Perihilar cholangiocarcinoma: a surgeon's perspective

Abstract

Perihilar cholangiocarcinoma (PHC) is defined anatomically as tumors located in the extrahepatic biliary tree proximal to the origin of the cystic duct. However, as the boundary between the extrahepatic and intrahepatic bile ducts is not well defined, PHC potentially includes two types of tumors: one is the extrahepatic type, which arises from the large hilar bile duct, and the other is the intrahepatic type, which has an intrahepatic component with invasion of the hepatic hilus. The new American Joint Commission on Cancer tumor-node-metastasis staging system for PHC was well revised; an important revision is that the Bismuth type IV was removed from T4 stage determinants. Most patients with PHC present with obstructive jaundice, and preoperative biliary drainage is mandatory. Previously, percutaneous transhepatic biliary drainage was used in many centers; however, it has been accepted that endoscopic nasobiliary drainage is the most suitable method for PHC. Recently, inside-stents have also been adopted for biliary drainage, with favorable results. Portal vein embolization is widely performed as presurgical treatment for patients undergoing extended hepatectomy to minimize postoperative liver dysfunction. Surgical resection of PHC is technically demanding and continues to be the most difficult challenge for hepatobiliary surgeons. Because of advances in diagnostic and surgical techniques, surgical outcomes and survival rates after resection have steadily improved. However, survival rates, especially for patients with lymph node metastasis, are unsatisfactory, and establishing a protocol for effective adjuvant chemotherapy is an urgent task. Further synergy of endoscopists, radiologists, oncologists, and surgeons is required to conquer this intractable disease.

1. Introduction

Cholangiocarcinoma is a rare, enigmatic, and challenging cancer originating at any point along the biliary tract epithelium. This cancer is classified into three main categories: intrahepatic, perihilar, and distal, depending on the anatomical location [1]. Of these, perihilar cholangiocarcinoma (PHC) is the most common type of cholangiocarcinoma, accounting for 50% [2] to 67% [3], followed by the distal and then the intrahepatic forms.

PHC is a devastating aggressive disease that is most often diagnosed at advanced stage, resulting in poor prognosis. Surgical resection of PHC is technically demanding and continues to be the most difficult challenge for hepatobiliary surgeons. However, as surgical resection is the only way to cure the disease, many surgeons have adopted an aggressive approach to PHC. In the past two decades, with advances in diagnostic and surgical techniques, surgical outcomes and survival rates have gradually improved [[2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24]]. Since the late 1970s, Nimura (my mentor, emeritus professor of Nagoya University Graduate School of Medicine) [25] and I have made a consistent effort to treat this intractable disease with aggressive surgical strategies [[26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41]]. In this review article, I refer to my experience at Nagoya University Hospital and discuss the current surgical treatment of PHC.

2. Definition of PHC

PHC is defined anatomically as tumors located in the extrahepatic biliary tree proximal to the origin of the cystic duct [1]. On the other hand, intrahepatic cholangiocarcinoma (IHC), malignant intrahepatic epithelial neoplasm with biliary differentiation, arises in the liver peripheral/proximal to the left and right hepatic ducts [42]. These definitions are unclear and confusing, as the boundary between the extrahepatic and intrahepatic bile ducts is vague, without clear anatomical landmarks. According to the World Health Organization (WHO) definition (5th edition) [42], IHC has two main subtypes: large duct type and small duct type. The WHO textbook states that large duct type IHC arises in the large intrahepatic bile ducts near the hepatic hilus and resembles PHC [42]. Considering this, PHCs potentially include two types of tumors: one is the extrahepatic type, which arises from the large hilar bile duct, and the other is the intrahepatic type, which has an intrahepatic component with invasion of the hepatic hilus [43]. These two types of tumors exhibit similar features on cholangiogram and require the same surgical management; therefore, it is better to group them together under the term PHC [43,44].

In 1965, Klatskin first described in detail the clinical features of 13 patients with adenocarcinomas of the hepatic hilus [45]. It is noteworthy that three of the 13 patients had a bulky hard mass, measuring 5–10 cm in diameter, centered on the bifurcation of and extending deeply into the liver parenchyma. These three cases were all intrahepatic type PHC. After that, Okuda et al. studied autopsy cases of IHCs, concluding that the hilar-type IHCs resembled extrahepatic bile duct cancers [46]. Aishima et al., who analyzed 87 resected IHCs, demonstrated that the incidence of perineural invasion, lymph node metastasis, and vascular invasion was significantly higher in hilar-type IHC than it was in peripheral-type IHC. This aggressive nature of hilar-type IHC was more like that of extrahepatic cholangiocarcinoma [47].

In 1996, the Johns Hopkins group proposed a new classification of cholangiocarcinoma based on the surgery required for its treatment [3]. According to this classification, PHC was defined as cholangiocarcinoma involving or requiring resection of the hepatic duct confluence. On the other hand, IHC was defined as cholangiocarcinoma confined to the liver without invasion of the hepatic hilus or requiring only hepatectomy. Thus, cholangiocarcinomas with a significant intrahepatic component as well as the involvement of the confluence were included in the perihilar category rather than in the intrahepatic category.

Blechacz et al. [48] noted that the distinction between the intrahepatic and extrahepatic types is artificial and without either a biological or a clinical meaning. Currently, the definition developed by Ebata et al. in 2014 is clear and straightforward and should therefore be used. PHC is defined as cholangiocarcinomas involving the hilar bile duct (the duct located topologically between the right side of the umbilical portion of the left portal vein and the left side of the origin of the right posterior portal vein); for tumors with a significant liver mass component, the center of the mass has to be located between the abovementioned portal landmarks (Fig. 1A-D) [44].

Fig. 1.

Illustration showing the porta hepatis and definition of perihilar cholangiocarcinoma in a patient with a major hepatic mass. (A) The black dotted line indicates the hilar bile duct, defined as the duct located between the right side of the umbilical portion of the left portal vein and the left side of the origin of the right posterior portal vein. An intrahepatic mass lesion with hilar bile duct involvement is considered clinically to be a perihilar cholangiocarcinoma only when the center of the hepatic mass is located between the right side of the umbilical portion of the left portal vein and the left side of the origin of the right posterior portal vein (white dotted line). (B–D) Mass-forming intrahepatic cholangiocarcinoma (MF) often invades the hilar bile duct (small dots). This tumor is broadly classified as perihilar cholangiocarcinoma according to Nakeeb's definition, whereas a mass with a center remote from the hilar bile duct is categorized as intrahepatic, not perihialr, cholangiocarcinoma in the current definition. UP, umbilical portion; P, right posterior portal vein [44].

