Abstract
Aim: To compare the cells of mucosal extension (ME) and radial invasion (RI) in hilar cholangiocarcinoma (HCCA) for optimal resection. Materials and Methods: Forty-six patients underwent surgery for HCCA between 1992 and 2004. Immunohistochemical expressions of p53, Ki-67, matrix metalloproteinase-7 (MMP7), mucin 1 (MUC1), and E-cadherin were assessed at five different sites of the tumour and compared between the recurrence and non-recurrence groups. Results: Expression of E-cadherin was significantly lower in RI cells than in ME cells, and that of MMP7 and MUC1 was significantly higher in RI cells than in ME cells. Ki-67 expression was higher in ME cells than in RI cells. During the 11-year follow-up, recurrence in patients with R0 resection was associated with significantly lower E-cadherin, higher MMP7, and higher Ki-67 expression. Conclusion: Removal of as many RI cells as possible should be a priority in resection of HCCA, followed by removal of ME cells. E-Cadherin appears to be associated with recurrence of HCCA.
Tumour cells of hilar cholangiocarcinoma (HCCA) invade radially into surrounding organs [defined as radial invasion (RI)] and extend along the mucosa (defined as [mucosal extension (ME)] (1, 2). Patients with RI cells remaining at the surgical margin after resection have poorer outcomes than those with remaining ME cells (3-7).
In the present study, five different sites of the tumour and its adjacent tissues were selected based on the dysplasia–carcinoma sequence: the tumour cells of the centre; the tumour RI and ME cells; the cells of dysplasia; and the cells of normal mucosa. The degree of malignancy of these sites was evaluated using five basic immunohistochemical markers (p53, Ki-67, matrix metalloproteinase-7 (MMP7), mucin1 (MUC1), and E-cadherin), which have been reported to be associated with malignancy in pancreatobiliary cancer (8-21). The sites of ME that were longer than 2 cm were defined as superficial spread (SS) (5) and assessed separately because the cell behaviour might differ and the procedures for resection may need to be different from that for short ME. The associations of these markers with recurrence were also assessed over the 11-year follow-up.
Patients and Methods
Patients. The subjects of this study were 46 patients (33 men, 13 women) diagnosed as having HCCA who underwent bile duct resection with or without hepatectomy at the Department of Gastroenterological Surgery, Yokohama City University Hospital between July 1992 and September 2004. The surgical procedures were bile duct resection alone in five patients, hepatectomy with bile duct resection in 36 patients, pancreatoduodenectomy with bile duct resection in three, and hepatopancreatoduodenectomy in two. Fourteen patients underwent vascular resection. R0 resection was performed in 32 patients (69.6%). R1 resection was performed in 14 patients (30.4%), of whom three had ME at the margin, and 11 had RI at the margin. There was no significant bile leakage during surgery. Perioperative mortality occurred within 2 months in three patients. The subjects were classified according to the International Union Against Cancer pTNM classification (22). Resected specimens of the 46 patients were reviewed retrospectively, and the recurrence status of these patients was evaluated as of December 2014 (Table I). The study was approved by the Institutional Review Board of Yokohama City University Hospital (# HI-31), and informed consent was obtained from all participants.
Specimens. The resected bile ducts were opened longitudinally from the distal margin to the proximal margin. The tumours were classified macroscopically as papillary, nodular, and flat type, and histologically as papillary with or without SS, nodular with or without SS, and flat type with or without SS (Figure 1A). The histological types of the tumours were classified as papillary, well-differentiated, moderately differentiated, or poorly differentiated adenocarcinoma, and mucinous or adenosquamous carcinoma (Table I) (23, 24).
Examined sites. Five different sites of the tumour and adjacent tissues were selected based on the dysplasia–carcinoma sequence and were examined histologically using haematoxylin and eosin staining (Figure 1B). They were classified using a textbook definition (24).
The molecular mechanisms of cholangiocarcinoma development differ by anatomical location (proximal/distal bile duct) in the biliary tree (25). Distal cholangiocarcinomas and intrahepatic cholangiocarcinomas were excluded, focusing on perihilar cholangiocarcinomas.
