Prognostic Factors After Pancreatoduodenectomy with Extended Lymphadenectomy for Distal Bile Duct Cancer

ORIGINAL ARTICLE Prognostic Factors After Pancreatoduodenectomy With Extended Lymphadenectomy for Distal Bile Duct Cancer

Takanori Yoshida, MD; Toshifumi Matsumoto, MD; Atsushi Sasaki, MD; Yuji Morii, MD; Masanori Aramaki, MD; Seigo Kitano, MD

Background: Since 1995, we have been performing pan- Main Outcome Measures: Clinicopathologic char- creatoduodenectomy with regional and para-aortic lymph acteristics and long-term results. node dissection for patients with distal bile duct cancer. Prognostic indicators after extended lymphadenectomy Results: In 6 patients (22%) major surgical complica- have not been fully understood. tions occurred including 1 in-hospital death (3.7%). For 26 patients, the survival rates were 65% for 1 year and Hypothesis: Pancreatoduodenectomy with extended 37% for 3 and 5 years. Univariate analysis found that the lymphadenectomy and adjuvant chemotherapy is the treat- absence of lymph node metastasis, no more than 2 in- ment of choice for patients with distal bile duct cancer. volved nodes, and negative resection margins were predictors of survival. Multivariate analysis with a Cox pro- Design: In a retrospective study, univariate and multi- portional hazards regression model revealed that favorable variate models were used to analyze the effect of patient factors for survival included up to 2 positive nodes, nega- demographics, tumor characteristics, and treatment fac- tive resection margins, and the use of postoperative ad- tors on long-term survival. juvant chemotherapy.

Setting: Oita Medical University and its affiliated hos- Conclusions: Patients with up to 2 positive lymph pitals in Japan. nodes had a more favorable prognosis than that of other patients. We recommend pancreatoduodenectomy with Patients: From 1995 to 1999, 27 patients with distal extended lymphadenectomy and adjuvant chemo- bile duct cancer underwent pancreatoduodenectomy with therapy for the treatment of patients with distal bile extended lymphadenectomy. In 9 patients fluorouracil duct cancer. (500 mg/d) was infused continuously for 14 days after surgery as adjuvant chemotherapy. Arch Surg. 2002;137:69-73

N PATIENTS with distal bile duct Results from randomized controlled cancer, lymph node metastasis is trials of patients with adenocarcinoma of the associated significantly with treat- pancreatic head indicated that extended ment outcomes.1-4 The extent of lymphadenectomy and retroperitoneal soft dissection and histological exami- tissue clearance with pancreatoduodenal re- Ination of lymph nodes, however, varies be- section might be accepted as the proce- tween countries, and this variation may re- dure of choice in the treatment of node- sult in erroneous nodal staging and may positive patients with similar morbidity and account for differences in surgical results. mortality to standard dissection.10,11 The use In Japan, some surgeons advocate the com- of postoperative adjuvant chemoradio- plete removal of the primary bile duct can- therapy and intraoperative blood loss are cer with connective tissue clearance, in- considered independent prognostic fac- cluding lymph node and neural plexus tors.12 Because of its rarity, there are a few From the Department of dissection.4-6 Since 1995, we have been per- detailed reports about treatments for distal General Surgery, Nakatsu forming pancreatoduodenectomy with re- bile duct cancer.1-5,13,14 This retrospective Municipal Hospital gional and para-aortic lymph node dissec- study identifies prognostic factors after pan- (Drs Yoshida and Morii), tion for patients with distal bile duct creatoduodenectomy with extended lymph- and the First Department 7-9 of Surgery, Oita Medical cancer. The goal of this radical surgery adenectomy in patients with distal bile duct University (Drs Matsumoto, with extended lymphadenectomy is to pri- cancer, using a series of 27 patients, a suf- Sasaki, Aramaki, and Kitano), marily achieve locoregional control and to ficiently large population to allow for sta- Oita, Japan. secondarily obtain accurate tumor staging. tistical evaluation, and relating to a single