3. Staging system

The Bismuth-Corlette classification for PHC [49,50] is well known in the literature and widely used in clinical practice. Although often mistakenly understood, this classification is not a staging system, as mentioned by the authors, because of the absence of longitudinal description of cancer extent, no relation with prognostic data, and no clear definition of resectability criteria. The primary purpose of the Bismuth-Corlette classification is to serve as a guide for surgical strategy based on anatomical considerations.

The Blumgart T staging system [51], proposed in 2001 by Memorial Sloan-Kettering Cancer Center, is also used for local tumor assessment for PHC, primarily in the United States, and pertains to selecting patients for surgery. This staging scheme involves biliary invasion, portal vein invasion, and hepatic lobar atrophy as the T determinants to predict resectability and survival in the presurgical period. Tumors of the most advanced stage, T3 tumors, are considered to indicate local irresectability and reported to have a median survival time of only 8–11 months [51,52]. On the other hand, Ebata et al., who analyzed 729 patients with PHC, reported that resection was possible in more than 60% (269/444) of patients with T3 tumors, and the 5-year survival of 269 resected patients was 30.7% [53]. The Blumgart T stage was incorporated in the updated guidelines by the American Hepato-Pancreato-Biliary Association in 2015 [54], which still impacts global surgical practice for PHC. However, the Blumgart T staging system does not correspond to the current concepts of resectability. Although this simple staging system may be useful as a presurgical staging system, the unresectable classification of T3 tumors should be revised [53].

A new staging system for PHC [55], introduced by DeOliveira et al. in 2011, aims to standardize the reporting of this intractable disease. A registry based on this new system was already launched. Regrettably, this staging system has some flaws, as I previously noted [56]. First, this new system is not a staging system but rather merely descriptive, as the staging numbers in this system do not correspond to the extent of the severity of the tumor. Another problem involves the assessment of the extent of vessel invasion. The figures depicting the classification of the portal vein and hepatic artery invasions are inconsistent with real-life situations [55]. It is greatly desired that these concerns will be revised near future.

The American Joint Commission on Cancer (AJCC) Cancer Staging Manual includes a tumor-node-metastasis (TNM) staging classification for PHC, which is accepted worldwide to define prognosis. Since the seventh edition of the TNM classification, PHC has been recognized as a separate disease from distal cholangiocarcinoma. In 2017, the AJCC revised the manual and published the eighth edition of the TNM classification [57]. The most significant revision was that Bismuth type IV tumors were removed from the T4 determinants so that the T4 stage is no longer linked to Bismuth type IV PHC. This epoch-making revision is underlined by a multicenter study, which demonstrated no difference in survival among pM0 N0 patients with Bismuth type I/II, III, and IV tumors (with 5-year survival rates of 63.1%, 65.6%, and 59.2%, respectively) [44]. Yamada et al., who analyzed 702 resected patients with PHC, found that the new AJCC TNM system had the greatest discriminability for resectability and the largest C-index in survival, compared with those of the Blumgart T system and the Bismuth-Corlette classification [58]. They concluded that the new TNM system is the optimal classification system for predicting resectability and survival in PHC [58]. Another critical revision is that the N stage was changed from a region-based to a number-based concept, which was similar to other malignancies, including gastric cancer, colon cancer, or breast cancer. Specifically, nodal status was divided into three categories based on the number of involved nodes: pN0, pN1 (1–3 involved nodes), and pN2 (at least four involved nodes) [57].

4. Diagnosis

The first step of diagnosis for PHC is to perform a hematological examination and ultrasonography (US). Although elevation in the liver and biliary enzymes is often observed, these findings are not specific for PHC. Carbohydrate antigen (CA) 19-9 is elevated in 85% of patients with PHC, but it has variable sensitivity and specificity with low positive predictive value [59]. As jaundice is a main confounding factor for CA19-9, this tumor marker should be re-evaluated after biliary drainage [59]. US can detect tumors in more than 70% of patients with cholangiocarcinoma and thus should be performed as the first step for patients with suspected PHC [60].

Multidetector-row computed tomography (MDCT) with contrast enhancement is the most essential tool required in the second step to diagnose PHC [54,[61], [62], [63], [64], [65], [66]]. Computed tomography (CT) with an adequate pitch of detector rotation and table speed produces multiplanar reformation images. This helps to achieve a precise assessment of the degree of tumor extension [63] and vascular invasion [64,65]. Sugiura et al. evaluated how well MDCT could identify invasion of the portal bifurcation by PHC and demonstrated that the sensitivity, specificity, and overall accuracy for macroscopic portal vein invasion were 96.3%, 92.6%, and 94.0%, respectively [64]. Fukami et al. also assessed the diagnostic ability of MDCT for right hepatic artery invasion in a predominantly left-sided tumor [65], as this artery is often involved because of its anatomical aspect. No macroscopic evidence of right hepatic artery invasion was found in patients presenting visible low-density planes on multiplanar reformation images between the right hepatic artery and the adjacent tumor. In contrast, more than 70% of patients without visible low-density planes presented macroscopic evidence of right hepatic artery invasion and underwent combined arterial resection and reconstruction. The longitudinal extension of PHC is usually assessed by either endoscopic retrograde or percutaneous transhepatic cholangiography. Although these direct images are the gold standard for diagnosis, MDCT is also useful for this purpose [61,63,66]. Senda et al. reported that an accurate assessment of a tumor extent on both the proximal and distal borders was possible in approximately 80% of patients with hyperattenuated PHC [66]. Importantly, MDCT should be performed before biliary drainage, as the presence of the drainage catheter markedly hampers an accurate evaluation of the bile duct wall thickness, which is an essential finding for the longitudinal tumor extension, because of inflammatory changes in the bile duct. On the other hand, the diagnostic ability of MDCT for the detection of lymph node metastasis is unsatisfactory, with a sensitivity of approximately 50% [61,67,68]. Noji et al. reviewed CT findings and pathological results in patients who underwent lymphadenectomy for biliary tract cancer, including PHC, and found no significant correlations, concluding that CT is not valid for predicting lymph node metastasis from biliary tract cancer [67].

