and underwent a left trisectionectomy, however, the biliary drain had been placed in the right anterior sector, which did not communicate with the posterior sectoral ducts. The second was a patient who had a right trisectionectomy without preoperative drainage of the left biliary tree. The third patient, who had a right trisectionectomy, had the FLR drained preoperatively and underwent PVE but transient liver failure developed postoperatively and resolved over time. There were 5 perioperative deaths (5% operative mortality). Two patients died from liver failure and 3 patients had an initial infectious complication (pneumonia, n ⫽ 2; subphrenic abscess, n ⫽ 1) leading to liver failure. One additional patient died after discharge from hospital and outside of the 30-day window but within 90 days of
608
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
J Am Coll Surg
Figure 3. Overall survival of 95 patients undergoing resection for hilar cholangiocarcinoma. Overall 5-year survival rate was 43%.
surgery from what appeared to be a myocardial event, although this was not confirmed by autopsy. Eighty of 95 patients resected (84%) had histologically negative margins (R0). The remaining 15 patients had microscopically positive margins. Twelve of these specimens were initially called negative by frozen-section analysis, but were found to be positive on permanent section. In 3 cases, it was known that the margins were microscopically positive at surgery; however, additional extension of the biliary resection margin was not technically possible. In 2 left trisectionectomies, the segment 6 and segment 7 ducts were already widely separate, and in 1 right trisectionectomy, resection had already been performed back to separate segment 2 and 3 ducts. Papillary tumors occurred in 5% of resected patients. Twenty-five percent of patients had well-differentiated tumors, with the remainder either moderately or poorly differentiated. Mean tumor size was 3.4 ⫾ 1.9 cm. Thirtyeight percent of patients had positive nodes in the hepatoduodenal ligament. Of the 42 patients who underwent portal vein resection, 17 (41%) had true invasion of the portal vein wall by tumor. The remaining 25 patients had dense fibrosis surrounding the portal vein, within which could be found tumor but no true vascular invasion. The distribution of patients by preoperative staging system is demonstrated in Table 1. An assessment of the 2 preoperative staging systems used demonstrated no correlation between the Bismuth-Corlette staging system5 or the staging system proposed by Jarnagin and Blumgart,3 and
survival. However, when patients who had lobar hypertrophy preoperatively because of tumor involvement of the contralateral liver or induced with PVE were grouped together and compared with those patients who did not have lobar hypertrophy, they had a significantly lower operative mortality than those patients without lobar hypertrophy (0% versus 9%; p ⫽ 0.02 ). Median actuarial survival for all patients was 38 months, with median follow-up time of 32 months. Five-year actuarial survival rate was 43% for all patients undergoing resection (Fig. 3) and 50% for those patients who were resected with negative margins. Patients who had an R0 resection survived longer than patients resected with microscopically positive margins (median survival 58 months; 95% CI, 36–81 months versus 26 months 95% CI, 14–31 months; p ⬍ 0.01; Fig. 4). There were no long-term survivors who had positive resection margins. Perhaps not surprisingly, patients resected with positive margins were more likely to have positive lymph nodes than patients with negative margins (87% versus 28%; p ⬍ 0.01). There was no difference in either operative mortality (2% versus 8%; p ⫽ 0.23) or long-term survival (Fig. 5) between patients who had portal vein resections and those who did not. Factors not associated with an effect on survival by univariate analysis in patients who were resected included age older than 70 years, sex, preoperative Bismuth-Corlette and Jarnagin-Blumgart stage, tumor size ⬎2.5 cm, type of resection, and postoperative American Joint Committee on
Vol. 212, No. 4, April 2011
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
609
Figure 4. Patients resected with negative margins had a 5-year survival rate of 50%, which was significantly better than those patients resected with positive margins, with no patient with positive margins surviving longer than 40 months (p ⬍ 0.05 log rank).
Cancer stage. Patients with lymph node involvement by tumor appeared to have a worse 5-year survival rates than patients with node-negative status; however, this difference was not statistically significant (23% versus 47%; p ⫽ 0.06; Fig. 6). Patients with well-differentiated tumors showed a trend toward improved survival compared with patients with moderately or poorly differentiated tumors, however, this trend was not significant (p ⫽ 0.10). Only R0 status demonstrated an effect on survival at a level of significance of ⬍0.05 by univariate analysis. A Cox regression analysis was performed on variables with p ⱕ 0.10, including R0 status, lymph node status, and tumor differentiation. This confirmed that only R0 status remained a substantial predictor of survival. Although there was no difference in overall survival between the first half of the study period and the second, there was a difference in 30-day mortality, with 10% operative mortality in the first half of the study and no operative mortality in the second half of the study (p ⫽ 0.04), when PVE and preoperative biliary drainage of the FLR were used as a standard.
DISCUSSION HCCA continues to be a difficult challenge for the surgeon. Achieving negative surgical margins when resecting this relatively uncommon tumor is technically demanding
because of the close proximity of the bile duct bifurcation to the vascular inflow of the liver. The addition of liver resection to the procedure has become the standard of care. The addition of liver resection increases the number of patients who can be offered potentially margin-free resection, and increases the oncologic clearance of the procedure in patients who would previously have been treated with bile duct resection only. Extended hepatectomy in the setting of biliary obstruction and elevated bilirubin can be associated with increased risk, however.14-16 Attempts to reduce the risks associated with biliary obstruction by preoperative biliary drainage remain controversial, with arguments for and against debated vigorously.17,18 There is little argument that the presence of preoperative biliary drainage increases the rate of bacterial colonization of the biliary tree and can increase the rate of perioperative infectious complications.3,19 Conversely, several recent series of hepatic resections for HCCA demonstrate a reduced incidence of liver failure when preoperative biliary drainage was used, although PVE was also used in those series.20-23 In practice, biliary drainage has occurred before referral in the majority of cases seen by the hepatobiliary surgeon. In our series, 84 of 95 patients already had biliary stents placed before surgical referral. In many cases, stents were placed without consideration of subsequent surgery and were placed into the side of the liver that would be removed rather than the side of the planned remnant liver after resection. Kennedy
610
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
J Am Coll Surg
Figure 5. There was no significant difference in survival between those patients who had portal vein resection and those that did not (p ⫽ 0.6 log rank).
