A New Pathological Classication of Intrahepatic Cholangiocarcinoma from an Embryological Viewpoint

Background: No universal classication method for intrahepatic cholangiocarcinoma (IHCC) has been reported based on the embryological origin of biliary epithelial cells. The aim of this study was to classify IHCC according to protein expression levels of somatostatin receptor 2 (SSTR2) and b-cell leukemia/lymphoma 2 (Bcl2) and to elucidate the clinicopathological features of each group. Methods: Fifty-two IHCC patients who underwent hepatic resection were enrolled in this study. Protein expression levels of SSTR2 and Bcl2 were examined using immunohistochemistry. Clinicopathological factors were compared between the three groups and prognostic factors were investigated.


Background
Intrahepatic cholangiocarcinoma (IHCC) is a primary adenocarcinoma of the liver that arises from the intrahepatic bile ducts. It is the second most common primary hepatic tumor, after hepatocellular carcinoma (HCC), and comprises about 5%-15% of total primary hepatic malignancies [1,2]. Compared with those suffering from other malignancies, IHCC patients have an extremely poor prognosis, even if they have had curative resections performed [1,2]. IHCC can arise from any portion of the intrahepatic biliary tree, and is classi ed as either perihilar or peripheral IHCC depending on where the tumor emerges [1,2]. Tumor type was often determined based on gross appearance [3,4]. Although most IHCC tumors are morphologically classi ed as adenocarcinomas, their molecular and biological features are heterogeneous. In general, the prognosis of patients with perihilar IHCC appears to be worse than those with peripheral IHCC [3,4]. This heterogeneity may have resulted from the embryological origin of IHCC [1,2,5]. However, it is di cult to determine where the IHCC emerged from based only on gross ndings and analysis of hematoxylin and eosin (H&E) stained histological tumor sections [6].
Previous studies have identi ed epidemiological and clinicopathological differences between perihilar and peripheral IHCC tumors [3][4][5][7][8][9]. Patients who develop peripheral IHCC often suffer from chronic liver diseases such as viral hepatitis or alcoholic cirrhosis. Perihilar IHCC, however, sometimes arises in individuals with hepatolithiasis and chronic in ammation of bile ducts, including primary sclerosing cholangitis and pancreaticobiliary maljunction. Lymph node metastasis, intrahepatic metastasis, and perineural invasion occur more often in perihilar IHCC patients [3,9]. Pathologically, higher expression levels of KRAS, S100P, and p53 are recognized in perihilar IHCC [4,9]. In contrast, higher expression levels of isocitrate dehydrogenase 1/2 (IDH1/2) and NRAS have been observed in peripheral IHCC [4]. Further clari cation of IHCC heterogeneity may provide critical information for the development of novel treatment methods. However, despite these studies, there are no universally accepted criteria for classi cation of an IHCC tumor as perihilar or peripheral.
In this study, we aimed to differentiate perihilar (large bile duct) and peripheral (small bile duct) carcinogenesis of IHCC through expression levels of two proteins, SSTR2 and Bcl2. We classi ed IHCC cases based on these criteria, then investigated the differences in clinicopathological characteristics, including prognosis, between the two groups.

Patient selection
Patients included in this study included 52 with IHCC, and 37 with extrahepatic cholangiocarcinoma (EHCC) as positive control was examined as control of large bile duct cancer, all of whom had undergone surgical resection at the Tokushima University Hospital between 1994 and 2017. All patients had surgical specimens available for immunohistochemistry and survived the surgery without any complications, such as postoperative liver failure. The IHCC patients included 34 men and 18 women ranging in age from 43 to 84 years old, with a mean age of 70.5 years. Following the Classi cation of Primary Liver Cancer by the Liver Cancer Study Group of Japan, T-factor was determined by number of tumors (one or more), size (no more than 2 cm in diameter), and vascular in ltration (present or absent). Tumor stage was determined by T-, N-, and M-factors. Curability of each patient was de ned as A, B, or C, as follows: A: no residual tumor in stage I or II IHCC; B: no residual tumor in stage III or IV IHCC; C: residual tumor in any stage IHCC. Based on these classi cations, 43 IHCC patients (82.7%) underwent resections with curability A or B. No patient received chemotherapy or irradiation before or after surgical resection. The 3-and 5-year survival rates of the IHCC patients were 40.9% and %, respectively, and the mean follow-up period was 24.7 months (range: 4.4-143.8 months).
De nition of peripheral IHCC, perihilar IHCC and EHCC according to gross appearance The cholangiocarcinomas were classi ed on the basis of nding of computed tomography (CT) into three groups, EHCC, perihilar and peripheral IHCC. In the present study, EHCC was de ned as the periductal in ltrating (PI) type of cholangiocarcinoma involving the left, right hepatic duct and those con uence [20]. Meanwhile, mass forming (MF) or MF plus PI type of cholangiocarcinoma, located in 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 tumor area, was de ned as perihilar IHCC [21]. And the other MF type cholangiocarcinoma was de ned as peripheral IHCC.