Magnetic resonance imaging (MRI) coupled with magnetic resonance cholangiopancreatography is also recommended as the second step to diagnose PHC. The advantage of MRI is that the whole biliary tree can be visualized without the use of a contrast medium, which helps to evaluate the precise location and local extension of PHC [54,61]. The meta-analysis of PHC reported that the accuracy of MRI in assessing ductal extension varied from 71% to 80% [61]. The disadvantage of MRI includes a lack of accuracy in evaluating vascular invasion and poor information regarding adenopathy. However, diffusion-weighted imaging strengthens the performance of MRI in visualizing these abnormalities [54]. When both MDCT and MRI are performed, the accuracy of predicting resectability is reported to exceed 75% [54].

Positron emission tomography (PET) has a marginal role in staging PHC. This modality can be used to identify metastatic lymph nodes and distant metastases or to clarify indeterminate lesions. However, the role of PET may be limited because of its false-positive results in cases of inflammation and its false-negative results because of a high desmoplastic reaction [48]; therefore, it is reasonable for PET to be indicated in some selected patients when needed.

With the evolution of cross-sectional imaging, the role of endoscopic retrograde cholangiopancreatography (ERCP) during preoperative workup has gradually evolved from a radiological diagnostic tool into an interventional procedure aimed at providing cytological specimens when pathological confirmation is required. In resectable cholangiocarcinoma, the diagnostic sensitivity/specificity of brushing cytology, forceps biopsy, and the comprehensive diagnosis of these two examinations performed in ERCP were reported to be 44%/99%, 48%/99%, and 59%/100%, respectively, in a meta-analysis [69]. Their procedure-related complications were pancreatitis, cholangitis, cholecystitis, and so on (3%–6%) [70,71]. The diagnostic accuracy of endoscopic ultrasonography (EUS) fine-needle aspiration (FNA) is reported to be higher than that of transpapillary biopsy and/or cytology (94% vs. 53%) [72]. EUS-FNA for distal cholangiocarcinoma can be performed safely as that for pancreatic head cancer. For PHC, however, this intervention would cause peritoneal dissemination [73], similar to transabdominal US-guided needle biopsy [74]. Therefore, only transpapillary biopsy and/or cytology are recommended. If a transpapillary approach is impossible, then bile aspiration cytology through an external biliary drainage catheter is useful [75,76].

5. Surgical anatomy of the liver

A complete understanding of the liver anatomy is mandatory to perform precise and safe resection of PHC. The bile duct, hepatic artery, and portal vein run intertwined at the hepatic hilum, forming the portal triad. Although understanding the origin, number, and branching pattern of these biliovascular vessels is essential, knowing three-dimensional (3D) anatomy, that is, spatial relationship among these vessels, is more important from a surgical perspective. The portal vein is an anatomical landmark, and the course of the hepatic artery and bile duct relative to the portal vein should be grasped. For this, MDCT-generated images are beneficial. The advent of MDCT has expanded conventional planar information into visually comprehensive 3D images [77,78]. In addition, recent technological advances have developed 3D image-processing software tools dedicated to computer assistance in liver surgery. Using such software packages, surgeons can now virtually perform various types of liver resections under realistic anatomic conditions [[79], [80], [81], [82]].

Yoshioka et al. evaluated the course of the right posterior hepatic artery to the right portal vein using MDCT-generated 3D arteriography and portography. The artery ran caudally to the portal vein (infraportal type) in 80% of patients, cranially (supraportal type) in 12%, and along both sides in 9% (Figures 2A and B) [77]. The supraportal right posterior hepatic artery runs just beneath the right hepatic duct, which may function as an anatomic trap during left-sided hepatectomy. Therefore, surgeons should understand the course of the right posterior hepatic artery before left-sided hepatectomy, especially left trisectionectomy for PHC [36]. Shimizu et al. studied the course of the left hepatic artery to the umbilical portion of the left portal vein using the same method mentioned previously. The artery ran into the left lateral sector from the left side to the umbilical portion in 91% of patients, the right side in 5%, and both sides in 3% (Figures 2C and D) [78]. This information is also essential when planning right-sided hepatectomy, especially right trisectionectomy for PHC [32].

Fig. 2.

Three-dimensional liver anatomy. (A and B) Course of the right posterior hepatic artery (RPHA). The artery runs caudally to the right portal vein (infraportal type, A), and the artery runs cranially to the right portal vein (supraportal type, B). (C and D) Course of the left hepatic artery (LHA). The artery runs the right side to the umbilical portion of the left portal vein (C), and the artery runs the left side to the umbilical portion of the left portal vein (D). (E–H) Exposure of the right hepatic vein (RHV) on the transected plane. A case in which the RHV was fully exposed (virtual hepatectomy, E; operative view, F). Note that in this case, the portal vein was resected and reconstructed using the external iliac vein graft (blue double-headed arrow), and the hepatic artery was also resected and reconstructed with end-to-end anastomosis (red arrow). A case in which the RHV was partially exposed (virtual hepatectomy, G; and operative view, H).

In addition, hepatobiliary surgeons should pay attention to the spatial relation between the portal vein and the hepatic vein. Sato et al. investigated the spatial anatomy of the right intersectional plane by virtual left trisectionectomy using 3D image-processing software in 200 patients who underwent MDCT [82]. They demonstrated that only 55% of the study patients showed complete exposure of the right hepatic vein on the transected plane (Figures 2E and F). In contrast, the remaining patients exhibited partial exposure because one or two portal branches of segment VI ran ventrally to the periphery of the right hepatic vein (Figures 2G and H) [82]. Furthermore, they showed that the extent of vein exposure is closely related to the type of inferior right hepatic vein; all patients with large inferior right hepatic vein showed partial exposure of the right hepatic vein. This finding means that the right hepatic vein does not always run along the right intersectional plane, which is one reason why left trisectionectomy is technically demanding [83].