and colleagues,24 reporting on the Memorial experience, demonstrated that biliary drainage of the FLR when it was ⬍30% of liver volume was beneficial in their series. In that article, there was no correlation between preoperative bilirubin levels and outcomes. This finding can be explained in part by the fact that in many cases in that series, stents were placed in the biliary tree ipsilateral to the tumor, allowing improved function, decreased bilirubin, and recovery of the side of the liver that was to be removed rather than the side that was to remain. Ideally, patients should undergo early referral, with the need for, and planning of, preoperative biliary drainage done in concert with the surgeon. The importance of the roles of remnant liver function and liver volume and the concept of lobar atrophy and subsequent contralateral lobar hypertrophy and its effect on operative mortality were again demonstrated in this study. Patients who had lobar hypertrophy induced by either contralateral involvement by tumor (JarnaginBlumgart T2) or by preoperative PVEs were less likely to die in the perioperative period. Only 2 of the 5 deaths in this series were from isolated liver failure without initial infectious complication. However, the 3 other patients who died had infection initially that led to development of liver failure as a component of their demise. Seyama and colleagues21 used preoperative PVE in a series of 58 pa-
tients; all patients undergoing resection of ⬎50% of initial liver volume for HCCA and reported a 0% operative mortality, although there was a 43% morbidity, which is similar to our series. Although speculative, it might be that the improved remnant liver function provided by preoperative lobar hypertrophy not only prevents primary liver failure but also provides a degree of hepatic functional reserve that allows patients to better tolerate a subsequent insult, such as an infectious complication. Complication rates for extended hepatectomy for HCCA range from 40% to 52% in recent series.3,20,22,25-27 There was a change in practice during the duration of the study after our initial report in 2005.11 All patients in the second half of the study received preoperative biliary drainage of the FLR with PVE performed if no evidence of lobar hypertrophy was present. Although this was clearly not a randomized prospective trial, there was a decreased operative mortality noted in the second half of the study. We would like to attribute the decrease in mortality to the initiation of standardized biliary drainage of the FLR and use of PVE; however, it might be that improved selection of patients or greater experience over time led to some of the improvement. Of note, the improvement in operative mortality did not translate into improved long-term survival. In any event, it would appear reasonable to use preoperative PVE in all patients under-
Vol. 212, No. 4, April 2011
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
611
Figure 6. Patients with lymph node involvement with tumor appeared to have a worse survival rate at 5 years than those patients with negative lymph nodes; however, this difference was not statistically significant (23% versus 47%; p ⫽ 0.06).
going planned extended hepatectomy for HCCA who do not have pre-existing lobar hypertrophy of the remnant liver from contralateral tumor involvement. Only the presence of an R0 resection proved to be predictive of an improvement in survival. Patients resected with negative margins had a 50% 5-year actuarial survival rate, and no patients with positive margins were alive past 48 months. These findings are similar to the larger Japanese experience reported from Nagoya that demonstrated a 52% 5-year survival rate in patients undergoing an R0 resection and in which there was liberal use of portal vein resection along with hepatic arterial resection performed.27 Patients in the current study who had lymph node involvement showed a trend toward having worse survival that did not reach statistical significance (p ⫽ 0.06). The power of this study was not sufficient to determine whether there was a true difference in survival associated with lymph node involvement, and the possibility of a type II error is relatively high. The literature provides conflicting results about the association of lymph node status on survival, with some authors showing a clear effect21,27 and some showing none.3,25 The most recent studies with larger numbers of patients in the series would indicate that lymph node involvement is independently associated with poor prognosis.22,28 Positive lymph node involvement suggests a
more advanced, biologically aggressive tumor that, intuitively, should have worse outcomes. More aggressive tumor biology might also explain the association between patients with positive nodes and positive resection margins noted in the current study, despite fairly aggressive resections with extended hepatectomy and vascular resection. Although suggesting a poorer prognosis, it is also clear that positive lymph nodes in the porta hepatis do not preclude a curative procedure if negative margins can be obtained. Although liver resection aids in the management of longitudinal extension of tumor out the biliary tree and also with unilateral involvement of vascular structures with tumor, involvement of the portal vein bifurcation remains difficult to manage. Some of the early reports of liver resection for HCCA recognized this difficulty and suggested portal vein resection with subsequent reconstruction in an attempt to obtain clear margins.29 The degree of difficulty and assumed risk of increased mortality associated with portal vein resection led to little enthusiasm for its adoption, however. Recent development of specialized hepatobiliary surgical groups, along with the use of techniques developed for liver transplantation have renewed interest in portal vein resection for HCCA. The operative mortality in our series, if anything, appeared lower in the group undergoing portal vein resection as compared with the group that
612
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
did not undergo portal vein resection (2% versus 8%), although this was not a statistically significant difference (p ⫽ 0.37). In previous reports, we have suggested that the relatively low mortality observed with portal vein resection might be secondary to compensatory hypertrophy of the remnant liver and improved postoperative liver function that occurs in portal vein resection patients due to tumor involvement of the contralateral portal vein.11,30 Other groups have also reported no difference in operative mortality in patients undergoing portal vein resection versus standard resection for HCCA.9,26,31 The impact of portal vein resection on long-term survival is less clear. Ebata and colleagues31 reported worse long-term survival in patients requiring portal vein resection; however, multivariate analysis demonstrated that portal vein resection itself did not worsen survival, but it was the presence of transluminal tumor or positive margins that had a negative impact. Lee and colleagues,22 in a series of 302 patients with HCCA, demonstrated no difference in long-term survival in the 38 patients who underwent portal vein resection versus the 264 patients who did not require portal vein resection. The Nagoya group performed portal vein resection in more than one-third of 298 patients undergoing resection for HCCA and found an unexpectedly better survival in those patients undergoing vascular resection.27 Neuhaus and colleagues9,32 have taken portal vein resection a step further. In a series of extended right hepatectomies combined with regional lymphadenectomy and routine portal vein resection and reconstruction, they have reported a 72% 5-year survival rate, which is by far the best result reported to date for resection of HCCA. The portal vein is resected in this technique in all patients to allow a “no-touch” resection of the tumor and adjacent tissue, and the authors suggest that it is the adherence to such oncologic principles that yields favorable results. In the last several years, we have adopted many of the no-touch principles espoused by Neuhaus and colleagues32 and, when performing extended right hepatectomies, will have a very low threshold to resect the portal vein bifurcation. Resecting the portal vein bifurcation when performing a left trisectionectomy is substantially more difficult because of the relatively short course of the right portal vein before branching and was not commonly performed in the current series. We could show neither a survival advantage nor disadvantage associated with en bloc resection of the portal vein. The results suggest that although portal vein resection should not necessarily be a standard part of every resection, the need for portal vein resection should not be a contraindication to resection for cure.
J Am Coll Surg
CONCLUSIONS In summary, the current study confirms that an aggressive surgical approach to HCCA that includes hepatic resection as standard therapy is warranted. The operative mortality was 5%, which is in line with the mortality for extended hepatectomies performed for other indications at our institution. A 50% 5-year survival rate can be expected if negative margins can be obtained, even if this requires resection and reconstruction of the hepatic vasculature. Obtaining negative resection margins is the single most important factor in achieving prolonged survival. The use of strategies to optimize the FLR before resection can allow more aggressive resections in pursuit of negative margins and reduce operative mortality. Author Contributions Study conception and design: Hemming, Mekeel, Khanna, Baquerizo, Kim Acquisition of data: Hemming, Baquerizo, Kim Analysis and interpretation of data: Hemming, Mekeel, Kim Drafting of manuscript: Hemming, Mekeel, Khanna, Baquerizo, Kim Critical revision: Hemming, Mekeel, Khanna, Baquerizo, Kim
REFERENCES 1. Olnes MJ, Erlich R. A review and update on cholangiocarcinoma. Oncology 2004;66:167–179. 2. Nakeeb A, Pitt HA, Sohn TA, et al. Cholangiocarcinoma. A spectrum of intrahepatic, perihilar, and distal tumors. Ann Surg 1996;224:463–473; discussion 473–475. 3. Jarnagin WR, Fong Y, DeMatteo RP, et al. Staging, resectability, and outcome in 225 patients with hilar cholangiocarcinoma. Ann Surg 2001;234:507–517; discussion 517–519. 4. Tompkins RK, Thomas D, Wile A, Longmire WP Jr. Prognostic factors in bile duct carcinoma: analysis of 96 cases. Ann Surg 1981;194:447–457. 5. Bismuth H, Corlette MB. Intrahepatic cholangioenteric anastomosis in carcinoma of the hilus of the liver. Surg Gynecol Obstet 1975;140:170–178. 6. Launois B, Campion JP, Brissot P, Gosselin M. Carcinoma of the hepatic hilus. Surgical management and the case for resection. Ann Surg 1979;190:151–157. 7. Beazley RM, Hadjis N, Benjamin IS, Blumgart LH. Clinicopathological aspects of high bile duct cancer. Experience with resection and bypass surgical treatments. Ann Surg 1984;199: 623–636. 8. Kosuge T, Yamamoto J, Shimada K, et al. Improved surgical results for hilar cholangiocarcinoma with procedures including major hepatic resection. Ann Surg 1999;230:663–671. 9. Neuhaus P, Jonas S, Bechstein WO, et al. Extended resections for hilar cholangiocarcinoma. Ann Surg 1999;230:808–818; discussion 819. 10. Launois B, Reding R, Lebeau G, Buard JL. Surgery for hilar
Vol. 212, No. 4, April 2011
11. 12. 13. 14. 15. 16. 17. 18. 19.