Surgical procedure
All patients underwent hepatic resection. Segmentectomy involving 3 segments was performed for 6 patients, segmentectomy involving 2 segments was performed for 33 patients, and segmentectomy involving a single segment or less was performed for 13 patients. Fifteen patients underwent extrahepatic bile duct resection with hepatic resection. Lymphadectomy was performed in 29 patients. Nine patients were not performed curative resection because of the positive surgical margin or para-aortic lymph nodes metastasis.

Immunohistochemical staining and assessment
Methods for immunohistochemical staining have been described previously [22]. Brie y, four mm thick tissue sections from each sample were depara nized and dehydrated. Next, 0.3% hydrogen peroxide and methanol were administered to the sections for 20 minutes to halt peroxidase activity, followed by heat treatment. The sections were incubated with a primary rabbit polyclonal antibody to SSTR2 (NB300-157, diluted 1:200 in PBS; NOVUS Biologicals LLC, Centennial, CO, USA) and a primary mouse monoclonal antibody to Bcl2 (M088701, diluted 1:40 in PBS; Dako, Santa Clara, CA, USA) overnight at 4˚C respectively.
Sections were then treated with the secondary antibody (the EnVision TM + Dual Link System-HRP, Dako) for 1 hour at room temperature. SSTR2 and Bcl2 expression was evaluated by scoring the staining intensity (0, negative; 1, low; 2, medium; 3, high). Score 0 were considered as negative staining and more than 1 as positive staining. All sections of the immunostaining were evaluated by a pathologist who had no patients' information.

Clinicopathological analysis
The IHCC patients were divided into three groups: pathological perihilar which meant large bile duct carcinogenesis (Group H), SSTR2 positive and Bcl2 negative; pathological peripheral which meant small bile duct carcinogenesis (Group P), SSTR2 negative and Bcl2 positive; and unclassi ed (Group U), SSTR2 positive and Bcl2 positive or SSTR2 negative and Bcl2 negative. Clinicopathological factors were compared between the three groups. Furthermore, prognostic factors were identi ed by univariate and multivariate analysis.

Statistical analysis
All statistical analysis was performed using statistical software (JMP 13, Cary, NC, USA). Relationships between SSTR2 and Bcl2 expression and the clinicopathological variables were analyzed with one-way ANOVA analysis followed by Tukey's test. Survival curves were generated using the Kaplan-Meier method and compared using the log-rank test. All factors found to be signi cant by univariate analysis were included in the Cox's proportional hazards model of multivariate analysis to identify independent factors in uencing patient survival. Statistical signi cance was de ned as p<0.05.

Results
Correlation between SSTR2 and Bcl2 expression and clinicopathological variables Tumor tissue was de ned by cell staining for SSTR2 in the cell membrane (Fig. 1A) and Bcl2 in the cytoplasm (Fig. 1B). In normal part of IHCC patient's liver, SSTR2 expression was detected in the large bile duct, while no staining in small bile duct (Fig. 1C). Bcl2 expression was detected only in small bile duct including bile ductule and interlobular bile duct (Fig. 1D). In EHCC tumors, 32 were positive for SSTR2 but negative for Bcl2, while 5 were positive for both molecules. In these ve EHCC, Bcl2 was expressed, but SSTR2 was not. Only one case stained negative for both molecules. Positive SSTR2 expression in cancer cells was present in 26 out of 52 IHCC cases (50.0%) and positive Bcl2 expression in cancer cells was present in 19 (36.5%). Of 52 total IHCC patients, 21 were categorized as Group H (40.4%) and 14 as Group P (26.9%). For the remaining 17 patients in the unclassi ed group (Group U), ve were positive for both SSTR2 and Bcl2 (9.6% of all patients) and 12 were negative for both SSTR2 and Bcl2 (23.1% of all patients). Table 1 presents a comparison of clinicopathological characteristics of IHCC patients categorized by SSTR2 and Bcl2 expression. Group P had a signi cantly better prognosis according to T classi cation.
Group H and Group U patients had tumor in ltration into bile ducts signi cantly more frequently than those in Group P. Figure 2A shows that the overall survival of patients in Group P was better than that of those in Group H and Group U (p=0.098, <0.05, respectively). Similarly, Figure 2B shows that disease-free survival of patients in Group P was signi cantly better than that of those in Group H and Group U (p<0.05). All ve cases based on gross classi cation that were classi ed as perihilar IHCC were included in Group H. However, in 47 cases gross classi ed as peripheral IHCC, 14 cases were included in Group P, 16 cases in Group H, and 17 cases in Group U. Comparing these 14 individuals in Group P with the 16 in Group H, those in Group H had lower curability (p=0.031), a higher T-factor (p=0.005), higher clinical stage (p=0.001) ( Table   2), higher incidence of periductal in ltration (p=0.005), and worse prognosis according to disease-free survival (p=0.014) (Fig. 3).