6. Preoperative biliary drainage

Several previous studies, including meta-analyses, showed that preoperative biliary drainage did not improve morbidity and mortality in patients undergoing resection of biliary tract cancer [84,85]. In 2000, Cherqui et al. reported a controversial study [86]: the authors performed major hepatectomy without biliary drainage in 20 consecutive jaundiced patients and compared their surgical outcomes with 27 matched nonjaundiced patients with normal underlying liver. They showed no significant differences in mortality (5% vs. 0%) and liver failure (5% vs. 0%). However, morbidity was significantly higher in the jaundiced group (50% vs. 15%), leading us to conclude that major hepatectomy without preoperative biliary drainage is safe in most patients with obstructive jaundice. However, several recent studies showed that preoperative biliary drainage in patients with PHC reduced mortality after right-sided hepatectomy [87,88], especially in cases of the future liver remnant (FLR) being < 30% [89]. Furthermore, incomplete drainage in patients with FLR of < 50% is one of the risk factors for mortality in resectable PHC [90]. Currently, to my knowledge, almost all surgeons agree that preoperative biliary drainage is necessary for patients scheduled to undergo major hepatectomy for PHC [91].

There are several methods of biliary drainage, including percutaneous transhepatic biliary drainage (PTBD), endoscopic biliary stenting (EBS), endoscopic nasobiliary drainage (ENBD), and endoscopic stenting above the papilla (inside-stent; Fig. 3A-D). Previously, PTBD was preferentially used as an established method in many Japanese centers. Approximately 10 years ago, however, the trend in performing biliary drainage changed mainly from PTBD and ENBD because of the potential risks associated with PTBD [91]. One risk is vascular injury, and another risk is seeding metastasis including catheter tract recurrence [92] or peritoneal/pleural dissemination [93,94]. Several authors reported significantly poorer survival after PTBD compared with after endoscopic drainage in patients with resected PHC [95,96]. In this sense, PTBD cannot be the first choice of the procedure. EBS is accompanied by a higher risk of ascending cholangitis [97], leading to an increased postoperative complication. On the other hand, ENBD has definitive advantages over PTBD and EBS, including having no risk of seeding metastasis, a lower risk of ascending cholangitis, and longer patency [98,99]. Overall, ENBD is the most suitable drainage method for PHC patients undergoing resection.

Fig. 3.

Preoperative biliary drainage methods. (A) Percutaneous transhepatic biliary drainage (PTBD). (B) Endoscopic biliary stenting (EBS). (C) Endoscopic nasobiliary drainage (ENBD). (D) Endoscopic stenting above the papilla (inside-stent).

Nevertheless, ENBD has some disadvantages, including the loss of the bile and the discomfort for patients. Having a drainage tube coming out of the nose for a long period poses great discomfort to the patients. Maeda et al. studied the preoperative course (median length between the first admission and surgery = 37 days, range 12–197 days) in 191 consecutive patients who underwent resection of PHC after ENBD [100]. The authors reported that no patients intentionally removed the ENBD catheter, which is widely tolerable with a relatively low incidence of cholangitis [100]. However, these favorable results may come from the national mentality of Japanese patients. In sharp contrast, ENBD is wholly rejected in Western countries. To resolve this troublesome issue, inside-stents have recently been used as preoperative biliary drainage for PHC. Several authors reported promising inside-stent results compared with those of ENBD [101,102].

Finally, I mention an issue with the self-expandable metallic stent (SEMS). Uncovered SEMSs are widely used as palliative measures for unresectable tumors. However, if there is any possibility of resection, then uncovered SEMS should never be placed because it makes a salvage hepatectomy very difficult [103,104].

7. Portal vein embolization

Portal vein embolization (PVE), a procedure devised by Makuuchi et al. in the 1980s [105], has been widely used in the presurgical preparation to reduce the risk of postoperative liver failure in patients undergoing extended hepatectomy. Because of a lack of randomized controlled studies on PVE, the evidence level on the clinical value of PVE is low. However, several studies have demonstrated the clinical utility of PVE before extended hepatectomy for PHC [6,7,9,38,106,107]. Mortality after hepatectomy is lower in reports from Japan [6,7,9,38] where PVE is liberally used, compared with that of reports from Western countries [[12], [13], [14], [15],20,22,23]. Quite recently, Muller et al. reported surgical and oncological results of a benchmark PHC study, where 24 expert centers around the world participated [108]. Benchmark (= low risk) patients were defined as patients who underwent standard major hepatectomy without vascular resection and who had no major comorbidities. When comparing Asian (n = 272) vs. non-Asian (n = 436) centers in this homogeneous benchmark cohort, Asian centers demonstrated a significantly better overall survival (hazard ratio 1.64, 95% CI 1.26–2.13, P < 0.001) with a lower rate of severe liver failure (1.7% vs. 7.7%, P = 0.018) and a lower tendency of in-hospital mortality (2.0% vs. 5.9%, P = 0.06). The authors speculated that these superior results of Asian centers were attributable to more frequent use of preoperative PVE (41.5% vs. 24.8%, P < 0.001) as well as the routine performance of preoperative liver function testing, including the indocyanine green (ICG) test (< 50% in non-Asian centers). According to a recent review survey [109], FLR of < 40% was a common indication of PVE before hepatectomy for PHC.

Approximately 30 years ago, inspired by Makuuchi's report [105], PVE was used for biliary tract cancer at Nagoya University Hospital. Thereafter, I developed a new approach, that is, the ipsilateral approach, for percutaneous transhepatic PVE [110,111]. This approach replaced the conventional contralateral approach [105], which is now used worldwide as the standard method of PVE. Furthermore, I introduced a method of trisegment PVE: in the right trisegment PVE, the right portal vein and left medial portal branches are embolized; in the left trisegment PVE, the left portal vein and right anterior portal branch are embolized [111,112]. These trisegment PVEs are possible only by the ipsilateral approach and are also widely used as presurgical preparations for hepatic trisegmentectomy [32,36,38,113].

Recently, some surgeons have used associating liver partition and portal vein ligation in staged hepatectomy (ALPPS) as an alternative to PVE for resection of PHC, despite very high mortality rates after ALPPS [114,115]. This strategy, however, may be better to reconsider because of the following reasons [116]. First, bile is colonized/infected in most patients with PHC because of preoperative biliary drainage [117]. This condition may induce infectious complications after ALPPS, leading to high morbidity and mortality. Second, bile in patients with PHC contains floating cancer cells [75,76]. In ALPPS, the cancer is not resected and left in situ in the first stage; consequently, bile leakage associated with ALPPS is a considerable risk for the future development of peritoneal dissemination. Third, the rapid expansive volume gain offered by ALPPS is unnecessary because PHC is a slow-growing cancer, as already described in the original paper by Klatskin [45]. Thus, in effect, there is time to wait until enough hypertrophy is achieved after PVE.