20.
21. 22. 23. 24. 25. 26.
27. 28. 29. 30.
cholangiocarcinoma: French experience in a collective survey of 552 extrahepatic bile duct cancers. J Hepatobiliary Pancreat Surg 2000;7:128–134. Hemming AW, Reed AI, Fujita S, et al. Surgical management of hilar cholangiocarcinoma. Ann Surg 2005;241:693–699; discussion 699–702. Hemming AW, Reed AI, Howard RJ, et al. Preoperative portal vein embolization for extended hepatectomy. Ann Surg 2003; 237:686–691; discussion 691–693. Mullen JT, Ribero D, Reddy SK, et al. Hepatic insufficiency and mortality in 1,059 noncirrhotic patients undergoing major hepatectomy. J Am Coll Surg 2007;204:854–862; discussion 862–864. Su CH, Tsay SH, Wu CC, et al. Factors influencing postoperative morbidity, mortality, and survival after resection for hilar cholangiocarcinoma. Ann Surg 1996;223:384–394. Melendez J, Ferri E, Zwillman M, et al. Extended hepatic resection: a 6-year retrospective study of risk factors for perioperative mortality. J Am Coll Surg 2001;192:47–53. Imamura H, Seyama Y, Kokudo N, et al. One thousand fifty-six hepatectomies without mortality in 8 years. Arch Surg 2003; 138:1198–1206; discussion 1206. Laurent A, Tayar C, Cherqui D. Cholangiocarcinoma: preoperative biliary drainage (Con). HPB (Oxford) 2008;10:126–129. Nimura Y. Preoperative biliary drainage before resection for cholangiocarcinoma (Pro). HPB (Oxford) 2008;10:130–133. Hochwald SN, Burke EC, Jarnagin WR, et al. Association of preoperative biliary stenting with increased postoperative infectious complications in proximal cholangiocarcinoma. Arch Surg 1999;134:261–266. Kawasaki S, Imamura H, Kobayashi A, et al. Results of surgical resection for patients with hilar bile duct cancer: application of extended hepatectomy after biliary drainage and hemihepatic portal vein embolization. Ann Surg 2003;238:84–92. Seyama Y, Kubota K, Sano K, et al. Long-term outcome of extended hemihepatectomy for hilar bile duct cancer with no mortality and high survival rate. Ann Surg 2003;238:73–83. Lee SG, Song GW, Hwang S, et al. Surgical treatment of hilar cholangiocarcinoma in the new era: the Asian experience. J Hepatobiliary Pancreat Sci 2010;17:476–489. Maguchi H, Takahashi K, Katanuma A, et al. Preoperative biliary drainage for hilar cholangiocarcinoma. J Hepatobiliary Pancreat Surg 2007;14:441–446. Kennedy TJ, Yopp A, Qin Y, et al. Role of preoperative biliary drainage of liver remnant prior to extended liver resection for hilar cholangiocarcinoma. HPB (Oxford) 2009;11:445–451. Rea DJ, Munoz-Juarez M, Farnell MB, et al. Major hepatic resection for hilar cholangiocarcinoma: analysis of 46 patients. Arch Surg 2004;139:514–523; discussion 523–525. Shimizu H, Kimura F, Yoshidome H, et al. Aggressive surgical resection for hilar cholangiocarcinoma of the left-side predominance: radicality and safety of left-sided hepatectomy. Ann Surg 2010;251:281–286. Igami T, Nishio H, Ebata T, et al. Surgical treatment of hilar cholangiocarcinoma in the “new era”: the Nagoya University experience. J Hepatobiliary Pancreat Sci 2010;17:449–454. Ito K, Ito H, Allen PJ, et al. Adequate lymph node assessment for extrahepatic bile duct adenocarcinoma. Ann Surg 2010;251: 675–681. Hadjis NS, Blenkharn JI, Alexander N, et al. Outcome of radical surgery in hilar cholangiocarcinoma. Surgery 1990;107:597– 604. Hemming AW, Kim RD, Mekeel KL, et al. Portal vein resection
Hemming et al
Discussion
613
for hilar cholangiocarcinoma. Am Surg 2006;72:599–604; discussion 605. 31. Ebata T, Nagino M, Kamiya J, et al. Hepatectomy with portal vein resection for hilar cholangiocarcinoma: audit of 52 consecutive cases. Ann Surg 2003;238:720–727. 32. Neuhaus P, Jonas S, Settmacher U, et al. Surgical management of proximal bile duct cancer: extended right lobe resection increases resectability and radicality. Langenbecks Arch Surg 2003;388:194–200.
