Long-term results (> 12 years) of surgical treatment of perihilar cholangiocarcinoma: Application of a modified Nagoya approach at a western center Authors: İlgin Özden, Orhan Bilge, Nergiz Dağoğlu, Ayşen Yavru, Mine Güllüoğlu, Arzu Poyanlı, Yaman Tekant, Ahmet Kızir Published: 2026 ABSTRACT PurposeTo analyze the results of a modified Nagoya approach in perihilar cholangiocarcinoma at a western center.MethodsThe first author trained for one year at Nagoya University. The records of the 53 patients who underwent major hepatobiliary resection between 2001 and 2012 were reviewed. All fundamental principles were strictly applied with following modifications: hepatoduodenal ligament dissection limited to the level cranial to the gastroduodenal artery; 8a-p lymph nodes removed only if macroscopically positive; adjuvant radiotherapy, chemoradiotherapy (CRT) and CRT followed by chemotherapy in selected patients starting in 2007,2009 and 2011 respectively. The results were compared with the 2021 international benchmark values.ResultsTwenty-seven patients were men; median (range) age was 54 (29–74). Percutaneous biliary drainage was performed in 45 patients (85%) and portal vein embolization in 8 (15%). Portal vein resection was required in 10 patients (19%). The 90-day mortality was 4% (2/53). Perihepatic fluid infection without an obvious biliary leak was the most frequent complication (42%). The frequencies of liver failure B-C and bile leak were 8% and 12% respectively. The calculated 1,3,5 and 10 -year survival rates were 87%,62%, 41% and 25% respectively. Thirteen patients (25%) survived longer than 10 years: seven had T1 and six had T2a tumors; none had a poorly differentiated tumor, positive surgical margins or lymph node metastasis.ConclusionsThe modified Nagoya approach yielded outcomes similar to 25th percentile values for 90-day mortality, postoperative liver failure B-C and bile leak and benchmark values for 5-year overall survival. Introduction Major hepatobiliary resection with caudate lobectomy is the primary radical treatment of perihilar cholangiocarcinoma, except for tumors in the setting of primary sclerosing cholangitis [1, 2]. The first radical procedures were very occasionally performed in the 1950–1970’s [3]; the foundations of the standard approach were established in the 1980 s [3, 4]. In the 1990 s, this principle had been applied in many Far Eastern [5, 8], particularly Japanese centers [5–7] and in some Western centers [9–11]. The first author (İÖ) visited Nagoya University, First Department of Surgery in 2000–2001 for one year under the supervision of Prof. Yuji Nimura. After his return, a modified version of the meticulous clinical bundle of the Nagoya University was applied at Istanbul University. In this paper, the results of this modified Nagoya approach at a western center were reported, based on the outcome of patients who underwent surgery between 2001 and 2012 and followed for at least 12 years. Materials and methods The records of the 53 patients who underwent major hepatobiliary resection between 2001 and 2012 were reviewed. The clinical bundle summarized in Table 1 was applied. The scheme on the practice at the Nagoya University reflects the approach in 2000–2001 when the first author received training. Of course, significant changes occurred over the last 2 decades [1, 2]: Table 1Clinical bundle for perihilar cholangiocarcinoma at Nagoya University (N) in 2000–2001 and the modifications at İstanbul University (İ) 1-N: The concept of perihilar cholangiocarcinoma endorsed İ:Same 2-N: Computed tomography(CT) was the primary staging method. İ: Both CT and magnetic resonance imaging were used. 3-N: All patients considered for major hepatobiliary resection with total caudate lobectomy İ: Same. 4-N: PTBD in jaundiced patients; enteral bile replacement and UDCA İ:Same 5-N: Catheter angiography for staging/PVE for prevention of postoperative liver failure İ:Initially same; CTangiography replaced catheter angiography over time 6-N: ICG clearance test and measurement of biliary bilirubin excretion from the future remnant liver for assessment of liver function I:Same7-N: Lymphadenectomy starting from superior group of the retropancreatic nodes (13a), including 8a and 8p; paraaortic dissection if 8a positiveİ:Limited to the cranial part of the GDA (12 a1, 12a2,12b1,12 b2, 12 h, 12p1,12p2);8a-8p removed if macroscopically positive 8-N: Selective approach to patients with bilateral arterial involvement from 1998 to 2004; standard indication afterwards İ:Patients with bilateral involvement considered inoperable until 2014. 