(REPRINTED) ARCH SURG/ VOL 137, JAN 2002 WWW.ARCHSURG.COM 69

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/27/2021 PATIENTS AND METHODS was adopted as the differentiation. Lymph nodes were considered positive if any lymph node in the resected specimen contained metastatic foci. The number and location of in- Between January 1, 1995, and December 31, 1999, at the volved nodes were carefully recorded. Lymph nodes were clas- First Department of Surgery, Oita Medical University, Oita, sified according to the General Rules for Surgical and Patho- Japan, and its affiliated hospitals, 100 patients with malig- logical Studies on Cancer of the Biliary Tract15: 5 indicates nant tumors underwent pancreatoduodenectomy with sys- suprapyloric; 6, infrapyloric; 7, left gastric artery; 8, com- tematic lymph node dissection including regional and para- mon hepatic artery; 9, celiac trunk; 11, splenic artery; 12, hepa- aortic nodes. The patient population consisted of 27 patients toduodenal ligament; 13, posterior pancreaticoduodenal; 14, with distal bile duct cancer, 24 with pancreatic head can- superior mesenteric artery; 15, middle colic artery; 16, para- cer, 21 with ampullary cancer, and 28 with other malig- aortic; and 17, anterior pancreaticoduodenal (Figure 1). Re- nancies. Distal bile duct cancer was defined as adenocar- section margins were considered positive if infiltrating ad- cinoma arising from the intrapancreatic bile duct. Pancreatic enocarcinoma was present at the pancreatic neck, proximal head, ampullary, and other adenocarcinomas were ex- bile duct, or periductal soft tissue margins. The lymphatic, cluded from this study. Of the 27 patients with distal bile venous, and perineural permeations were examined rou- duct cancer, 16 were treated by pancreatoduodenectomy tinely with cut sections. with conventional gastrectomy, and 11 underwent a pylorus- Intraoperative blood loss, transfused units of erythro- preserving procedure. Four patients underwent portal vein cytes, and operative time were investigated by reviewing the resection and reconstruction. Para-aortic lymph node dis- medical records. Patients were offered 2 options for post- section extended from the celiac axis to the inferior mes- operative treatment—chemotherapy or no chemotherapy— enteric artery and from the right margin of the inferior vena according to their condition. As adjuvant chemotherapy, fluo- cava to the left margin of the abdominal aorta. One pa- rouracil (500 mg/d) was infused continuously and tient died of an intra-abdominal abscess after the opera- systemically for 14 days after the pancreatoduodenec- tion and was excluded from survival analysis. The remain- tomy.16 ing 26 patients included 16 men and 10 women who had a Follow-up to December 31, 2000, was obtained for all mean age of 63.5 years (age range, 35-80 years). 26 patients and ranged from 12 to 71 months (mean du- The resected specimens with attached choledochal and ration of follow-up, 52.2 months; median duration of follow- peripancreatic lymph nodes were examined by the prepara- up, 54.0 months). Patient demographics, tumor charac- tion of serial tissue sections at 5-mm intervals and were stained teristics, and treatment factors were evaluated using with hematoxylin-eosin. All specimens were reviewed by a univariate and multivariate models to determine their effect single surgical pathologist (A.S.) according to the General Rules on long-term survival. The cumulative survival rate was cal- for Surgical and Pathological Studies on Cancer of the Biliary culated by the Kaplan-Meier method, and the survival curves Tract15 outlined by the Japanese Society of Biliary Surgery. were compared using the Mantel-Haenszel test. Multivar- Local extent of the primary tumor was classified into 2 groups iate analysis was performed using the stepwise forward Cox with the classification based on the presence of pancreatic proportional hazards regression model. Differences were parenchymal invasion.7-9 Predominant histological grading considered statistically significant at PϽ.05.