Discussion DR WILLIAM JARNAGIN (New York, NY): This is a very good paper, illustrating the results that are possible in hilar cholangiocarcinoma from a technically proficient surgeon with a good understanding of the disease, working with an experienced team. Ninetyfive resections with an 84% R0 rate over approximately 10 years may seem like a small number, but anyone with any familiarity with this problem realizes that this is, in fact, a very large experience, and it doesn’t take into account those that were unresectable, which is probably around twice that number. Furthermore, anyone who has performed this operation with any frequency understands how technically difficult it can be and also just how good these results are with respect to morbidity and mortality. I have a few questions. I want to emphasize the important point that you made, which is that in the face of hypertrophy of the future liver remnant (FLR), due to contralateral atrophy caused by tumor, portal vein embolization is unnecessary, and in fact, preoperative biliary drainage probably adds little to improving perioperative morbidity, and indeed in this situation, placing biliary catheters likely exposes patients to the risks of the procedure with little benefit. You state it is your current practice not to drain the future liver remnant if there is hypertrophy, but you don’t specify the level required. In our experience, an FLR greater than 30% seemed to be the cut-off value. But I wonder what level you used? Also, you showed that patients with hypertrophy of the FLR noted preoperatively, whether due to tumor involvement or from an embolization, had 0% operative mortality, compared with 9% for those without. Was this independent of preoperative biliary drainage? I suspect that most of the latter group had undergone preoperative biliary decompression, which underscores the point that it is the quality of the future liver remnant that is ultimately the most important factor, and that biliary drainage in and of itself is not necessarily as protective as many would like to believe. You talked about the Blumgart preoperative staging system, which is not really useful for predicting survival after resection, but is really for predicting the likelihood that a resection is possible and the likelihood of finding metastatic disease. As such, it really requires inclusion of the entire denominator of patients, including those who were unresectable. Over the period of study, how many patients had unresectable disease at presentation and how many did you explore without resecting? Finally, you show that operative mortality was no different in the portal vein resection versus no resection groups. What about the morbidity? I suspect it was higher in the group with vascular resec-
Cholangiocarcinoma is a relatively rare tumor, with an estimated incidence of 3,000 cases per year in the United States.1 A review of close to 300 patients with cholangiocarcinoma demonstrated that two-thirds of patients had hilar tumors, 27% of tumors occurred in the distal bile duct, and 6% were intrahepatic.2 An estimate of potentially
resectable patients who present with hilar cholangiocarcinoma (HCCA) was reported in a series by Jarnagin and colleagues;3 approximately 30% of patients with HCCA seen at their multidisciplinary hepatobiliary cancer center presented with unresectable disease, and 70% went on to some attempt at surgical management. Of those patients managed surgically, only 50% had an attempted curative resection. Initial reports of resection of HCCA consisted mainly of resections of the biliary tree and bilioenteric anastomosis to the intrahepatic ducts.4,5 Resection can be complicated by anatomic issues that make resection with negative margins for cure problematic with bile duct resection alone. Cholangiocarcinoma frequently extends along the hepatic ducts and along the hilar lymphatics and vessels, making the addition of liver resection essentially mandatory for cure in this disease. Blumgart and Launois, pioneers of modern hepatobiliary surgery, were early proponents of the addi-
Disclosure Information: Nothing to disclose. Presented at Southern Surgical Association 122nd Annual Meeting, Palm Beach, FL, December 2010. Received December 10, 2010; Accepted December 15, 2010. From the Department of Surgery, Center for Hepatobiliary Disease and Abdominal Transplantation, University of California San Diego, San Diego, CA (Hemming, Mekeel, Khanna, Baquerizo) and the Department of Surgery, University of Florida, Gainesville, FL (Kim). Correspondence address: Alan W Hemming, MD, MSc, FACS, Center for Hepatobiliary Diseases, University of California San Diego, Suite 2-280, 200 West Arbor Dr, #8401, San Diego, CA 92013-8401. email: ahemming@ ucsd.edu
© 2011 by the American College of Surgeons Published by Elsevier Inc.
604
ISSN 1072-7515/11/$36.00 doi:10.1016/j.jamcollsurg.2010.12.028
Vol. 212, No. 4, April 2011
Hemming et al
Abbreviations and Acronyms
FLR ⫽ future liver remnant HCCA ⫽ hilar cholangiocarcinoma PVE ⫽ portal vein embolization
tion of liver resection to bile duct resection for this disease6,7 to manage those tumors with direct hepatic invasion as well as to obtain oncologic margins on tumors that frequently extend longitudinally out the hepatic ducts.3,8-10 Initially, the concept was greeted with little enthusiasm by surgeons who, at the time, had a relatively nihilistic view of any surgical treatment for HCCA. Recent advances in perioperative management, such as preoperative portal vein embolization (PVE) and, although controversial, preoperative biliary drainage of the planned remnant liver, have to some degree ameliorated many of the concerns about major liver resection in the setting of biliary obstruction. The involvement of major hilar vascular structures (ie, portal vein and hepatic artery) by tumor makes resection with negative margins possible only with concomitant resection of the liver and biliary tree along with vascular resection and reconstruction. The extent of disease suggested by this involvement, coupled with the additional risk of vascular resection in this setting, leads to questions about the use of the procedure. This study examines the results of surgical management of HCCA during a time when liver resection in addition to bile duct resection was considered the standard management for curative resection of HCCA. The role of portal vein resection for HCCA is assessed.
METHODS Patients
Ninety-five consecutive patients with HCCA undergoing surgical resection with curative intent between July 1999 and July 2010 at the University of Florida and University of California San Diego by the hepatobiliary surgery/ transplant service were retrospectively reviewed with IRB approval. Patients were classified as HCCA if there was an adenocarcinoma originating from the biliary confluence or the left or right hepatic duct. Tumors that arose from the proximal hepatic duct were included only if they extended to involve the right or left hepatic duct. All patients had HCCA confirmed by pathology after resection. Fifty-nine patients were male (62%; male-to-female ratio of 1.6:1). Mean patient age was 64 ⫾ 9 years (range 24 to 85 years). Preoperative assessment
The standard preoperative workup consisted of triphasic CT to assess biliary, portal venous, and hepatic arterial
Portal Vein Resection in Hilar Cholangiocarcinoma
605
Table 1. Patients as Classified by the Bismuth-Corlette Staging System5 and the Staging System of Jarnagin and Blumgart3 Stage
Bismuth-Corlette 2 3a 3b 4 Blumgart T stage T1 T2 T3
Patients, n
10 57 18 10 59 36 0
involvement. Ideally, imaging was performed before any interventions on the biliary tree. Stents had been placed before referral to our center in 84 of 95 patients. Chest and abdominal CTs were also performed to assess for the presence of extrahepatic disease. After an assessment of the CT scan, resectability was determined. Imaging with magnetic resonance cholangiopancreatography was also obtained in 64 patients, either as part of the outside workup or to more precisely define the degree of biliary tree involvement. Before 2005, if patients were already stented but the biliary tree on the side of the liver to be left in was not drained, then percutaneous transhepatic stents were placed to decompress the biliary tree of the future liver remnant (FLR), regardless of the patient’s bilirubin level. Patients who presented with bilirubin ⬎5.0 mg/dL and had not been stented previously underwent percutaneous transhepatic stenting of the biliary tree of the FLR. Only 1 patient presented with a bilirubin level of ⬍5 mg/dL and had not been stented before referral and this patient subsequently underwent left trisegmentectomy without preoperative biliary intervention. After publication of our initial series in 2005,11 we adjusted our preoperative management such that all patients who did not show hypertrophy of the FLR because of contralateral tumor involvement undergo PVE of the side to be resected, and biliary drainage of the FLR was performed regardless of preoperative bilirubin level. Patients were staged preoperatively (Table 1) according to both the Bismuth-Corlette staging system5 and the T staging system of Jarnagin and Blumgart.3 In 38 cases where there was no pre-existing lobar hypertrophy and it could be definitively determined which side of the liver was to be resected, preoperative PVE was performed on the side of the liver that was to be resected 4 to 6 weeks before surgery.12 Thirty-six patients had embolization of the right portal vein for subsequent right trisectionectomy and 2 patients had embolization of the left portal vein for subsequent left trisectionectomy.