8. Assessment of hepatic functional reserve of FLR

The hepatic functional reserve depends on both liver volume and hepatic function: the former is measured by CT volumetry, and the latter is assessed by the ICG test. Unfortunately, there is little available evidence on the relationship between the hepatic functional reserve and surgical outcomes. Ribero et al. reported that FLR of < 30% and preoperative cholangitis triggered hepatic insufficiency and were determinants of postoperative liver failure–related death in patients who underwent hepatectomy for PHC [118]. Olthof et al. showed that FLR of < 30%, preoperative cholangitis, jaundice at initial presentation, and total bilirubin of > 2.9 mg/dL at surgery were risk factors for posthepatectomy liver failure [119]. A major drawback in these studies from Western countries was a lack of liver functional assessment, as Western surgeons did not perform preoperative ICG tests in principle.

In 2006, I analyzed the surgical outcomes of 240 patients with biliary tract cancer who underwent major hepatectomy after PVE [107] and found that postoperative mortality was significantly higher in patients whose ICG disappearance rate of FLR (ICGK-F) was < 0.05 than it was in those whose ICGK-F was ≥ 0.05 (28.6% vs. 5.5%, P < 0.001). Furthermore, Yokoyama et al. demonstrated that an ICGK-F with a cutoff value of 0.05 works as a predictor of postoperative liver failure and mortality [120,121]. This new index is easy to calculate (ICGK-F= ICGK × proportion of FLR) and rational, as both volume and function are integrated, thus being widely used in Japan [122]. However, a standardized approach for determining the hepatic functional reserve of the FLR has yet to be established. Multiple factors are linked to liver failure and mortality; therefore, institutional expertise [108], patients’ background (age, performance status, and comorbidity), and combined procedure (concomitant pancreatoduodenectomy, portal vein, and/or hepatic artery resection) should also be considered for safety assessment [91].

9. Surgery

9.1. Basic policy

The resection procedures depend on the location of the primary tumor [49,50]. In principle, right hemihepatectomy is applied to Bismuth type I, II, and IIIa tumors [6,7,9,33,38], whereas left hemihepatectomy is applied to Bismuth type IIIb tumors. Regarding the choice of procedures for Bismuth type I or II tumors, which are inevitably located near the right hepatic artery, local hilar resections have often been performed [[2], [3], [4], [5]], but such limited resections often result in R1/2 resection or local recurrence frequently occurs even after R0 resection, leading to a bleak prognosis [[2], [3], [4], [5]]. Therefore, right hemihepatectomy is recommended, especially for nodular or infiltrating type PHCs [33]. Although this strategy is still deemed rational, Sugiura et al. showed that left hemihepatectomy with combined resection and reconstruction of the right hepatic artery is a valid alternative to right hemihepatectomy, especially in patients with an insufficient left liver functional reserve [124]. This concept is attractive because arterial reconstruction in left hemihepatectomy is relatively easy because of the short-range involvement and large caliber of the distal artery. In Bismuth type IV tumors, the type of hepatectomy is determined by considering the predominant tumor location, the presence or absence of portal vein and/or hepatic artery invasion, and liver function. Right trisectionectomy is applied to Bismuth type IV tumors with a right-sided predominance or an even extension [4,6,7,9,32,38]. In contrast, left trisectionectomy is selected for Bismuth type IV tumors with a left-sided predominance [36,84,123]. Left trisectionectomy is still rarely performed because of its technically demanding nature. However, this hepatectomy, compared with left hemihepatectomy, can increase the number of negative proximal ductal margins, leading to a high proportion of R0 resections and, in turn, to improved survival for patients with advanced left-sided PHC [36,84].

Figure 4 shows changes in the type of hepatectomy over 43 years at Nagoya University Hospital. More than 20 years ago, bile duct resection alone or central hepatectomy (H1458, H14, or H1, according to the New World Terminology [125]) was often performed, while recently these limited resections have rarely been selected. Currently, four types of major hepatectomy, that is, right hemihepatectomy (H15678), right trisectionectomy (H145678), left hemihepatectomy (H1234), and left trisectionectomy (H123458), are standard procedures for PHC. Many hepatobiliary surgeons still believe that right-sided hepatectomy is oncologically superior to left-sided hepatectomy [4,6,7,22,126]. However, this is a fallacy, which should be exploded [127]. Recent studies on the issue of right- or left-sided resections for PHC reported comparable long-term survival [108,[128], [129], [130], [131]], although mortality was much higher with right-sided hepatectomy [108,[129], [130], [131]]. When the resection procedure is flexibly selected, oncological sidedness is not associated with the type of hepatectomy.

Fig. 4.

Changes in types of hepatectomy between 1977 and 2020 at Nagoya University Hospital (n = 1078).

9.2. Extended resection

For locally advanced PHC, extended resection added to standard major hepatectomy is required (Fig. 5). When a tumor has longitudinal extension with downward spread to the intrapancreatic bile duct, major hepatectomy with pancreatoduodenectomy, that is, hepatopancreatoduodenectomy (HPD), is applied [37,38]. On the other hand, when a tumor exhibits vertical extension with involvement of the portal vein and/or hepatic artery, combined vascular resection is performed [[16], [17], [18], [19],35,38,41]. In addition, superextended resection, that is, HPD with vascular resection, is indicated for advanced PHC with both longitudinal and vertical extensions. Among them, HPD with simultaneous resection of the portal vein and hepatic artery is the most complicated and much more demanding procedure, which is why it is referred to as ultimate superextended resection [40,132].

Fig. 5.

Surgical procedures for perihilar cholangiocarcinoma according to cancer extent. Four major hepatectomies (left or right hemihepatectomy, left or right trisectionectomy) are standard procedures for perihilar cholangiocarcinoma. In the case of longitudinal extension, hepatopancreatoduodenectomy is necessary. In the case of vertical extension, combined vascular resection is essential. In the case of both longitudinal and vertical extension, superextended resection, that is, hepatopancreatoduodenectomy with vascular resection, is required.