9-N: No adjuvant treatment İ:Adjuvant radiotherapy, chemoradiotherapy (CRT) and CRT followed by chemotherapy in selected patients starting in 2007,2009 and 2011 respectively 10-N: Persevere unless the tumor is clearly inoperable. İ:Same.N: Nagoya İ: İstanbul; PTBD (percutaneous transhepatic biliary drainage); UDCA: ursodeoxycholic acid, PVE: portal vein embolization; ICG: indocyanine green; GDA: gastroduodenal artery Table 1Clinical bundle for perihilar cholangiocarcinoma at Nagoya University (N) in 2000–2001 and the modifications at İstanbul University (İ) 1-N: The concept of perihilar cholangiocarcinoma endorsed İ:Same 2-N: Computed tomography(CT) was the primary staging method. İ: Both CT and magnetic resonance imaging were used. 3-N: All patients considered for major hepatobiliary resection with total caudate lobectomy İ: Same. 4-N: PTBD in jaundiced patients; enteral bile replacement and UDCA İ:Same 5-N: Catheter angiography for staging/PVE for prevention of postoperative liver failure İ:Initially same; CTangiography replaced catheter angiography over time 6-N: ICG clearance test and measurement of biliary bilirubin excretion from the future remnant liver for assessment of liver function I:Same7-N: Lymphadenectomy starting from superior group of the retropancreatic nodes (13a), including 8a and 8p; paraaortic dissection if 8a positiveİ:Limited to the cranial part of the GDA (12 a1, 12a2,12b1,12 b2, 12 h, 12p1,12p2);8a-8p removed if macroscopically positive 8-N: Selective approach to patients with bilateral arterial involvement from 1998 to 2004; standard indication afterwards İ:Patients with bilateral involvement considered inoperable until 2014. 9-N: No adjuvant treatment İ:Adjuvant radiotherapy, chemoradiotherapy (CRT) and CRT followed by chemotherapy in selected patients starting in 2007,2009 and 2011 respectively 10-N: Persevere unless the tumor is clearly inoperable. İ:Same.N: Nagoya İ: İstanbul; PTBD (percutaneous transhepatic biliary drainage); UDCA: ursodeoxycholic acid, PVE: portal vein embolization; ICG: indocyanine green; GDA: gastroduodenal artery Clinical bundle for perihilar cholangiocarcinoma at Nagoya University (N) in 2000–2001 and the modifications at İstanbul University (İ) N: Nagoya İ: İstanbul; PTBD (percutaneous transhepatic biliary drainage); UDCA: ursodeoxycholic acid, PVE: portal vein embolization; ICG: indocyanine green; GDA: gastroduodenal artery The concept of perihilar cholangiocarcinoma was endorsed [12].After history, physical examination and ultrasonography, computed tomography was used as the principal staging examination at Nagoya University; magnetic resonance imaging (MRI) was not performed routinely. Both were used at Istanbul University.All patients were considered for major hepatobiliary resection with total caudate lobectomy.At Nagoya University, for jaundiced patients (total bilirubin > 2 mg/dL) without cholangitis, there was a shift from bilateral to unilateral (future liver remnant) percutaneous drainage of the biliary tree. Complete biliary drainage was attempted in patients with cholangitis. Initially, bilateral drainage was performed at İstanbul University. The shift was eventually adopted. Oral bile replacement and ursodeoxycholic acid were given to decrease bacterial translocation and improve liver function [1].At Nagoya University, surgeon-performed catheter angiography was used for staging. Also, surgeon-performed percutaneous portal vein embolization was used to prevent of postoperative liver failure in patients who had a future remnant liver smaller than 40%. At Istanbul University, these procedures were performed by interventional radiologists. As was the case all over the world, CT-angiography replaced catheter angiography for diagnostic purposes during the study period.The indocyanine green (ICG) clearance test was used to assess liver function [1]. Also, bilirubin level of the bile from the future remnant liver was measured.At Nagoya University, extensive lymphadenectomy of the hepatoduodenal ligament starting from the superior group of the retropancreatic nodes (13a) was performed routinely [13]. Macroscopic involvement of the paraaortic lymph nodes was not an absolute criterion for inoperability. Paraaortic lymph node dissection was performed if the 8a lymph node was positive for cancer on microscopy. At Istanbul University, paraaortic lymph node involvement was a criterion for inoperability. Hepatoduodenal ligament dissection was limited to the level cranial to the gastroduodenal artery (GDA). 8a and 8p lymph nodes were removed if they were macroscopically positive (Fig. 1).At Nagoya University, the first hepatobiliary resection with combined resection-reconstruction of the portal vein and hepatic artery was performed in March 1998 (Prof. Yuji Nimura, personal communication). Patients with bilateral arterial involvement were approached on a case-by-case basis in 2000–2001; it became a part of standard practice in 2004 [1]. At İstanbul University, these patients were considered inoperable until 2014, when the first hepatic artery resection and reconstruction were performed.No adjuvant treatment was recommended at Nagoya University in 2000–2001. In İstanbul, for patients with R1/R2 resection, lymphovascular/perineural invasion and lymph node metastasis, intensity modulated radiotherapy (IMRT) was introduced in 2007: 45 Gy in 25 fractions were used. Also, concurrent 5-fluorouracil (5-FU) was added in 2009 to perform chemoradiotherapy (CRT); 5-FU and gemcitabine were given following CRT from 2011.At Nagoya University, vigorous effort was spent to provide a safe resection to all patients. A similar approach was taken at İstanbul University. For example, a patient who had undergone endoscopic stenting at another institution, attended İstanbul University for a second opinion but declined surgery. He returned 7 months later with cholangitis. The endoscopic stent was removed and bilateral PTBD was performed. Angiography showed