606
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
J Am Coll Surg
Surgical technique and procedures
An initial exploration to detect disseminated intraabdominal disease was performed. Laparoscopy was used selectively in patients with bulky disease or obvious nodal involvement on preoperative imaging. Patients with clearly disseminated disease had no additional surgical intervention and were managed with stents, chemotherapy, and/or radiotherapy, depending on patient-specific factors. Patients without disseminated disease underwent a standardized assessment of resectability. Intraoperative ultrasound was used in all cases. If the tumor was predominantly rightsided, then dissection of the left hepatic duct and left portal vein at the base of the falciform ligament was performed. The presence of an uninvolved hepatic duct at the segment 2 to 3 junction along with a patent left portal vein would allow us to proceed with resection, although this did not preclude the necessity of reconstructing the left portal vein due to more proximal involvement by the tumor. If the tumor was predominantly left-sided and a left trisectionectomy was contemplated, then ultrasound examination of the right liver became more important. In most cases, it could be determined that the tumor clearly did not extend to the posterior division of the right hepatic duct by using a combination of dissection along the posterior aspect of the right portal pedicle, manual palpation, and intraoperative ultrasound. The division into segment 6 and 7 ducts occurs relatively intrahepatically, however, and without lowering the hilar plate a definite identification of tumor at the segment 6 to 7 takeoffs without dividing the liver is difficult. We did not lower the hilar plate because of concerns about broaching tumor planes and the segment 6 to 7 confluence was approached from within the hepatic parenchyma during the phase of liver transection later in the case. Patients with localized disease who were believed to be unresectable had either a segment 3 or segment 6 bypass performed. Only patients who underwent resection are included in the current study. Once a decision about resectability within the liver was made, dissection of the hepatic artery was performed, ensuring arterial supply to the planned liver remnant. The common bile duct was divided at the level of the pancreas and reflected superiorly. Lymph nodes of the celiac axis, common hepatic artery, and all lymphatic structures in the hepatoduodenal ligament were resected. Caudate lobectomy was performed routinely. Frozen-section analysis of margins was used to guide resection; additional resection was performed, if possible, when margins were initially positive. In suitable candidates who had initial positive margins of the distal bile duct or those with disease that clearly extended into the intrapancreatic portion of the bile duct, pancreaticoduodenectomy was also performed.
Figure 1. The bifurcation of the portal vein has been resected and the left portal vein brought down in end to end fashion. LH artery, left hepatic artery; LH duct, left hepatic duct; PV anastomosis, portal vein anastomosis.
Roux-en-Y biliary enteric reconstruction was performed using a 60-cm-long segment of jejunum. Patients received postoperative chemoradiotherapy if clear margins (R0 resection) were not obtained. Patients with positive lymph nodes also received adjuvant chemoradiotherapy. In the 95 patients who underwent resection, liver resections performed with biliary resection included 84 trisectionectomies (63 right, 21 left) and 11 lobectomies (8 left, 3 right). Thirteen patients also had simultaneous pancreaticoduodenectomy performed for extension of disease into the intrapancreatic portion of the bile duct. Forty-two patients required portal vein resection to achieve negative margins. Portal vein reconstructions were from main portal vein to left portal vein in 36 cases (Fig. 1), main portal vein to right portal vein in 3 cases, and from main portal vein to the posterior branch of the right portal vein in 3 cases. Five patients required arterial resection and reconstruction; 4 reconstructions of the right hepatic artery (main right hepatic artery to main right hepatic artery in 2 patients and main right hepatic artery to posterior branch right hepatic artery in 2 patients) and 1 of the main hepatic artery. Portal vein resection
Portal vein resection was performed in end-to-end fashion. No interposition grafts of either autologous or prosthetic material were required. The portal vein resection was performed at 2 different time points during the procedure, depending on the extent of mobility of the portal vein, extent of tumor, and accessibility in the patient. In the majority of patients undergoing right-sided trisectionectomies in which access was reasonable and the amount of portal vein expected to be resected relatively short, the portal vein resection and reconstruction were performed before hepatic transection. After dissection of the left hepatic
Vol. 212, No. 4, April 2011
Hemming et al
Figure 2. The transverse portion of the left portal vein is dissected such that vascular clamps can be applied proximally and distally to the bifurcation and tumor (solid white lines). The tumor is left en bloc with the bifurcation of the portal vein. CBD, distal common bile duct reflected superiorly; LHA, left hepatic artery; LPV, left portal vein; PV, portal vein.
artery to its segmental branches and ligation of the right hepatic artery, the main portal vein below the tumor was isolated. The left portal vein above the tumor was dissected out past the transverse portion of the vein up to the ascending segment just before branching into main segmental branches. Multiple caudate branches require division to free up enough left portal vein to work with. The area of the main portal venous bifurcation was left en bloc with the tumor. Vascular clamps were placed on the main portal vein and left portal vein just before its branching to main segmental branches (Fig. 2). The vein was then divided and left attached to the tumor. The left portal vein was then brought down to the main portal vein with a primary endto-end anastomosis (Fig. 1). Arterial perfusion was maintained throughout the resection and reconstruction. The hepatic transection and specimen removal occurred after reconstruction with maintenance of both portal and hepatic arterial flow. In some patients, when access was difficult or the tumor extensive, the portal vein would be dissected as much as possible before hepatic transaction, but not divided until hepatic parenchymal transection was complete. Dividing the portal vein at this later stage removes the specimen and allows tremendous improvement in mobility of the hepatic side of the portal vein, which can be rotated down considerably from its original position to minimize tension on the venous anastomosis. Left-sided resections with anastomosis of the main portal vein to either right portal vein branch or posterior sectoral portal vein branch were all done after hepatic parenchymal transection. Parametric statistical analysis was performed using Student’s t-tests and nonparametric analysis of data was per-
Portal Vein Resection in Hilar Cholangiocarcinoma
607
formed using Kendall’s tau, chi-square, or Fisher’s exact test when appropriate. Survival was calculated using the Kaplan-Meier method, with differences in survival assessed using the log-rank test. Cox regression analysis was used on variables that were suggested by univariate analysis to be significant. Statistical analysis was performed using SPSS 10.0 software (SPSS Inc). Operative mortality was defined as death within 30 days or within the same hospital admission. Death within 90 days was also assessed as a separate variable. Operative blood use was defined as intraoperative transfusion or blood received within 48 hours of surgery. Liver failure was defined as postoperative bilirubin ⬎7.0 mg/dL or an international normalized ratio ⬎2.0 requiring fresh-frozen plasma longer than 48 hours after resection.13 Results are reported as mean ⫾ 1 SD unless otherwise specified.
RESULTS Median blood loss in patients undergoing resection was 675 ⫾ 329 mL (range 100 to 1,800 mL). Thirty percent of patients received no blood, but the median number of transfused units was 2 (range 0 to 8 U). Hospital stay was 13 ⫾ 5 days (range 6 to 32 days). Complications occurred in 32 of the 95 patients (36%) who underwent resection. Some patients had more than one complication. There were 9 bile leaks managed conservatively, 5 cases of pleural effusions large enough to require drainage, and 9 intra-abdominal abscesses that required drainage. One patient undergoing hepatopancreaticoduodenectomy developed a controlled pancreatic fistula that resolved with conservative management. Wound infections occurred in 14 patients. Two patients developed renal failure requiring temporary dialysis and 3 patients developed pneumonia. Liver failure not related to an infectious complication occurred in 3 patients, 2 of whom did not have the FLR’s biliary tree drained preoperatively. One of these patients had a preoperatively placed biliary drain and underwent a left trisectionectomy, however, the biliary drain had been placed in the right anterior sector, which did not communicate with the posterior sectoral ducts. The second was a patient who had a right trisectionectomy without preoperative drainage of the left biliary tree. The third patient, who had a right trisectionectomy, had the FLR drained preoperatively and underwent PVE but transient liver failure developed postoperatively and resolved over time. There were 5 perioperative deaths (5% operative mortality). Two patients died from liver failure and 3 patients had an initial infectious complication (pneumonia, n ⫽ 2; subphrenic abscess, n ⫽ 1) leading to liver failure. One additional patient died after discharge from hospital and outside of the 30-day window but within 90 days of
608
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
J Am Coll Surg
Figure 3. Overall survival of 95 patients undergoing resection for hilar cholangiocarcinoma. Overall 5-year survival rate was 43%.