Representative cases
According to typical histological ndings described by Akita and Liau [6,9], perihilar IHCC consists of ductforming and tall columnar tumor cells, as well as an abundant brotic stroma [6,9]. This type of tumor has a clear cytoplasm and tubular components similar to reactive bile ductules at the tumor-liver interface of the invasive front. In contrast, peripheral IHCC consists of cuboidal to low columnar cells, which form irregularly anastomosing tubular architecture with scant mucin. This type of IHCC typically has a hepatoid appearance. Figure 4 shows a representative case. According to the CT images and gross appearance, the tumor appeared to be located in the periphery of the liver (Fig. 4A-D). However, morphologically, the tumor consisted of duct-forming, abundant brotic stroma and little tubular architecture and cancer cells had a clear cytoplasm, which suggested it to be the perihilar type (Fig. 4E). Additionally, immunohistochemical analysis indicated that SSTR2 was expressed in the tumor, but Bcl2 was not (Group H) (Fig. 4F, G).
Considering these observations, this tumor was classi ed as embryological perihilar immunohistologically, but as peripheral based on its gross appearance.
Relationship between the bile duct and IHCC tumor according to 3D imaging Figure 5 shows the relationship between the bile duct and IHCC tumor according to 3D imaging. The CT image was reconstructed using SYNAPSE VINCENT® (Fuji lm, Japan) and we can identify larger vein and artery than the subsegmental branch. The distance between tumor and vessels was evaluated using 3D imaging for 17 cases of IHCC (Group H: 10 cases, Group P: 7 cases). The tumor contacted the bile duct in all cases in Group H (Fig. 5A). However, in Group P, the tumor contacted the bile duct in only four of the seven cases (Fig. 5B). Table 3 shows the results of the univariate analysis of overall survival. The stage, curability, T-factor, lymph node metastasis, tumor thrombus in the portal vein, intrahepatic metastasis, and carbohydrate antigen 19-9 (CA19-9) were found to be signi cant prognostic factors for overall survival, as well as the SSTR2 and Bcl2 expression pattern (Group H and U). Table 3 also shows the results of the multivariate analysis of overall survival. Lymph node metastasis (hazard ratio = 3.091) was identi ed as independent prognostic factors. However, the SSTR2 and Bcl2 expression pattern was not found to be an independent prognostic factor.