center, to minimize the problem of homogeneity of treat- fied as 12, 13, and/or 17. Seven patients had metastatic ment and follow-up. nodes in group 8 or 14 or both. Para-aortic lymph node involvement was observed in 5 patients. RESULTS There were 12 tumors (46%) without pancreatic invasion and 14 (54%) with pancreatic invasion (Table 2). Major surgery-related complications occurred in 6 (22%) Four patients (15%) had positive resection margins (proxi- of 27 patients, including leakage of the pancreaticojeju- mal bile duct in 2 and periductal soft tissue margins in nostomy in 5 (1 death), delayed bleeding in 2, myocar- the other), and 22 (85%) had negative margins. Most car- dial infarction in 1, and obstruction of the efferent loop cinomas either were well differentiated in 15 (58%) or of gastrojejunostomy in 1. One patient, in whom adju- moderately differentiated in 7 (27%). The median intra- vant chemotherapy was not administered, died of intraoperative blood loss was 1000 mL, the median units of abdominal sepsis due to leakage of the pancreaticojeju- erythrocytes transfused was 1, and the median opera- nostomy after 38 postoperative days. The in-hospital death tive time was 7 hours. Nine patients who underwent ad- rate was 3.7%. Two patients with delayed intraperitoneal juvant chemotherapy had no chemotherapy-related com- hemorrhage were treated successfully with transcatheter plications. arterial embolization and conservative treatment.17 Reop- The survival rates (Figure 2) were 65% for 1 year eration was needed 20 days after pancreatoduodenec- and 37% for 3 and 5 years with survival time ranging from tomy in 1 patient with obstruction of the efferent loop of 4 to 64 months (mean, 26.4 months; median, 20.5 gastrojejunostomy. months). Of 26 patients, 16 (4 with positive and 12 with In 15 patients (58%), metastatic adenocarcinoma was negative margins) died of either the tumor or a recur- identified in resected lymph nodes (Table 1). Of the 15 rence. The cause of death was peritoneal dissemination node-positive patients, 6 had 1 or 2 nodes involved, and in 5, liver metastases in 4, local recurrence due to posi- 9 had 3 or more nodes involved. All node-positive pa- tive resection margins in 4, and retroperitoneal recur- tients had involved lymph nodes in the groups classi- rence in 3. In 9 patients with postoperative adjuvant che-

(REPRINTED) ARCH SURG/ VOL 137, JAN 2002 WWW.ARCHSURG.COM 70

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/27/2021 motherapy, 2 had local and 2 had retroperitoneal recur- rences. Liver metastases occurred only in patients who did not receive adjuvant chemotherapy. Univariate analysis (Table 2) found prognostic fac- 7 tors for survival to include the absence of lymph node 9 11 8 metastasis (P=.02), no more than 2 involved lymph nodes 16 (PϽ.001), and negative resection margins (P=.002). For 12 patients without positive lymph nodes, survival rates were 91% for 1 year and 61% for 3 and 5 years (Figure 3). For patients with up to 2 positive nodes, survival rates were 83% for 1 year and 50% for 3 and 5 years. The survival rate for patients with 3 or more positive nodes was 17 22% for 1 year, and it decreased to 0% by 3 years. Sur- 14 vival rates were statistically similar between patients with- 13

out positive nodes and those with 1 or 2 positive nodes 15 (P=.61). Significant differences existed between patients with 3 or more positive nodes and those without positive nodes (PϽ.001) as well as those with 1 or 2 involved nodes (P=.04). Figure 1. Classification of lymph node groups in the management of carcinoma of the distal bile duct. They are as follows: 7, left gastric artery; Multivariate analysis with a stepwise regression 8, common hepatic artery; 9, celiac trunk; 11, splenic artery; 12, model showed that prognostic factors for survival in- hepatoduodenal ligament; 13, posterior pancreaticoduodenal; 14, superior cluded the number of lymph nodes involved, the status mesenteric artery; 15, middle colic artery; 16, para-aortic; and 17, anterior of resection margins, and the use of postoperative adju- pancreaticoduodenal. vant chemotherapy (Table 3). Although the use of adjuvant chemotherapy was a significant prognostic factor for survival by multivariate analysis, no difference was Table 1. Patterns of Lymph Node Metastasis found by the Mantel-Haenszel test (P=.12). in Node-Positive Patients