surgery from what appeared to be a myocardial event, although this was not confirmed by autopsy. Eighty of 95 patients resected (84%) had histologically negative margins (R0). The remaining 15 patients had microscopically positive margins. Twelve of these specimens were initially called negative by frozen-section analysis, but were found to be positive on permanent section. In 3 cases, it was known that the margins were microscopically positive at surgery; however, additional extension of the biliary resection margin was not technically possible. In 2 left trisectionectomies, the segment 6 and segment 7 ducts were already widely separate, and in 1 right trisectionectomy, resection had already been performed back to separate segment 2 and 3 ducts. Papillary tumors occurred in 5% of resected patients. Twenty-five percent of patients had well-differentiated tumors, with the remainder either moderately or poorly differentiated. Mean tumor size was 3.4 ⫾ 1.9 cm. Thirtyeight percent of patients had positive nodes in the hepatoduodenal ligament. Of the 42 patients who underwent portal vein resection, 17 (41%) had true invasion of the portal vein wall by tumor. The remaining 25 patients had dense fibrosis surrounding the portal vein, within which could be found tumor but no true vascular invasion. The distribution of patients by preoperative staging system is demonstrated in Table 1. An assessment of the 2 preoperative staging systems used demonstrated no correlation between the Bismuth-Corlette staging system5 or the staging system proposed by Jarnagin and Blumgart,3 and
survival. However, when patients who had lobar hypertrophy preoperatively because of tumor involvement of the contralateral liver or induced with PVE were grouped together and compared with those patients who did not have lobar hypertrophy, they had a significantly lower operative mortality than those patients without lobar hypertrophy (0% versus 9%; p ⫽ 0.02 ). Median actuarial survival for all patients was 38 months, with median follow-up time of 32 months. Five-year actuarial survival rate was 43% for all patients undergoing resection (Fig. 3) and 50% for those patients who were resected with negative margins. Patients who had an R0 resection survived longer than patients resected with microscopically positive margins (median survival 58 months; 95% CI, 36–81 months versus 26 months 95% CI, 14–31 months; p ⬍ 0.01; Fig. 4). There were no long-term survivors who had positive resection margins. Perhaps not surprisingly, patients resected with positive margins were more likely to have positive lymph nodes than patients with negative margins (87% versus 28%; p ⬍ 0.01). There was no difference in either operative mortality (2% versus 8%; p ⫽ 0.23) or long-term survival (Fig. 5) between patients who had portal vein resections and those who did not. Factors not associated with an effect on survival by univariate analysis in patients who were resected included age older than 70 years, sex, preoperative Bismuth-Corlette and Jarnagin-Blumgart stage, tumor size ⬎2.5 cm, type of resection, and postoperative American Joint Committee on
Vol. 212, No. 4, April 2011
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
609
Figure 4. Patients resected with negative margins had a 5-year survival rate of 50%, which was significantly better than those patients resected with positive margins, with no patient with positive margins surviving longer than 40 months (p ⬍ 0.05 log rank).
Cancer stage. Patients with lymph node involvement by tumor appeared to have a worse 5-year survival rates than patients with node-negative status; however, this difference was not statistically significant (23% versus 47%; p ⫽ 0.06; Fig. 6). Patients with well-differentiated tumors showed a trend toward improved survival compared with patients with moderately or poorly differentiated tumors, however, this trend was not significant (p ⫽ 0.10). Only R0 status demonstrated an effect on survival at a level of significance of ⬍0.05 by univariate analysis. A Cox regression analysis was performed on variables with p ⱕ 0.10, including R0 status, lymph node status, and tumor differentiation. This confirmed that only R0 status remained a substantial predictor of survival. Although there was no difference in overall survival between the first half of the study period and the second, there was a difference in 30-day mortality, with 10% operative mortality in the first half of the study and no operative mortality in the second half of the study (p ⫽ 0.04), when PVE and preoperative biliary drainage of the FLR were used as a standard.
DISCUSSION HCCA continues to be a difficult challenge for the surgeon. Achieving negative surgical margins when resecting this relatively uncommon tumor is technically demanding
because of the close proximity of the bile duct bifurcation to the vascular inflow of the liver. The addition of liver resection to the procedure has become the standard of care. The addition of liver resection increases the number of patients who can be offered potentially margin-free resection, and increases the oncologic clearance of the procedure in patients who would previously have been treated with bile duct resection only. Extended hepatectomy in the setting of biliary obstruction and elevated bilirubin can be associated with increased risk, however.14-16 Attempts to reduce the risks associated with biliary obstruction by preoperative biliary drainage remain controversial, with arguments for and against debated vigorously.17,18 There is little argument that the presence of preoperative biliary drainage increases the rate of bacterial colonization of the biliary tree and can increase the rate of perioperative infectious complications.3,19 Conversely, several recent series of hepatic resections for HCCA demonstrate a reduced incidence of liver failure when preoperative biliary drainage was used, although PVE was also used in those series.20-23 In practice, biliary drainage has occurred before referral in the majority of cases seen by the hepatobiliary surgeon. In our series, 84 of 95 patients already had biliary stents placed before surgical referral. In many cases, stents were placed without consideration of subsequent surgery and were placed into the side of the liver that would be removed rather than the side of the planned remnant liver after resection. Kennedy
610
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
J Am Coll Surg
Figure 5. There was no significant difference in survival between those patients who had portal vein resection and those that did not (p ⫽ 0.6 log rank).
and colleagues,24 reporting on the Memorial experience, demonstrated that biliary drainage of the FLR when it was ⬍30% of liver volume was beneficial in their series. In that article, there was no correlation between preoperative bilirubin levels and outcomes. This finding can be explained in part by the fact that in many cases in that series, stents were placed in the biliary tree ipsilateral to the tumor, allowing improved function, decreased bilirubin, and recovery of the side of the liver that was to be removed rather than the side that was to remain. Ideally, patients should undergo early referral, with the need for, and planning of, preoperative biliary drainage done in concert with the surgeon. The importance of the roles of remnant liver function and liver volume and the concept of lobar atrophy and subsequent contralateral lobar hypertrophy and its effect on operative mortality were again demonstrated in this study. Patients who had lobar hypertrophy induced by either contralateral involvement by tumor (JarnaginBlumgart T2) or by preoperative PVEs were less likely to die in the perioperative period. Only 2 of the 5 deaths in this series were from isolated liver failure without initial infectious complication. However, the 3 other patients who died had infection initially that led to development of liver failure as a component of their demise. Seyama and colleagues21 used preoperative PVE in a series of 58 pa-
tients; all patients undergoing resection of ⬎50% of initial liver volume for HCCA and reported a 0% operative mortality, although there was a 43% morbidity, which is similar to our series. Although speculative, it might be that the improved remnant liver function provided by preoperative lobar hypertrophy not only prevents primary liver failure but also provides a degree of hepatic functional reserve that allows patients to better tolerate a subsequent insult, such as an infectious complication. Complication rates for extended hepatectomy for HCCA range from 40% to 52% in recent series.3,20,22,25-27 There was a change in practice during the duration of the study after our initial report in 2005.11 All patients in the second half of the study received preoperative biliary drainage of the FLR with PVE performed if no evidence of lobar hypertrophy was present. Although this was clearly not a randomized prospective trial, there was a decreased operative mortality noted in the second half of the study. We would like to attribute the decrease in mortality to the initiation of standardized biliary drainage of the FLR and use of PVE; however, it might be that improved selection of patients or greater experience over time led to some of the improvement. Of note, the improvement in operative mortality did not translate into improved long-term survival. In any event, it would appear reasonable to use preoperative PVE in all patients under-
Vol. 212, No. 4, April 2011
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
611
Figure 6. Patients with lymph node involvement with tumor appeared to have a worse survival rate at 5 years than those patients with negative lymph nodes; however, this difference was not statistically significant (23% versus 47%; p ⫽ 0.06).