Discussion
In this study, we examined the expression of SSTR2 and Bcl2 in IHCC tumors. The data suggested that the expression pattern of these molecules may correlate with the location of the tumors. Our embryological classi cation approach may be considered reasonable clinicopathologically and useful for the classi cation of cases that are di cult to identify histologically.
Our study is based on previous investigations that have described the heterogeneity of normal bile ducts according to their location [10][11][12][13][14][15][16][17][18][19]. Currently, there are no criteria for classi cation of these IHCC tumors based on any molecules that potentially display differential expression when compared with normal bile ducts. As mentioned above, we believe that changes to pathological characteristics will inevitably occur when a bile duct develops into a tumor, which potentially threaten diagnostic precision. For example, S100 protein (S100) is expressed both in tumor tissue and in normal bile ducts [8]. However, there was a report that higher expression levels of S100P was detected in cholangiocarcinoma compared with the benign biliary strictures [23]. Additionally, mucin production also varies during each stage of carcinogenesis [8,24]. To address this effect, we selected SSTR2 and Bcl2. To best of our knowledge, there is no report suggesting that the expression of SSTR2 or Bcl2 relates to tumor malignancy, and we believe that this new method may be useful for IHCC classi cation. According to numerous studies, the prognosis of perihilar IHCC patients is worse than that of peripheral IHCC patients, and there is also an increased incidence of lymph node metastasis in these individuals [3,4,6,7]. These facts demonstrate why careful and accurate classi cation of IHCC type is necessary.
Previous studies have used various criteria to classify IHCC tumors, including gross appearance and imaging ndings [3,4], histopathologic appearance [3,4,6,9], and immunophenotypes [7]. Although the gross appearance and imaging classi cation approach is considered to be the most practical, some tumors derived from the peripheral duct can move into the hepatic portal region, making it di cult to determine the origin of the IHCC using this method. Aishima et al. classi ed IHCC tumors that were smaller than 5 cm in diameter based on gross appearance, de ning perihilar tumors as those involving segmental or larger bile ducts and peripheral tumors as those only affecting smaller ducts [3]. Ruzzenente et al. also classi ed IHCC on the basis of gross appearance, speci cally by the location of the center of the tumor. However, these criteria could not speci cally identify the origin of larger tumors. When this occurred, evaluation was also performed by a pathologist and radiologist [3,4]. These studies suggest that classifying an IHCC tumor based only on gross appearance is a di cult task [4].
Some reports have described histopathological classi cations of IHCC tumors [3,4,6,9]. A publication by Liau et al. stated that the perihilar type is composed of tall columnar cells, while the peripheral type consists of cuboidal to low columnar cells [9]. Akita et al. reported that perihilar tumors display ductal adenocarcinoma, but perihilar ones have tubular components in the central parts of the tumor. Although the authors could classify these IHCC tumors using histopathological criteria, reproducibility is still an issue.
They reported that their standard had a kappa value of less than 0.6, and diagnosing the origin of IHCC is a di cult task, even for some pathologists [6].
Immunohistochemical classi cation, such as our use of SSTR2 and Bcl2 expression, is a relatively simple method for both pathologists and non-pathologists alike. However, some cases still have unclear results and are categorized as indeterminate. In fact, our study had 17 indeterminate cases that were placed in the unclassi ed group (Group U) -tumors that were positive for both SSTR2 and Bcl2 or negative for both SSTR2 and Bcl2. Although characteristics of Group U is similar to Group H according to prognostic factor including T classi cation and the tumor in ltration into bile duct, these 17 cases were di cult to classify into Group P or Group H based on our method. Only 35 cases (67%) of IHCC could be reliably classi ed as perihilar or peripheral, which is the most serious limitation of our study. It is possible that certain characteristics, including expression of several molecules, are altered when normal tissue becomes malignant. In contrast, Hayashi's criteria for classi cation utilizes a scoring system which is based on the mucin productivity and immunophenotype of the cells (S100P, N-cadherin, and neural cell adhesion molecule (NCAM). The authors reported that 98 of 102 cases (96%) could be classi ed into two types of IHCC [7]. Compared with our results, this rate is extremely high. However, scoring the cases after performing immunostaining many times is complicated.
Our results suggest that pathological classi cation may partially correlate with gross classi cation. A limitation of gross classi cation using 3D imaging is that large tumors often appear to be connected to the bile duct in both perihilar and peripheral IHCC. However, 3D imaging may be effective when used in combination with the pathological classi cation method.

Conclusions
Our work demonstrates a novel approach to classify IHCC tumors as peripheral or perihilar based on the expression patterns of SSTR2 and Bcl2 embryologically. Yet, about 30% of cases could not be classi ed by this method. Further research is needed to determine whether adding more molecules to the expression analysis will improve the success of our IHCC classi cation technique. Declarations Ethics approval and consent to participate The protocol for this research project has been approved by a suitably constituted Ethics Committee of the institution and it conforms to the provisions of the Declaration of Helsinki. Committee of the Tokushima University, TOCMS ID: 3215. All informed consent was obtained from the subjects.

Consent for publication
Not applicable Availability of data and materials The datasets generated and analysed during the current study are not publicly available due to protect the individual privacy but are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests. Authors' contributions SY performed immunohistochemistry examination of IHCC patients' liver and evaluate the result and wrote this manuscript.YM analyzed above data and was a major contributor in writing the manuscript. All authors read and approved the nal manuscript. Tables   Table 1. Comparison of various factors among groups of IHCC patients classi ed by SSTR2 and Bcl2 expression