possible involvement of the left hepatic artery. Left hepatectomy+caudate lobectomy and Roux-Y cholangiojejunostomy were performed 9 months after the first endoscopic stenting. The surgical margins were negative and there was no lymph node metastasis. He died of tumor recurrence 9 years and 5 months later.With few exceptions, the first author was either the operator or first assistant to more senior surgeons in all operations. The concept of perihilar cholangiocarcinoma was endorsed [12]. The concept of perihilar cholangiocarcinoma was endorsed [12]. After history, physical examination and ultrasonography, computed tomography was used as the principal staging examination at Nagoya University; magnetic resonance imaging (MRI) was not performed routinely. Both were used at Istanbul University. After history, physical examination and ultrasonography, computed tomography was used as the principal staging examination at Nagoya University; magnetic resonance imaging (MRI) was not performed routinely. Both were used at Istanbul University. All patients were considered for major hepatobiliary resection with total caudate lobectomy. All patients were considered for major hepatobiliary resection with total caudate lobectomy. At Nagoya University, for jaundiced patients (total bilirubin > 2 mg/dL) without cholangitis, there was a shift from bilateral to unilateral (future liver remnant) percutaneous drainage of the biliary tree. Complete biliary drainage was attempted in patients with cholangitis. Initially, bilateral drainage was performed at İstanbul University. The shift was eventually adopted. Oral bile replacement and ursodeoxycholic acid were given to decrease bacterial translocation and improve liver function [1]. At Nagoya University, for jaundiced patients (total bilirubin > 2 mg/dL) without cholangitis, there was a shift from bilateral to unilateral (future liver remnant) percutaneous drainage of the biliary tree. Complete biliary drainage was attempted in patients with cholangitis. Initially, bilateral drainage was performed at İstanbul University. The shift was eventually adopted. Oral bile replacement and ursodeoxycholic acid were given to decrease bacterial translocation and improve liver function [1]. At Nagoya University, surgeon-performed catheter angiography was used for staging. Also, surgeon-performed percutaneous portal vein embolization was used to prevent of postoperative liver failure in patients who had a future remnant liver smaller than 40%. At Istanbul University, these procedures were performed by interventional radiologists. As was the case all over the world, CT-angiography replaced catheter angiography for diagnostic purposes during the study period. At Nagoya University, surgeon-performed catheter angiography was used for staging. Also, surgeon-performed percutaneous portal vein embolization was used to prevent of postoperative liver failure in patients who had a future remnant liver smaller than 40%. At Istanbul University, these procedures were performed by interventional radiologists. As was the case all over the world, CT-angiography replaced catheter angiography for diagnostic purposes during the study period. The indocyanine green (ICG) clearance test was used to assess liver function [1]. Also, bilirubin level of the bile from the future remnant liver was measured. The indocyanine green (ICG) clearance test was used to assess liver function [1]. Also, bilirubin level of the bile from the future remnant liver was measured. At Nagoya University, extensive lymphadenectomy of the hepatoduodenal ligament starting from the superior group of the retropancreatic nodes (13a) was performed routinely [13]. Macroscopic involvement of the paraaortic lymph nodes was not an absolute criterion for inoperability. Paraaortic lymph node dissection was performed if the 8a lymph node was positive for cancer on microscopy. At Istanbul University, paraaortic lymph node involvement was a criterion for inoperability. Hepatoduodenal ligament dissection was limited to the level cranial to the gastroduodenal artery (GDA). 8a and 8p lymph nodes were removed if they were macroscopically positive (Fig. 1). At Nagoya University, extensive lymphadenectomy of the hepatoduodenal ligament starting from the superior group of the retropancreatic nodes (13a) was performed routinely [13]. Macroscopic involvement of the paraaortic lymph nodes was not an absolute criterion for inoperability. Paraaortic lymph node dissection was performed if the 8a lymph node was positive for cancer on microscopy. At Istanbul University, paraaortic lymph node involvement was a criterion for inoperability. Hepatoduodenal ligament dissection was limited to the level cranial to the gastroduodenal artery (GDA). 