going planned extended hepatectomy for HCCA who do not have pre-existing lobar hypertrophy of the remnant liver from contralateral tumor involvement. Only the presence of an R0 resection proved to be predictive of an improvement in survival. Patients resected with negative margins had a 50% 5-year actuarial survival rate, and no patients with positive margins were alive past 48 months. These findings are similar to the larger Japanese experience reported from Nagoya that demonstrated a 52% 5-year survival rate in patients undergoing an R0 resection and in which there was liberal use of portal vein resection along with hepatic arterial resection performed.27 Patients in the current study who had lymph node involvement showed a trend toward having worse survival that did not reach statistical significance (p ⫽ 0.06). The power of this study was not sufficient to determine whether there was a true difference in survival associated with lymph node involvement, and the possibility of a type II error is relatively high. The literature provides conflicting results about the association of lymph node status on survival, with some authors showing a clear effect21,27 and some showing none.3,25 The most recent studies with larger numbers of patients in the series would indicate that lymph node involvement is independently associated with poor prognosis.22,28 Positive lymph node involvement suggests a
more advanced, biologically aggressive tumor that, intuitively, should have worse outcomes. More aggressive tumor biology might also explain the association between patients with positive nodes and positive resection margins noted in the current study, despite fairly aggressive resections with extended hepatectomy and vascular resection. Although suggesting a poorer prognosis, it is also clear that positive lymph nodes in the porta hepatis do not preclude a curative procedure if negative margins can be obtained. Although liver resection aids in the management of longitudinal extension of tumor out the biliary tree and also with unilateral involvement of vascular structures with tumor, involvement of the portal vein bifurcation remains difficult to manage. Some of the early reports of liver resection for HCCA recognized this difficulty and suggested portal vein resection with subsequent reconstruction in an attempt to obtain clear margins.29 The degree of difficulty and assumed risk of increased mortality associated with portal vein resection led to little enthusiasm for its adoption, however. Recent development of specialized hepatobiliary surgical groups, along with the use of techniques developed for liver transplantation have renewed interest in portal vein resection for HCCA. The operative mortality in our series, if anything, appeared lower in the group undergoing portal vein resection as compared with the group that
612
Hemming et al
Portal Vein Resection in Hilar Cholangiocarcinoma
did not undergo portal vein resection (2% versus 8%), although this was not a statistically significant difference (p ⫽ 0.37). In previous reports, we have suggested that the relatively low mortality observed with portal vein resection might be secondary to compensatory hypertrophy of the remnant liver and improved postoperative liver function that occurs in portal vein resection patients due to tumor involvement of the contralateral portal vein.11,30 Other groups have also reported no difference in operative mortality in patients undergoing portal vein resection versus standard resection for HCCA.9,26,31 The impact of portal vein resection on long-term survival is less clear. Ebata and colleagues31 reported worse long-term survival in patients requiring portal vein resection; however, multivariate analysis demonstrated that portal vein resection itself did not worsen survival, but it was the presence of transluminal tumor or positive margins that had a negative impact. Lee and colleagues,22 in a series of 302 patients with HCCA, demonstrated no difference in long-term survival in the 38 patients who underwent portal vein resection versus the 264 patients who did not require portal vein resection. The Nagoya group performed portal vein resection in more than one-third of 298 patients undergoing resection for HCCA and found an unexpectedly better survival in those patients undergoing vascular resection.27 Neuhaus and colleagues9,32 have taken portal vein resection a step further. In a series of extended right hepatectomies combined with regional lymphadenectomy and routine portal vein resection and reconstruction, they have reported a 72% 5-year survival rate, which is by far the best result reported to date for resection of HCCA. The portal vein is resected in this technique in all patients to allow a “no-touch” resection of the tumor and adjacent tissue, and the authors suggest that it is the adherence to such oncologic principles that yields favorable results. In the last several years, we have adopted many of the no-touch principles espoused by Neuhaus and colleagues32 and, when performing extended right hepatectomies, will have a very low threshold to resect the portal vein bifurcation. Resecting the portal vein bifurcation when performing a left trisectionectomy is substantially more difficult because of the relatively short course of the right portal vein before branching and was not commonly performed in the current series. We could show neither a survival advantage nor disadvantage associated with en bloc resection of the portal vein. The results suggest that although portal vein resection should not necessarily be a standard part of every resection, the need for portal vein resection should not be a contraindication to resection for cure.
J Am Coll Surg
CONCLUSIONS In summary, the current study confirms that an aggressive surgical approach to HCCA that includes hepatic resection as standard therapy is warranted. The operative mortality was 5%, which is in line with the mortality for extended hepatectomies performed for other indications at our institution. A 50% 5-year survival rate can be expected if negative margins can be obtained, even if this requires resection and reconstruction of the hepatic vasculature. Obtaining negative resection margins is the single most important factor in achieving prolonged survival. The use of strategies to optimize the FLR before resection can allow more aggressive resections in pursuit of negative margins and reduce operative mortality. Author Contributions Study conception and design: Hemming, Mekeel, Khanna, Baquerizo, Kim Acquisition of data: Hemming, Baquerizo, Kim Analysis and interpretation of data: Hemming, Mekeel, Kim Drafting of manuscript: Hemming, Mekeel, Khanna, Baquerizo, Kim Critical revision: Hemming, Mekeel, Khanna, Baquerizo, Kim
REFERENCES 1. Olnes MJ, Erlich R. A review and update on cholangiocarcinoma. Oncology 2004;66:167–179. 2. Nakeeb A, Pitt HA, Sohn TA, et al. Cholangiocarcinoma. A spectrum of intrahepatic, perihilar, and distal tumors. Ann Surg 1996;224:463–473; discussion 473–475. 3. Jarnagin WR, Fong Y, DeMatteo RP, et al. Staging, resectability, and outcome in 225 patients with hilar cholangiocarcinoma. Ann Surg 2001;234:507–517; discussion 517–519. 4. Tompkins RK, Thomas D, Wile A, Longmire WP Jr. Prognostic factors in bile duct carcinoma: analysis of 96 cases. Ann Surg 1981;194:447–457. 5. Bismuth H, Corlette MB. Intrahepatic cholangioenteric anastomosis in carcinoma of the hilus of the liver. Surg Gynecol Obstet 1975;140:170–178. 6. Launois B, Campion JP, Brissot P, Gosselin M. Carcinoma of the hepatic hilus. Surgical management and the case for resection. Ann Surg 1979;190:151–157. 7. Beazley RM, Hadjis N, Benjamin IS, Blumgart LH. Clinicopathological aspects of high bile duct cancer. Experience with resection and bypass surgical treatments. Ann Surg 1984;199: 623–636. 8. Kosuge T, Yamamoto J, Shimada K, et al. Improved surgical results for hilar cholangiocarcinoma with procedures including major hepatic resection. Ann Surg 1999;230:663–671. 9. Neuhaus P, Jonas S, Bechstein WO, et al. Extended resections for hilar cholangiocarcinoma. Ann Surg 1999;230:808–818; discussion 819. 10. Launois B, Reding R, Lebeau G, Buard JL. Surgery for hilar
Vol. 212, No. 4, April 2011
11. 12. 13. 14. 15. 16. 17. 18. 19.