No. (%) of Range of Involved COMMENT Node-Positive Lymph Node Lymph Node Group* Patients (Mean, Median) The presence of lymph node metastasis is reportedly sig- 12 and/or 13 and/or 17 15 (100) 1-6 (2.3, 2.0) nificantly associated with the outcome of patients with 8 and/or 14 7 (47) 0-9 (1.2, 0.0) 1-4 distal bile duct cancer. In our study, however, the 5-year 16 5 (33) 0-3 (0.7, 0.0) survival rate for node-positive patients was 20%, and Total 15 1-17 (4.2, 3.0) lymph node involvement failed to affect survival adversely if no more than 2 positive nodes were detected. *The group designators indicate the following: 8, common hepatic artery; 12, hepatoduodenal ligament; 13, posterior pancreaticoduodenal; These results suggest that the extent of lymph node dis- 14, superior mesenteric artery; 16, para-aortic; and 17, anterior section is an important component of radical surgery. In pancreaticoduodenal lymph nodes. patients with distal bile duct cancer, extended lymph node dissection can contribute to locoregional control and accurate nodal staging and result in favorable outcomes.4 tastases associated with the primary tumor. At present, We found no record of technical difficulties with the a pathologic N category has been defined as the number extended lymphadenectomy that included regional and of involved nodes in the stomach (N1 indicates 1-6 posi- para-aortic nodes. Our results of median intraoperative tive nodes; N2, 7-15 positive nodes; and N3, Ն16 posi- blood loss, median transfused units of erythrocytes, and tive nodes) and colorectum (N1 indicates 1-3 positive median operative time are almost comparable with those nodes; N2, Ն4 positive nodes), and this has been ac- of other reports.10-12 With this procedure, early postop- cepted as a simple and useful classification system.19,20 erative complications are problematic although recent oc- In addition, ampullary carcinoma patients with up to 2 currences are lacking. We only have performed lymph positive nodes have a better survival rate than that of other node dissection around the superior mesenteric artery patients.21,22 We, therefore, analyzed the number of in- with either total circular or left hemicircular preserva- volved nodes in distal bile duct cancer. A lymph node tion of the nerve plexus, and this helps to prevent in- staging system based on the number of involved nodes tractable diarrhea and digestive malnutrition.4,18 Intra- may be applied to pancreatoduodenal malignancies af- operative blood loss of less than 700 mL and the absence ter further investigation. of intraoperative blood transfusion were previously re- Tumor involvement of the surgical margins is an im- ported as predictors of improved survival in patients with portant prognostic factor in patients with cancer of the pancreatic head cancer.12 In our study, however, intrapancreatic head and distal bile duct.1,4,12 Our results show operative blood loss and blood transfusion failed to be that the survival rates for patients with negative resec- important prognostic factors; this may have been owing tion margins are significantly higher than those for pa- to the small sample size. tients with positive surgical margins and that the status In the past, staging systems for malignancy deter- of the resection margins is an independent prognostic fac- mined the nodal status by the location of lymph node me- tor. Of 27 patients, 2 (7.4%) had a positive proximal bile

(REPRINTED) ARCH SURG/ VOL 137, JAN 2002 WWW.ARCHSURG.COM 71

Survival Rate, %

Factor No. of Patients 1-Year 3-Year 5-Year P Value Demographics Age, y Ͻ65 11 73 42 42 .47 Ն65 15 60 33 33 Sex Male 16 69 41 41 .56 Female 10 60 30 30 Tumor characteristics Pancreatic invasion Negative 12 75 58 58 .10 Positive 14 57 16 16 Lymph node metastasis Negative 11 91 61 61 .02 Positive 15 47 20 20 No. of involved lymph nodes Յ217885757Ͻ.001 Ն392200 Differentiation* Well 15 67 33 33 .90 Others 11 64 42 42 Lymphatic permeation Negative 5 80 60 60 .33 Positive 21 62 31 31 Venous permeation Negative 11 73 64 64 .06 Positive 15 60 15 . . .† Perineural permeation Negative 4 100 75 75 .14 Positive 22 59 30 30 Resection margins Negative 22 73 44 44 .002 Positive 4 25 0 0 Treatment factors Blood loss, mL Ͻ1000 12 83 46 . . . .15 Ն1000 14 50 29 29 Transfusions received No 7 86 51 . . . .30 Yes 19 58 32 32 Operative time, h Ͻ7 13 77 51 51 .12 Ն7 13 54 23 ... Adjuvant chemotherapy received No 17 53 27 27 .12 Yes 9 89 56 56

*Well indicates well-differentiated tubular adenocarcinoma; others, moderately or poorly differentiated adenocarcinoma. †Ellipses indicate not applicable.