HPD is indicated in the following cases: (1) diffusely infiltrating cholangiocarcinoma of the whole extrahepatic bile duct, (2) PHC with downward superficial spreading, or (3) bulky nodal metastases of the pancreatoduodenal region. Since Takasaki et al. first introduced HPD to treat locally advanced gallbladder cancer in 1980, some aggressive surgeons have used this extended procedure to treat biliary tract cancer [25]. However, a limited number of patients and a high rate of mortality ranging from 13 to 60% hindered our ability to reach a definitive conclusion regarding the survival benefit of HPD [25]. Even today, HPD for biliary tract cancer remains controversial and is the most challenging surgery. In 2012, Ebata et al. reported the surgical outcomes of 85 patients with cholangiocarcinoma (59 perihilar and 26 distal tumors) who underwent HPD [37]. The types of hepatectomy performed included right-sided hepatectomy in 55 patients, left-sided hepatectomy in 23 patients, and central hepatectomy in seven patients. The 90-day mortality rate was only 2.4%, and the overall 5-year survival rate was 37.4%. After that, Aoki et al. also showed better surgical and oncological outcomes of 52 patients with biliary tract cancer (39 cholangiocarcinomas and 13 gallbladder cancers) who underwent HPD [133]. These reports demonstrate that HPD can be performed with low mortality and offers a better oncological outcome.

Combined portal vein resection is now a routine procedure in leading centers in both the East and West, and its clinical benefit has been validated by many studies [16,19,31,38,41]. Portal vein resection is performed primarily in cases of right-sided hepatectomy, most likely because of technical feasibility. However, surgeons should note that vessel resection is often required in cases of left-sided hepatectomy for advanced PHC with left-sided predominance [35,36,38]. Portal vein resection should be performed only when the vessel adheres to and cannot be freed from the tumor during skeletonization of the hepatoduodenal ligament [38]. Hilar en bloc resection by the no-touch technique, introduced by Neuhaus et al. in 1999 [4], should not be used [134], as this preemptive extended vascular resection is highly associated with mortality and lacks oncological benefit [108]. A recent study from Nagoya University Hospital has demonstrated mortality of < 3% and 5-year survival rate of 25% in patients with PHC who underwent hepatectomy with combined portal vein resection [41].

Recent advances in surgical techniques and knowledge, which have been gained from experiences with liver transplantation, have facilitated the performance of hepatic artery resection with reconstruction. However, most previous studies showed negative results and did not recommend combined hepatic artery resection for biliary tract cancer. For example, Miyazaki et al. reported a mortality rate of 33.3% (3/9) in patients who underwent hepatectomy with hepatic artery resection, and none of these patients survived for more than 3 years [10]. From these observations, they concluded that hepatic artery resection, unlike portal vein resection, cannot be justified [10]. In 2010, I reported my experiences with major hepatectomies with simultaneous resections and reconstructions of the portal vein and hepatic artery (n = 50) and showed that this challenging surgery can be performed with an acceptable mortality rate of 2% and offers a better chance of long-term survival, with a 5-year survival rate of 30% [35]. Thereafter, the number of patients who underwent combined hepatic artery resection for PHC has increased, now reaching nearly 150 [41]. The overall mortality rate was 4%, and the 5-year survival rate was 29.5% [41]. Sugiura et al. also reported similar favorable outcomes of combined hepatic artery resection for PHC [21]. Although the clinical significance of hepatic artery resection is still debatable, this extended resection will be disseminated to treat advanced PHC.

A further challenge is ultimate superextended surgery, that is, HPD with simultaneous resection of the portal vein and hepatic artery (Fig. 5). Between April 2007 and March 2020, this surgery was performed on nine patients with PHC, including one patient with recurrent PHC after extrahepatic bile duct resection for cholangiocarcinoma (Fig. 6) [40]. One patient died of multiple organ failure caused by pancreatic fistula-related bleeding on postoperative day 86, whereas the remaining eight patients were discharged in good health, giving 90-day mortality of 11% (1/9). They all died of recurrence, but two patients survived for more than 7 years (86 months and 104 months, respectively) [40]. During the same period, HPD with PV resection alone (n = 17) or HA resection alone (n = 4) was performed in another 21 patients with advanced PHC. In this cohort, the 90-day mortality was 0%, and five patients survived for more than 5 years. Of these, two patients are still alive more than 10 years after surgery [40]. These outcomes are not satisfactory but promising and thus may encourage hepatobiliary surgeons to attempt superextended surgery.

Fig. 6.

A case of superextended hepatopancreatoduodenectomy (left trisectionectomy, caudate lobectomy, conventional pancreatoduodenectomy, and simultaneous resection of the portal vein and hepatic artery). (A) Preoperative scheme of cancer extent. Hilar tumor with left-sided predominance extensively involves the hepatoduodenal ligament. (B) Pancreatoduodenectomy was performed first, preserving the plexus around the superior mesenteric artery. (C) Liver transection (left trisectionectomy) was completed. The right hepatic vein was exposed on the transected plane. After that, the vasculobiliary system was divided in the following order: (1) the right posterior hepatic duct; (2) the common hepatic artery; (3) the right posterior hepatic artery; (4) the middle and left hepatic veins; (5) the main portal vein; and (6) the right posterior portal vein. After extirpation of the specimen, the portal vein was first reconstructed, and then the hepatic artery was reconstructed. (D) The portal vein was reconstructed using an external iliac vein graft. The hepatic artery was reconstructed using the rotating left gastric artery. (E) Scheme of superextended surgery. CHA, common hepatic artery; SA, splenic artery; SV, splenic vein; SMV, superior mesenteric vein; RHV, right hepatic vein; MHV, middle hepatic vein; LHV, left hepatic vein; RPHD, right posterior hepatic duct; RPPV, right posterior portal vein; RPHA, right posterior hepatic artery; EIV, external iliac vein; and LGA, left gastric artery (cited from reference No. 40).