20.
21. 22. 23. 24. 25. 26.
27. 28. 29. 30.
cholangiocarcinoma: French experience in a collective survey of 552 extrahepatic bile duct cancers. J Hepatobiliary Pancreat Surg 2000;7:128–134. Hemming AW, Reed AI, Fujita S, et al. Surgical management of hilar cholangiocarcinoma. Ann Surg 2005;241:693–699; discussion 699–702. Hemming AW, Reed AI, Howard RJ, et al. Preoperative portal vein embolization for extended hepatectomy. Ann Surg 2003; 237:686–691; discussion 691–693. Mullen JT, Ribero D, Reddy SK, et al. Hepatic insufficiency and mortality in 1,059 noncirrhotic patients undergoing major hepatectomy. J Am Coll Surg 2007;204:854–862; discussion 862–864. Su CH, Tsay SH, Wu CC, et al. Factors influencing postoperative morbidity, mortality, and survival after resection for hilar cholangiocarcinoma. Ann Surg 1996;223:384–394. Melendez J, Ferri E, Zwillman M, et al. Extended hepatic resection: a 6-year retrospective study of risk factors for perioperative mortality. J Am Coll Surg 2001;192:47–53. Imamura H, Seyama Y, Kokudo N, et al. One thousand fifty-six hepatectomies without mortality in 8 years. Arch Surg 2003; 138:1198–1206; discussion 1206. Laurent A, Tayar C, Cherqui D. Cholangiocarcinoma: preoperative biliary drainage (Con). HPB (Oxford) 2008;10:126–129. Nimura Y. Preoperative biliary drainage before resection for cholangiocarcinoma (Pro). HPB (Oxford) 2008;10:130–133. Hochwald SN, Burke EC, Jarnagin WR, et al. Association of preoperative biliary stenting with increased postoperative infectious complications in proximal cholangiocarcinoma. Arch Surg 1999;134:261–266. Kawasaki S, Imamura H, Kobayashi A, et al. Results of surgical resection for patients with hilar bile duct cancer: application of extended hepatectomy after biliary drainage and hemihepatic portal vein embolization. Ann Surg 2003;238:84–92. Seyama Y, Kubota K, Sano K, et al. Long-term outcome of extended hemihepatectomy for hilar bile duct cancer with no mortality and high survival rate. Ann Surg 2003;238:73–83. Lee SG, Song GW, Hwang S, et al. Surgical treatment of hilar cholangiocarcinoma in the new era: the Asian experience. J Hepatobiliary Pancreat Sci 2010;17:476–489. Maguchi H, Takahashi K, Katanuma A, et al. Preoperative biliary drainage for hilar cholangiocarcinoma. J Hepatobiliary Pancreat Surg 2007;14:441–446. Kennedy TJ, Yopp A, Qin Y, et al. Role of preoperative biliary drainage of liver remnant prior to extended liver resection for hilar cholangiocarcinoma. HPB (Oxford) 2009;11:445–451. Rea DJ, Munoz-Juarez M, Farnell MB, et al. Major hepatic resection for hilar cholangiocarcinoma: analysis of 46 patients. Arch Surg 2004;139:514–523; discussion 523–525. Shimizu H, Kimura F, Yoshidome H, et al. Aggressive surgical resection for hilar cholangiocarcinoma of the left-side predominance: radicality and safety of left-sided hepatectomy. Ann Surg 2010;251:281–286. Igami T, Nishio H, Ebata T, et al. Surgical treatment of hilar cholangiocarcinoma in the “new era”: the Nagoya University experience. J Hepatobiliary Pancreat Sci 2010;17:449–454. Ito K, Ito H, Allen PJ, et al. Adequate lymph node assessment for extrahepatic bile duct adenocarcinoma. Ann Surg 2010;251: 675–681. Hadjis NS, Blenkharn JI, Alexander N, et al. Outcome of radical surgery in hilar cholangiocarcinoma. Surgery 1990;107:597– 604. Hemming AW, Kim RD, Mekeel KL, et al. Portal vein resection
Hemming et al
Discussion
613
for hilar cholangiocarcinoma. Am Surg 2006;72:599–604; discussion 605. 31. Ebata T, Nagino M, Kamiya J, et al. Hepatectomy with portal vein resection for hilar cholangiocarcinoma: audit of 52 consecutive cases. Ann Surg 2003;238:720–727. 32. Neuhaus P, Jonas S, Settmacher U, et al. Surgical management of proximal bile duct cancer: extended right lobe resection increases resectability and radicality. Langenbecks Arch Surg 2003;388:194–200.
Discussion DR WILLIAM JARNAGIN (New York, NY): This is a very good paper, illustrating the results that are possible in hilar cholangiocarcinoma from a technically proficient surgeon with a good understanding of the disease, working with an experienced team. Ninetyfive resections with an 84% R0 rate over approximately 10 years may seem like a small number, but anyone with any familiarity with this problem realizes that this is, in fact, a very large experience, and it doesn’t take into account those that were unresectable, which is probably around twice that number. Furthermore, anyone who has performed this operation with any frequency understands how technically difficult it can be and also just how good these results are with respect to morbidity and mortality. I have a few questions. I want to emphasize the important point that you made, which is that in the face of hypertrophy of the future liver remnant (FLR), due to contralateral atrophy caused by tumor, portal vein embolization is unnecessary, and in fact, preoperative biliary drainage probably adds little to improving perioperative morbidity, and indeed in this situation, placing biliary catheters likely exposes patients to the risks of the procedure with little benefit. You state it is your current practice not to drain the future liver remnant if there is hypertrophy, but you don’t specify the level required. In our experience, an FLR greater than 30% seemed to be the cut-off value. But I wonder what level you used? Also, you showed that patients with hypertrophy of the FLR noted preoperatively, whether due to tumor involvement or from an embolization, had 0% operative mortality, compared with 9% for those without. Was this independent of preoperative biliary drainage? I suspect that most of the latter group had undergone preoperative biliary decompression, which underscores the point that it is the quality of the future liver remnant that is ultimately the most important factor, and that biliary drainage in and of itself is not necessarily as protective as many would like to believe. You talked about the Blumgart preoperative staging system, which is not really useful for predicting survival after resection, but is really for predicting the likelihood that a resection is possible and the likelihood of finding metastatic disease. As such, it really requires inclusion of the entire denominator of patients, including those who were unresectable. Over the period of study, how many patients had unresectable disease at presentation and how many did you explore without resecting? Finally, you show that operative mortality was no different in the portal vein resection versus no resection groups. What about the morbidity? I suspect it was higher in the group with vascular resec-