100 duct transection margin owing to superficial spreading of more than 30 mm from the main papillary tumor, and 80 these patients died of anastomotic recurrence. Al- though major hepatic resection with pancreatoduode- 60 nectomy may allow for a negative proximal bile duct transection margin, careful assessment and treatment are 40 necessary because of high morbidity and mortality Survival Rate, % rates.23-25 20 Locoregional recurrence and distant (mainly hepatic) metastasis are major causes of treatment failures. Sev- 0 1 2 3 4 5 6 eral studies of patients with pancreatic cancer reported that Time After the Operation, y adjuvant chemotherapy or chemoradiotherapy after cura- Figure 2. Survival rates after pancreatoduodenectomy with extended lymph tive pancreatectomy improved survival by decreasing node dissection in 26 patients with distal bile duct cancer. hepatic metastasis.12,16 These results led us to believe that

(REPRINTED) ARCH SURG/ VOL 137, JAN 2002 WWW.ARCHSURG.COM 72

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/27/2021 Presented in part at the 101st Congress of the Japan Table 3. Prognostic Factors for Survival Surgical Society, Sendai, Japan, April 12, 2001. by Multivariate Analysis Corresponding author and reprints: Takanori Yoshida, MD, Department of General Surgery, Nakatsu Municipal Hazard Factor Coefficient P Value Ratio 95% CI* Hospital, 173 Shimoikenaga, Nakatsu, Oita 871-8511, Japan. No. of involved nodes Յ2 1.0 REFERENCES Ն3 2.015 .003 7.502 1.988-28.318 Resection margins 1. Langer JC, Langer B, Taylor BR, Zeldin R, Cummings B. Carcinoma of the ex- Negative 1.0 trahepatic bile ducts: results of an aggressive surgical approach. Surgery. 1985; Positive 1.525 .04 4.594 1.107-19.072 98:752-759. Adjuvant 2. Ouchi K, Matsuno S, Sato T. Long-term survival in carcinoma of the biliary tract: chemotherapy analysis of prognostic factors in 146 resections. Arch Surg. 1989;124:248-252. Yes 1.0 3. Fong Y, Blumgart LH, Lin E, Fortner JG, Brennan MF. Outcome of treatment for No 1.353 .04 3.869 1.045-14.332 distal bile duct cancer. Br J Surg. 1996;83:1712-1715. 4. Kayahara M, Nagakawa T, Ohta T, Kitagawa H, Tajima H, Miwa K. Role of nodal *CI indicates confidence interval. involvement and the periductal soft-tissue margin in middle and distal bile duct cancer. Ann Surg. 1999;229:76-83. 5. Kurosaki I, Tsukada K, Hatakeyama K, Muto T. The mode of lymphatic spread in 100 carcinoma of the bile duct. Am J Surg. 1996;172:239-243. No Involved Nodes (n = 11 Patients) 1 or 2 Involved Nodes (n = 6 Patients) 6. Nakao A, Harada A, Nonami T, et al. Lymph node metastases in carcinoma of ≥ 80 3 Involved Nodes (n = 9 Patients) the head of the pancreas region. Br J Surg. 1995;82:399-402. 7. Yoshida T, Aramaki M, Matsumoto T, Morii Y, Sasaki A, Kitano S. The pattern of lymphatic spread in carcinoma of the distal bile duct. Int Surg. 1998;83:124-127. 60 8. Yoshida T, Aramaki M, Bandoh T, et al. Para-aortic lymph node metastasis in carcinoma of the distal bile duct. Hepatogastroenterology. 1998;45:2388-2391. 40 9. Yoshida T, Shibata K, Yokoyama H, et al. Patterns of lymph node metastasis in