9.3. Nagoya experience

During the abovementioned 13 years, 646 consecutive patients with PHC and four patients with local recurrence (at or near the hepatic hilum) after resection for cholangiocarcinoma underwent surgical resection. Of these 650 patients, 636 (97.8%) patients underwent hepatectomy with (n = 83) or without pancreatoduodenectomy (n = 553) (Table 1). A total of 248 (38.2%) patients underwent combined vascular resection, including portal vein resection alone (n = 122), hepatic artery resection alone (n = 42), and both vascular resections (n = 84). The median operative time was 566 min (range, 303–1167 min), and the median blood loss was 1095 mL (range, 46–10,799 mL). Regarding surgical curability, 462 (71.1%) patients with pM0 disease had R0 resection, and 120 (18.5%) patients with pM0 disease had R1/2 resection. The remaining 68 (10.5%) patients had pM1 disease, including periaortic lymph node metastasis (n = 32), ipsilateral liver metastasis (n = 30), and limited peritoneal dissemination (n = 6). Table 2 shows the 90-day mortality and overall survival according to the type of procedure [40]. The 90-day mortality of all cohorts was only 1.4% (9/650), and their 5-year overall survival rate was more than 40%.

Table 1.

Surgical procedures performed in 650 patients (April 2007–March 2020, Nagoya University Hospital).

Procedures		Combined vascular resection
		Without	PV alone	HA alone	PV + HA
Hepatopancreatoduodenectomy	(n = 83)	53	17	4	9
H123458-B-PD	(n = 15)	7	0	2	6
H1234-B-PD	(n = 16)	14	0	2	0
H145678-B-PD	(n = 3)	1	1	0	1
H15678-B-PD	(n = 48)	31	16	0	1
H158-B-PD	(n = 1)	0	0	0	1a
Hepatectomy alone	(n = 553)	335	105	38	75
H123458-B	(n = 158)	68	18	22	50
H1234-B	(n = 166)	112	19	12	23
H145678-B	(n = 66)	32	34	0	0
H15678-B	(n = 155)	117	33	3	2
H1458-B	(n = 6)	4	1	1	0
Other hepatectomy	(n = 2)	2	0	0	0
Pancreatoduodenectomy alone	(n = 4)	4	0	0	0
Bile duct resection alone	(n = 10)	10	0	0	0
Total	(n = 650)	402	122	42	84

B, extrahepatic bile duct resection; PD, pancreatoduodenectomy; PV, portal vein resection; HA, hepatic artery resection.

Note that procedures are described according to the New World Terminology [125].

^aThis patient had undergone left hepatectomy for intrahepatic stones 18 years before.

Table 2.

Mortality and survival in various kinds of surgery (April 2007 to March 2020, Nagoya University Hospital).

Procedures		Mortality (90-day)	Overall survival
			3 years	5 years	7 years	10 years	MST (months)
Hepatopancreatoduodenectomy
With PV and HA	(n = 9)	1 (11.1%)	33%	22%	22%	0%	22
With PV alone or HA alone	(n = 21)	0	44%	36%	36%	36%	21
Without vascular resection	(n = 53)	0	68%	46%	34%	34%	58
Hepatectomy alone
With PV and HA	(n = 75)	2 (2.7%)	54%	27%	19%	9%	37
With PV alone or HA alone	(n = 143)	3 (2.1%)	45%	28%	22%	20%	32
Without vascular resection	(n = 335)	3 (0.9%)	68%	56%	48%	39%	82
PD or bile duct resection alone
Without vascular resection	(n = 14)	0	90%	56%	45%	Not reached	62
Total	(n = 650)	9 (1.4%)	59%	43%	35%	27%	49

PV, portal vein resection; HA, hepatic artery resection; PD, pancreatoduodenectomy; MST, median survival time.

9.4. Liver transplantation for PHC

Liver transplantation was introduced for PHC more than 30 years ago. In the last decade, the Mayo Clinic group reported promising results with a neoadjuvant protocol consisting of multimodal chemoradiation therapy under strict inclusion criteria [[135], [136], [137]]. Their early report was superb, with a 5-year survival rate of above 80% [135]. Since the establishment of the Mayo Protocol, several other groups have reported their experience with the same or similar protocols. A multicenter study of 12 US centers reported a 5-year recurrence-free survival rate of 65% after liver transplantation for PHC (n = 216) [138]. According to Mayo criteria [137], good candidates for liver transplantation have “solitary, less than 3 cm in diameter, no lymph node metastasis, but unresectable tumor.” However, it is difficult to imagine an unresectable tumor at such a relatively early stage. I speculate that most PHCs meeting the Mayo criteria are not unresectable, but resectable [139]. Extended resections, including trisectionectomy [32,36,38], vascular resections [31,35,38,40,41], and HPD [37,40,133], can circumvent unnecessary transplantation and offer a favorable outcome even in some patients with locally advanced PHC. Patients with primary sclerosing cholangitis developing PHC are good candidates for liver transplantation. On the other hand, further studies are needed to evaluate whether liver transplantation is truly superior to resection as a surgical treatment of de novo PHC without any identifiable underlying cause [139].

10. Postoperative management

Postoperative management at Nagoya University Hospital has changed with the times [38], and recent (approximately after 2007) institutional protocol is as follows. Antibiotics for prophylaxis were selected based on preoperative bile culture [38,91]: when the culture was negative, first- or second-generation cephalosporins were used; and when it was positive, antibiotics were selected according to the susceptibility of the specific microorganisms detected. A laboratory blood test and chest/abdominal X-ray examination were performed every day until day 7. The abdominal drain fluid was measured for total bilirubin and amylase concentrations and sampled for bacterial culture on days 1, 3, and 7 [38,91]. MDCT on day 7 was routinely performed in all patients to survey abdominal fluid collection, liver perfusion, thrombosis, pleural effusion, and pneumonia [38]. Based on these findings, abdominal drains were removed or repositioned to an appropriate site to maximize the drainage effect, and US-guided abscess/fluid drainage was performed on demand. Enteral tube feeding started on day 1 and decreased inversely as oral intake increased around day 4 [38]. Synbiotic treatment [140] and exercise program [141] also started on day 1 and continued for at least 2 weeks.