8a and 8p lymph nodes were removed if they were macroscopically positive (Fig. 1). At Nagoya University, the first hepatobiliary resection with combined resection-reconstruction of the portal vein and hepatic artery was performed in March 1998 (Prof. Yuji Nimura, personal communication). Patients with bilateral arterial involvement were approached on a case-by-case basis in 2000–2001; it became a part of standard practice in 2004 [1]. At İstanbul University, these patients were considered inoperable until 2014, when the first hepatic artery resection and reconstruction were performed. At Nagoya University, the first hepatobiliary resection with combined resection-reconstruction of the portal vein and hepatic artery was performed in March 1998 (Prof. Yuji Nimura, personal communication). Patients with bilateral arterial involvement were approached on a case-by-case basis in 2000–2001; it became a part of standard practice in 2004 [1]. At İstanbul University, these patients were considered inoperable until 2014, when the first hepatic artery resection and reconstruction were performed. No adjuvant treatment was recommended at Nagoya University in 2000–2001. In İstanbul, for patients with R1/R2 resection, lymphovascular/perineural invasion and lymph node metastasis, intensity modulated radiotherapy (IMRT) was introduced in 2007: 45 Gy in 25 fractions were used. Also, concurrent 5-fluorouracil (5-FU) was added in 2009 to perform chemoradiotherapy (CRT); 5-FU and gemcitabine were given following CRT from 2011. No adjuvant treatment was recommended at Nagoya University in 2000–2001. In İstanbul, for patients with R1/R2 resection, lymphovascular/perineural invasion and lymph node metastasis, intensity modulated radiotherapy (IMRT) was introduced in 2007: 45 Gy in 25 fractions were used. Also, concurrent 5-fluorouracil (5-FU) was added in 2009 to perform chemoradiotherapy (CRT); 5-FU and gemcitabine were given following CRT from 2011. At Nagoya University, vigorous effort was spent to provide a safe resection to all patients. A similar approach was taken at İstanbul University. For example, a patient who had undergone endoscopic stenting at another institution, attended İstanbul University for a second opinion but declined surgery. He returned 7 months later with cholangitis. The endoscopic stent was removed and bilateral PTBD was performed. Angiography showed possible involvement of the left hepatic artery. Left hepatectomy+caudate lobectomy and Roux-Y cholangiojejunostomy were performed 9 months after the first endoscopic stenting. The surgical margins were negative and there was no lymph node metastasis. He died of tumor recurrence 9 years and 5 months later.With few exceptions, the first author was either the operator or first assistant to more senior surgeons in all operations. At Nagoya University, vigorous effort was spent to provide a safe resection to all patients. A similar approach was taken at İstanbul University. For example, a patient who had undergone endoscopic stenting at another institution, attended İstanbul University for a second opinion but declined surgery. He returned 7 months later with cholangitis. The endoscopic stent was removed and bilateral PTBD was performed. Angiography showed possible involvement of the left hepatic artery. Left hepatectomy+caudate lobectomy and Roux-Y cholangiojejunostomy were performed 9 months after the first endoscopic stenting. The surgical margins were negative and there was no lymph node metastasis. He died of tumor recurrence 9 years and 5 months later. With few exceptions, the first author was either the operator or first assistant to more senior surgeons in all operations. Fig. 1The hepatoduodenal ligament after right trisectionectomy (H145678-B) The hepatoduodenal ligament after right trisectionectomy (H145678-B) Definitions The liver resections were reported according to standard and ‘New World’ terminology [14].Post-hepatectomy liver failure was diagnosed and graded according to the International Study Group of Liver Surgery (ISGLS) criteria [15]. Bile leaks were evaluated by inspection rather than routine drain fluid bilirubin measurement but graded according to ISGLS criteria [16]. Complications were graded according to the Clavien-Dindo classification [17]. Follow up The patients were followed with standard blood tests, tumor markers (carcinoembryonic antigen and CA 19 − 9) and imaging studies (CT or MRI) every 4 months in the first year, every 6 months until the end of five years, annually until 10 years and every two years afterwards. Survival analysis Survival rates were calculated according to the Kaplan-Meier method. Frequencies were compared with Fischer’s exact test. Comparison The results obtained in patients treated between 2001 and 2012 were compared with the international benchmark values published in 2021 [18]. Results During the study period, 147 patients were evaluated as candidates for possible major hepatobiliary resection: the tumors of 27 patients were considered to be inoperable based on radiologic findings; 23 patients were considered as unfit for surgery due to a combination of factors such as age (> 75), accompanying comorbidities and poor liver function; 7 patients refused surgery or decided to seek treatment elsewhere. Laparotomy was performed in 90 patients and the tumor was considered to be inoperable due to the following reasons in 37 patients: peritoneal implants or gross paraaortic lymph node metastases (19), very extensive tumor extension in the hepatoduodenal ligament and the around the pedicle of the future remnant liver (14), liver metastases (4). Eventually, major hepatobiliary resection could be performed in 53 patients. Twenty-seven patients were men and 26 were women; median (range) age 54 (29–74). The Bismuth classification of the tumors were as follows: I: 2(4%) II:15 (28%), IIIa:15 (28%), IIIb: 13(25%) and IV: 8 (15%). PTBD was performed in 45 patients (85%) and portal vein branch embolization in 8 (15%). Right hepatectomy (H15678-B) was performed in 27 (51%) patients, left hepatectomy (H1234-B) in 25 (47%) and right trisectionectomy (H145678-B) in one. Portal vein resection was performed in 10 patients (19%): resection and end-to-end anastomosis in six patients, wedge resection and reconstruction in four. The 90-day mortality was 4% (2/53). In one patient, unexplained septic shock, possibly due to cholangitis, developed during parenchymal dissection following right hepatic artery ligation, resection and reconstruction of the portal vein and division of the right hepatic vein; the operation had to be completed under catecholamine support. He died on the following day. The other patient developed sepsis on the 12 th postoperative day on the background of grade B post-hepatectomy liver failure (INR 1.65, total bilirubin 5.6 mg/dL); emergency laparotomy revealed markedly dilated intestinal loops in the absence of obstruction; a minimal amount of infected fluid (no biliary fistula) with the intraoperatively placed drain in place was found in the right subdiaphragmatic space only; the abdomen was closed with a Bogota bag. Despite appropriate antibiotic treatment and re-exploration 3 days later, he died on the 17th postoperative day. There was no surgical mortality between 2008 and 2012. The resection was evaluated as R0 in 37 patients (70%), R1 in 14 patients (26%) and R2 in 2 patients (4%). The Clavien-Dindo II-IVb complications in the first 90 days were listed in Table 2; they were graded as II in 30 patients, IIIa in 11, grade IIIb in 5 and IVa in 6. Infection of the perihepatic fluid in the absence of an obvious biliary leak was the most frequent problem and was treated with antibiotics. Bile leaks were graded as A-B in five patients (none required percutaneous drainage, all healed with gradual drain removal) and C in one patient (evisceration leading to laparotomy: localized biloma adjacent to the biliodigestive anastomosis). Cholangitis attacks resolved after removal of the intraoperatively placed transjejunal biliary drainage catheters in one patient. Postoperative imaging in the other patient showed that the B5 and B8 had been misidentified as the right anterior and posterior sectional duct. PTBD for cholangitis and surgical revision of the biliodigestive anastomosis were performed. The patient with postoperative portal vein thrombosis had developed intraoperative thrombosis after portal vein resection and reconstruction; thrombectomy and reanastomosis had been performed. Asymptomatic thrombosis extending from the intrahepatic portal branches to the splenomesenteric junction developed on the third postoperative day; the lactate levels remained at the upper level of the normal range and there was no deterioration in standard biochemistry results. Her condition was managed with high-dose low-molecular-weight heparin. She survived one year and needed chronic diuretic treatment for ascites tendency. All surviving patients except one (lost to follow up at 7 months) were followed for at least 12 years. Contact was lost with one patient who had been alive at 172 months after surgery. Table 2Early Clavien-Dindo II-IVb complications (out of 53 − 1 = 52 patients)Intraabdominal fluid infection22(42 %)Collection requiring percutaneous/surgical drainage5/2(13 %)Biliary leak6(12 %)Postoperative liver dysfunction grade B-C4(8 %)Pleurisy requiring tube thoracostomy/empyemaa3/1(8 %)Evisceration/evantration1/2(6 %)Cholangitis2(4 %)Portal vein thrombosis1(2 %)Hepatic artery pseudoaneurysmb1(2 %)Postoperative delirium1(2 %)Vagal hypotension1(2 %)Thirty-nine patients (75%) experienced Clavien II-IVb complicationsaThe empyema was treated by thoracoscopic drainage and decorticationbTreated by angiographic coil embolization Table 2Early Clavien-Dindo II-IVb complications (out of 53 − 1 = 52 patients)Intraabdominal fluid infection22(42 %)Collection requiring percutaneous/surgical drainage5/2(13 %)Biliary leak6(12 %)Postoperative liver dysfunction grade B-C4(8 %)Pleurisy requiring tube thoracostomy/empyemaa3/1(8 %)Evisceration/evantration1/2(6 %)Cholangitis2(4 %)Portal vein thrombosis1(2 %)Hepatic artery pseudoaneurysmb1(2 %)Postoperative delirium1(2 %)Vagal hypotension1(2 %)Thirty-nine patients (75%) experienced Clavien II-IVb complicationsaThe empyema was treated by thoracoscopic drainage and decorticationbTreated by angiographic coil embolization Early Clavien-Dindo II-IVb complications (out of 53 − 1 = 52 patients) Thirty-nine patients (75%) experienced Clavien II-IVb complications aThe empyema was treated by thoracoscopic drainage and decortication bTreated by angiographic coil embolization Major long-term complications were PTBD tract recurrences (n:2; 5% =2/43), biliary anastomotic strictures, (n:2; 4% =2/51) and late portal vein stenosis presenting with variceal hemorrhage (n:2; 4% =2/51). One of the patients with PTBD tract recurrence was lost to follow up. The other patient underwent resection and reconstruction of the thoracic wall at the end of the second postoperative year. Despite radiotherapy, he died of disseminated tumor recurrence 17 months later. This is the only