Survival Rate, % carcinoma of the distal bile duct. Hepatogastroenterology. 1999;46:1595-1598. 20 10. Pedrazzoli S, DiCarlo V, Dionigi R, et al, for the Lymphadenectomy Study Group. Standard versus extended lymphadenectomy associated with pancreatoduodenectomy in the surgical treatment of adenocarcinoma of the head of the pancreas: a 0 1 2 3 4 5 6 multicenter, prospective, randomized study. Ann Surg. 1998;228:508-517. Time After the Operation, y 11. Yeo CJ, Cameron JL, Sohn TA, et al. Pancreaticoduodenectomy with or without extended retroperitoneal lymphadenectomy for periampullary adenocarcinoma: Figure 3. Survival rates of patients with 0, 1 or 2, and 3 or more involved comparison of morbidity and mortality and short-term outcome. Ann Surg. 1999; lymph nodes. Significant differences existed between patients with no 229:613-624. Ͻ involved nodes and those with 3 or more involved nodes (P .001) and 12. Yeo CJ, Abrams RA, Grochow LB, et al. Pancreaticoduodenectomy for pancreatic between those with 1 or 2 involved nodes and those with 3 or more involved adenocarcinoma: postoperative adjuvant chemoradiation improves survival: a pro- nodes (P=.04). Significant difference was absent between patients with no spective, single-institution experience. Ann Surg. 1997;225:621-636. involved nodes and those with 1 or 2 involved nodes (P=.61). 13. 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. radical resection with wide lymphadenectomy needed to 14. Nagorney DM, Donohue JH, Farnell MB, Schleck CD, Ilstrup DM. Outcomes af- be combined with an additional therapy to prevent he- ter curative resections of cholangiocarcinoma. Arch Surg. 1993;128:871-879. patic metastasis. Therefore, in 1995, we began adjuvant che- 15. Japanese Society of Biliary Surgery. General Rules for Surgical and Pathologi- cal Studies on Cancer of the Biliary Tract [in Japanese]. 4th ed. Tokyo, Japan: motherapy for patients in whom distal bile duct cancer was Kanehara Shuppan; 1997. resected by extended operation according to their condi- 16. Ishikawa O, Ohigashi H, Sasaki Y, et al. Liver perfusion chemotherapy via both the tion. In our present study, the use of adjuvant chemo- hepatic artery and portal vein to prevent hepatic metastasis after extended pancre- therapy was a prognostic factor by multivariate analysis al- atectomy for adenocarcinoma of the pancreas. Am J Surg. 1994;168:361-364. though no difference was found by univariate analysis. 17. Yoshida T, Matsumoto T, Morii Y, et al. Delayed massive intraperitoneal hemorrhage after pancreatoduodenectomy. Int Surg. 1998;83:131-135. Further investigations are needed that examine larger groups 18. Kawabata A, Hamanaka Y, Suzuki T. Potentiality of dissection of the lymph nodes of patients who undergo adjuvant chemotherapy and are with preservation of the nerve plexus around the superior mesenteric artery. Hepa- observed for long-term follow-up periods. To confirm our togastroenterology. 1998;45:236-241. results, a prospective, randomized trial should be under- 19. Sobin LH, Wittekind C, eds. International Union Against Cancer (UICC): TNM Clas- taken. sification of Malignant Tumours. 5th ed. New York, NY: John Wiley & Sons Inc; 1997. 20. Kodera Y, Yamamura Y, Shimizu Y, et al. The number of metastatic lymph nodes: Although the limited number of cases observed does a promising prognostic determinant for gastric carcinoma in the latest edition not allow for definite conclusions, the results of this study of the TNM classification. J Am Coll Surg. 1998;187:597-603. would seem to recommend pancreatoduodenectomy with 21. Roder JD, Schneider PM, Stein HJ, Siewert JR. Number of lymph node metas- extended lymphadenectomy and adjuvant chemo- tases is significantly associated with survival in patients with radically resected carcinoma of the ampulla of Vater. Br J Surg. 1995;82:1693-1696. therapy as the treatment of choice in the surgical man- 22. Shirai Y, Ohtani T, Tsukada K, Hatakeyama K. Patterns of lymphatic spread of agement of patients with distal bile duct cancer. Lymph carcinoma of the ampulla of Vater. Br J Surg. 1997;84:1012-1016. node involvement failed to affect survival adversely if no 23. Tsukada K, Yoshida K, Aono T, et al. Major hepatectomy and pancreatoduodenec- more than 2 positive nodes were detected. tomy for advanced carcinoma of the biliary tract. Br J Surg. 1994;81:108-110. 24. Sakamoto E, Nimura Y, Hayakawa N, et al. The pattern of infiltration at the proximal border of hilar bile duct carcinoma: a histologic analysis of 62 resected cases. This investigation was supported in part by Grant-in-Aid Ann Surg. 1998;227:405-411. for Scientific Research C-2-11671255 from the Ministry of 25. Nagino M, Kamiya J, Kanai M, et al. Right trisegment portal vein embolization for Education, Science, and Culture of Japan, Tokyo. biliary tract carcinoma: technique and clinical utility. Surgery. 2000;127:155-160.

(REPRINTED) ARCH SURG/ VOL 137, JAN 2002 WWW.ARCHSURG.COM 73