Complex hepatectomy required for resection of PHC has been a surgical challenge due to technical demand, impaired hepatic function, and potential biliary infection, leading to frequent postoperative complications ranging from 43% to 69% [[11], [12], [13],38,88]. According to our previous study which analyzed postoperative complications in 484 resected PHC patients [142], all patients had any complications (Clavien-Dindo classification [143] grade I, n = 27; grade II, n = 132; grade III, n = 294; grade IV, n = 22; and grade V, n = 9), and major complications (Clavien-Dindo classification grade III ≤) occurred in 67.1% (325/484) of the patients, but mortality was only 1.9% (9/484). This means that our meticulous, preemptive postoperative management greatly contributed to the low failure to rescue rate. Ghaferi et al. stressed that a high-mortality hospital is characterized by a high failure to rescue rate [144]. Endo et al., who studied postoperative complications in 422 patients who underwent HPD, reported that the incidence of grade III≤ complications in high-volume centers was significantly higher than in low-volume centers, but postoperative mortality was significantly better in the former than in the latter [145]. Nagoya University Hospital, a high-volume specialized center for biliary cancer, has well-organized multidisciplinary teams that facilitate the detection of signs of major complications, enabling early intervention by involving endoscopists, radiologists, infection doctors, intensivists, and nurse practitioners. Therefore, good teamwork is the key to achieve low failure to rescue rate.

11. Adjuvant chemotherapy

Unfortunately, it is well known that the incidence of cancer recurrence is high even after curative-intent resection of PHC [146,147]. Nakahashi et al. reported that approximately 60% of patients with PHC experience recurrence even after R0 resection and that lymph node metastasis (hazard ratio 2.80, P < 0.001) and microscopic venous invasion (hazard ratio 1.80, P < 0.001) were independent risk factors for recurrence-free survival in R0 patients [147]. Even in patients without such a risk factor, the probability of recurrence is high, reaching 30% at 5 years and 50% at 10 years [147]. These harsh findings strongly indicate that close surveillance for 10 years is necessary and that establishment of effective adjuvant chemotherapy is essential to prolong the survival of patients even after R0 resection of PHC.

The ESPAC-3 randomized controlled trial, published in 2012, explored the role of adjuvant chemotherapy for resected pancreaticobiliary tumors [148]. In this trial, 428 patients with periampullary cancers (a heterogeneous group including 297 periampullary cancers, 96 cholangiocarcinomas, and 35 other subtypes) were randomized after curative resection to observation alone, adjuvant 5-FU or adjuvant gemcitabine chemotherapy. In the 96 patients with cholangiocarcinoma, adjuvant chemotherapy did not improve overall survival (27.2 months with 95% CI 15.4–31.9 months vs. 18.3 months with 95% CI 12.9–28.7 months vs. 19.5 months with 95% CI 16.2–36.1 months for the observation, 5-FU, and gemcitabine groups, respectively).

Since 2017, three randomized phase III clinical trials have been reported [[149], [150], [151]]. All trials recruited patients with resected biliary tract cancer (cholangiocarcinoma and gallbladder cancer) who were randomized to observation alone or chemotherapy in the form of gemcitabine (BCAT [149], including 226 patients with extrahepatic cholangiocarcinoma only), gemcitabine and oxaliplatin (PRODIGE-12/ACCORD-18 [150], including 196 patients with cholangiocarcinoma and gallbladder cancer), or capecitabine (BILCAP [151], including 447 patients with cholangiocarcinoma and gallbladder cancer). Although gemcitabine-based chemotherapy failed to show an impact on relapse-free survival or overall survival, the BILCAP study showed a benefit from adjuvant capecitabine in terms of overall survival (preplanned sensitivity analysis of the intention-to-treat population and in the per-protocol analysis), with confirmed benefit in terms of reference-free survival. Based on the BILCAP trial, international guidelines recommend adjuvant capecitabine for a period of 6 months following potentially curative resection as the current standard of care for resected cholangiocarcinoma and gallbladder cancer [152]. However, BILCAP failed to show an overall survival benefit in the intention-to-treat (nonsensitivity analysis) population. This finding and some inconsistencies between studies have been criticized and have led to confusion in the biliary tract cancer medical community [153].

Recently, Seita et al. evaluated the efficacy of adjuvant S-1 chemotherapy (80–120 mg/day on 14 days of tri-weekly cycle for 6 months) in 50 patients who underwent resection of histologically proven node-positive biliary tract cancer (43 cholangiocarcinomas and seven gallbladder cancers) [154]. The 3-year overall survival and recurrence-free survival rates were 50% (95% CI, 40.9–59.1%) and 32.0% (95% CI, 19.1–44.9%), respectively, with median survival times of 34.6 months (95% CI, 19.3–49.8 months) and 18.4 months (95% CI, 11.9–24.9 months). Although this study is a single-arm phase II trial, the results imply that adjuvant S-1 chemotherapy is promising for node-positive biliary tract cancer. In addition, Takahashi et al. retrospectively analyzed the efficacy of adjuvant chemotherapy (S-1 or gemcitabine) in patients who underwent resection of node-positive PHC [155]. Recurrence-free survival was significantly longer in S-1 patients than in gemcitabine patients (24.4 vs. 14.9 months, P = 0.44), whereas overall survival was not (48.5 vs. 35.0 months, P = 0.324). After propensity score adjustment, the differences were more evident (hazard ratio 2.696, 95% CI 1.739–4.180, P < 0.001; hazard ratio 1.988, 95% CI 1.221–3.238, P < 0.001). These observations support the idea that fluoropyrimidine is more effective than gemcitabine or gemcitabine-combined regimen and may improve the survival of node-positive PHC patients. Data from an ongoing phase III clinical trial (JCOG1202 “ASCOT” trial: adjuvant S1 vs observation alone in resected biliary tract cancer) [156] are awaited to demonstrate the efficacy of adjuvant S-1 monotherapy.

12. Closing remarks

Surgical and oncological outcomes of PHC have steadily improved, but there are still several issues to be resolved in the surgical treatment of PHC. Further synergy of endoscopists, radiologists, oncologists, and surgeons is essential to conquer this intractable disease. Unfortunately, chemotherapy currently has little effect on PHC, and surgical resection is the only treatment option for cure. Therefore, the most important matter is to pursue the possibility of surgical resection. All hepatobiliary surgeons should make an effort to refine their surgical skills to safely and properly perform the complicated hepatobiliary resections, which leads to expanded surgical indications and improved survival of PHC. Under the veil of empty multidisciplinary treatment, the surgical challenge should not quickly be abandoned. Another important matter is to develop more effective anticancer drugs and to establish more valid neoadjuvant as well as adjuvant chemotherapy. For this, adequately designed and properly powered randomized phase III studies with sufficient follow-up are required [153].

Declaration of competing interest

The author declare no conflicts of interest.