patient who was eligible for resection of a recurrent tumor. Late, nonmalignant biliary strictures (diagnosed at 52 and 108 months) were treated by PTBD and balloon dilation. The two patients who developed portal vein stenosis presenting with variceal hemorrhage at 22 and 192 months were treated successfully with portal vein stenting and anticoagulant treatment. For the entire cohort, 1,3,5 and 10-year overall survival rates were 87%, 62%, 41% and 25% respectively (Fig. 2). The corresponding recurrence-free survival rates were 77%, 45%, 37% and 20% respectively (Fig. 3). Median overall survival was 48 months (95% confidence interval: 36–60). All deaths except five were due to cholangiocarcinoma recurrence: aggressive rectal adenocarcinoma (1), pneumonia (1), suspected thromboembolism after removal of the chemotherapy port (1), abscess in the left lung (1) and unknown cause (1). Fig. 2Kaplan-Meier survival graph of the entire cohort, including hospital deaths Kaplan-Meier survival graph of the entire cohort, including hospital deaths Fig. 3Kaplan-Meier survival graph of recurrence-free survival in patients who could be discharged from hospital Kaplan-Meier survival graph of recurrence-free survival in patients who could be discharged from hospital Cholangiocarcinoma recurred in 7 patients after 5 years and caused death in all but one (the patient with suspected thromboembolism). Thirteen patients (25%) survived longer than 10 years (Table 3); 9 were alive at last follow up. Of the four cases who had lymphovascular and/or perineural invasion, one surviving patient had received adjuvant RT, one surviving and one non-surviving had received adjuvant CRT. The other patient had undergone surgery before 2007. Comparison of the surviving patients (no recurrence) and patients who developed recurrence after 5 years was presented in Table 4. The frequency of negative prognostic factors was significantly higher in the recurrence group (2/9 vs. 7/7; p < 0.01). Table 3Characteristics of the patients who survived longer than 10 yearsAlive (n:9)Dead (n:4)ªT stageT161 T2a33Negative prognostic factors Lymph node metastasis00 Lymphovascular and/or perineural invasion22 Positive surgical margin00 Poorly differentiated00aOne of these patients died at 159 months at another institution. She had chosen to be followed at another hospital in the last 5 years. Information on the cause of death could not be obtained. The other patients died of cancer recurrence Table 3Characteristics of the patients who survived longer than 10 yearsAlive (n:9)Dead (n:4)ªT stageT161 T2a33Negative prognostic factors Lymph node metastasis00 Lymphovascular and/or perineural invasion22 Positive surgical margin00 Poorly differentiated00aOne of these patients died at 159 months at another institution. She had chosen to be followed at another hospital in the last 5 years. Information on the cause of death could not be obtained. The other patients died of cancer recurrence Characteristics of the patients who survived longer than 10 years aOne of these patients died at 159 months at another institution. She had chosen to be followed at another hospital in the last 5 years. Information on the cause of death could not be obtained. The other patients died of cancer recurrence Of note, two of the > 10-year survivors, developed nonresectable intraabdominal recurrences that showed indolent courses. In one patient with a T2a tumor and perineural invasion, a recurrence was suspected at 112 months based on an increase of the Ca 19 − 9 level to 99 U/mL (5–85) and a soft tissue mass surrounding the pedicle of the remnant liver. The mass showed no significant uptake on FDG PET; a retrospective inspection of the films taken on the 4th and 5th postoperative years revealed that the mass had been stable over the years and had been interpreted as postoperative inflammatory change. Fine-needle aspiration biopsy yielded insufficient material. The GGT level was within normal limits. Because an effective treatment could not be offered (the lesion was technically unresectable and the lack of objective growth limited the potential benefit of chemotherapy). The Ca 19 − 9 levels of the patient increased steadily to 394 U/mL at 166 months; new findings were dilation of the end of the jejunal loop and a further increased GGT level (564 U/L). Transhepatic drainage of the jejunal loop decreased GGT to 201 U/L. Laparotomy was performed to resolve the intriguing course; a solitary recurrence surrounding the pedicle and invading the jejunal loop was observed; frozen section examination demonstrated adenocarcinoma and paraffin-section examination revealed histopathological findings compatible with recurrence of the original tumor. The patient received palliative radiotherapy and chemotherapy but died at 207 months. The other patient also had a T2a tumor with lymphovascular and perineural invasion. He developed multiple peritoneal metastases at 54 months, (Ca 19 − 9: 489 U/mL) showed partial compliance to chemotherapy and died at 138 months. None of the patients developed a recurrence after 10 years. Table 4Comparison of the patients who survived recurrence-free for longer than 10 years and patients whose tumors recurred after 5 yearsNo recurrence (n:9)Recurrence after 5y (n:7)T stage T161 T2a34 T2b−1 T3−1Negative prognostic factors Lymph node metastasis0/92/7 Lymphovascular and/or perineural invasion2/97/7 Positive surgical margin0/91/7 Poorly differentiateda0/92/7aEither poorly differentiated or showing areas of poor differentiation Table 4Comparison of the patients who survived recurrence-free for longer than 10 years and patients whose tumors recurred after 5 yearsNo recurrence (n:9)Recurrence after 5y (n:7)T stage T161 T2a34 T2b−1 T3−1Negative prognostic factors Lymph node metastasis0/92/7 Lymphovascular and/or perineural invasion2/97/7 Positive surgical margin0/91/7 Poorly differentiateda0/92/7aEither poorly differentiated or showing areas of poor differentiation Comparison of the patients who survived recurrence-free for longer than 10 years and patients whose tumors recurred after 5 years aEither poorly differentiated or showing areas of poor differentiation Discussion Extensive effort has been spent to improve the safety and radicality of major hepatobiliary resection for perihilar cholangiocarcinoma. Even series including 40, 58 and 102 patients without mortality have been reported [19–21]. Still, “Despite ongoing efforts, postoperative mortality and morbidity remains high after complex liver surgery for PHC” as reported in the 2021 benchmark values study based on the data of patients treated between 2014 and 2018 at 24 high-volume centers (> 50 resections during the study period) for tumors “without the need for vascular reconstruction due to tumor invasion”) [18]. One of the important results is that “Centers from Asia disclosed better outcomes.” [18]. The present report shows that application of the modified Nagoya approach yields a successful outcome, based on multiple parameters, at a western center- an original aspect that has not been clearly established before. Of note, the results of the present study (2001–2012) and the 25th percentile values for 90-day mortality (4% vs 3.7%), postoperative liver failure B-C(8% vs 8%), bile leak (12% vs 33.7%) were comparable. Although the mortality was acceptable, the complication rate was still high: 39 patients experienced (75%) Clavien II-IVB complications (Table 2); the 25th percentile frequencies for any complication and Clavien-Dindo Grade ≥3a complications are 76.6% and 56.8% respectively. As to oncologic outcomes, the results for R0-R1-R2 resection rates (70%, 26% and 4% respectively) were similar to the benchmark values of ≥ 56.7%, < 43% ≤ 0.8% respectively. The 1-, 3- and 5 year overall survival figures of 87%, 62% and 41% were similar to the benchmark values of ≥ 77.5%, ≥ 48.4% and ≥ 39.7% respectively. Although a statistical comparison is not possible, the 1-, 3- and 5 year disease-free survival figures of 77%, 45% and 37% were better than the benchmark values of ≥ 33.3%, ≥ 0% and ≥ 0% respectively. Nakahashi et al. have shown that perihilar cholangiocarcinoma may recur after 5 years following an R0 resection and recommended close surveillance for 10 years [22]. Notably, in their series, only two of the 242 patients with recurrence, were diagnosed after this threshold: at 10.1 and 11.3 years. To the best of our information, there are only two reports on actual 10-year survivors after major hepatobiliary resection for perihilar cholangiocarcinoma [1, 23]. Also, Rassam et al. have published in 2018, the results of the patients who underwent surgery between 2000 and 2016 and provided data on patients at risk at 10 years on the survival curve [24]. First, in the single-center study by Nagino et al., published in 2013, two time periods (1977–2000 vs. 2001–2010) were compared. For the period 1977–2000, 25 actual 10-year survivors out of 188 patients (13% actual survival) were reported [1]. The second paper is by van Keulen et al. on 406 patients treated in 22 centers between 2000 and 2009, 29 (6%) actual 10-year survivors were reported [23]. In the third paper – a single-center report by Rassam et al. – the survival curve shows that 18 out of 170 patients (11%) have reached the 10-year mark, but the real figure for the series will be higher since some patients who underwent surgery after 2008 will be added [24]. In the paper by Nagino et al., the Kaplan-Meier-calculated 10-year survival rates of the two time periods were 14% (1977–2000) and 21% (2001–2010) respectively [1]. The roughly corresponds to the differences in surgical mortality (19/188, 10% vs. 8/386, 2%) between the two intervals. This is of course, an oversimplification. More extensive procedures were performed on more advanced tumors in the latter period. Still, the parallel course of the calculated survival curves suggests that in patients who have undergone safe surgery according to oncologic principles, long-term survival is determined primarily by tumor biology. In a 2025 paper from the same institution [25], on patients who underwent surgery between 2002 and 2019, the survival curve shows that 113 out of 681 (17%) patients have reached the 10-year mark but the real figure for the series will be higher since some patients who underwent surgery after 2015 will be added. The 25% actual 10-year survival rate in the present study may be partially explained by the frequency of patients with favorable tumor characteristics as well as the surgical mortality of 2/53 (4%). The hospital mortality in the paper by van Keulen et al. was not specifically stated but the survival curves suggest a figure of 9–10%; age above 70, Bismuth-Corlette type IV tumors, hepatic artery reconstruction, positive resection margins, positive lymph nodes and poor tumor differentiation were found to render cure unlikely [23]. The 90-day mortality of major hepatobiliary resection was 9% in the study by Rassam et al. [24]. In other words, decreasing surgical mortality and performing R0 resection whenever possible seem to be the only two factors within the surgical team’s control to achieve actual survivorship since even carcinoma in situ residue at the surgical margin is detrimental to long-term survival although not markedly as microscopic invasive carcinoma [25]. Also, as shown by the experience reported here, secondary liver failure [26], may be avoided by vigilant monitoring and treatment of perihepatic fluid infection. Resection of operable recurrent tumors is unfortunately possible in less than 10% of the cases [22]; it could be performed in only one patient in the present series. Recurrent tumors are usually disseminated and aggressive [22]. However, the two 10-year survivors in this report have shown that an indolent course is possible in recurrent tumors as well: recurrence at 112 and 58 months and death at 207 and 138 months respectively. Similarly, Okabe et al. reported a patient whose local recurrence at one year after R1 resection could be treated for more than 7 years after the relapse [27]. Long-term follow up by an experienced team is vital as major complications may occur as late as 192 months after surgery. One important limitation of the clinical bundle is that it reflects the 2000–2001 Nagoya practice which was developed further in the subsequent years [1, 2]. This is inevitable in a paper on > 12-year actual survivors. Still, the patients did not benefit from the developments in the subsequent years. Preoperative symbiotic treatment [1] could have served to decrease bacterial translocation and possibly the risk of perihepatic fluid infection. The PTBD tract recurrences may have been prevented by preferring the endoscopic route [1, 2]. An active prehabilitation program could have contributed to enhanced recovery [2]. Another limitation is that this report reflects both the initiation (‘the learning curve’) and establishment of a perihilar cholangiocarcinoma surgery program (no surgical mortality between 2008 and 2012) across 12 years at a rate of 4–5 resections per year. Also, adjuvant treatments were introduced in stepwise manner in the second half of the study period. On the other hand, the results still have implications for current practice according to the Milan Jury-Based Consensus of 2025 [28]: 1-Probiotic treatment was not included and adjuvant radiotherapy was not recommended. To put it in another way, the possible positive contributions of these measures were considered as ‘to be proven’. 2-The following statements in the document reflect that the clinical bundle of this study is within the spectrum of international practice in 2025. “International guidelines recommend EBD as the primary modality76 or both approaches equally77 based on larger scale retrospective evidence78,79, but PBD remains the preferred approach in the majority of centers9.” “In patients with resected pCCA, a six-month course of adjuvant fluoropyrimidine therapy (capecitabine or S1) is recommended.” “The role of perioperative (neoadjuvant or adjuvant) therapy with targeted or immunotherapy remains investigational. Participation in clinical trials is suggested.” Furthermore, adjuvant treatments in individual centers were not considered in the development of 1-,3- and 5-year survival standards in the 2021 benchmark study [18], which suggests that their inevitable heterogeneity among 24 centers has not been considered to be critical.The present results compare favorably with the benchmark values derived from the data of the leading centers [18]. Finally, an important contribution is the practice of vigilant monitoring of perihepatic fluid infection in the absence of a bile leak which increases the reported infection rate but contributes to the low mortality – a paradoxical situation, also encountered in our liver transplantation program [29]. Non-PSC-related perihilar cholangiocarcinoma surgery has probably approached its technical limits in dedicated centers in terms of extent of surgery and mortality [2, 19–21, 30–32]. There is still room to decrease the morbidity [18, 30]. The 25th percentile value for frequency of any complication is still 76.6% [18]. Five-year survival rates as high as 67% have been reported for R0,N0,M0 patients [1]. Further improvement may be possible with early diagnosis, identification of patients whose tumor biology may render upfront surgery potentially futile, neoadjuvant treatment in selected patients [2], expansion of knowledge on actionable mutations and better adjuvant treatments [2]. Conclusion The present report shows that application of the modified Nagoya approach in perihilar cholangiocarcinoma yields a successful outcome according to international benchmark criteria, at a western center. Surgery alone may achieve long-term survival-even cure- in patients with favorable tumor characteristics.