Immunohistochemical staining for p53, Ki-67, MMP7, MUC1, and E-cadherin. Tissue specimens were fixed in 20% buffered formalin overnight after resection, embedded in paraffin, sliced into 3-μm-thick sections, and mounted on MAS-GP type A-coated glass slides (Matsunami Glass Ind., Ltd., Osaka, Japan). Immunohistochemical staining was performed by the avidin-biotin complex method on sections from each specimen. Endogenous peroxidase was blocked by incubating the sections in 0.03% hydrogen peroxidase in methanol at room temperature for 30 min. After washing three times with phosphate-buffered saline (PBS), pH 7.2, for 5 minutes, sections were blocked with 10% normal rabbit serum in PBS for 15 minutes at 37°C. The blocking serum was decanted, and the different primary antibodies for p53 (clone DO-7, mouse monoclonal, diluted 1:100; Dako, Glostrup, Denmark), Ki-67 (MIB-1, mouse monoclonal, diluted 1:100; Dako), MMP7 (141-7B2, mouse monoclonal, diluted 1:500; Kyowa Pharma Chemical Co., Ltd., Toyama, Japan), MUC1 (Ma695, mouse monoclonal, diluted 1:100; Leica Biosystems Newcastle Ltd, Newcastle Upon Tyne, UK), and E-cadherin (HECD-1, mouse monoclonal, diluted 1:100; Takara Bio Inc., Shiga, Japan) in PBS were applied, respectively. They were applied overnight at 4°C for p53, for 15 min at 37°C for Ki-67 and MUC1, for 60 minutes at 37°C for MMP7, and overnight at 4°C for E-cadherin. After the slides were washed three times in PBS for 5 min, the Dako EnVision+ System (Dako) was used for labelling. After the final PBS rinse, tissue sections were treated with 0.06% diaminobenzidine (DAB) for 5 minutes, rinsed, counterstained with haematoxylin, dehydrated, cleared with xylene, and coverslipped.
Counting. Cells were considered positively stained when the following was observed: unequivocal diffuse brown nuclear staining for p53 and Ki-67 (8); cytoplasmic staining for MMP7 (9); and staining of the cytoplasm and luminal border for MUC1 (10). For E-cadherin, cells were considered negatively stained when membranous expression was markedly reduced compared to non-cancerous bile ducts (11). The number of positively stained cells out of 300 cells at each of the five sites was counted for each markers and recorded as labelling indices, representing the percentage of positively stained cells at each of the sites for each stain. Within each of the five sites, the area that had a substantial number of positively stained cells was selected for counting and obtaining the labelling indices. When the number of cells eligible for assessment was insufficient, the labelling indices were calculated from the number of obtainable cells. When there was no significant immunohistochemical expressions of p53, Ki-67, MMP7, MUC1, and E-cadherin at any of the sites, they were considered as having no expression. Thus, the average indices of the five markers at each of the five sites of the 46 patients were obtained for each stain.
Statistics. Labelling indices are expressed as means±standard error. Statistical analysis was carried out with one-way analysis of variance and the Bonferroni method to compare average percentage scores (labelling indices) among each of the five sites and among each of the three macroscopic types, and with the Mann–Whitney U-test to compare average percentage scores between the groups with recurrence and non-recurrence. When the p-value was less than 0.05, the difference was considered significant. All statistical analyses were performed using SPSS statistical software version 22.0 for Windows (IBM Corp., Armonk, NY, USA).
Results
Forty-six resected specimens from patients with HCCA were investigated. The histologic types of the tumours are shown in Table I. Of the 46 patients, 11 did not have eligible ME cells, and five did not have eligible cells of dysplasia histologically. The number of patients who did not have significant immunohistochemical expression of p53, Ki-67, MMP7, MUC1, and E-cadherin was 20, 1, 6, 5, and 1, respectively.
Average labelling indices at five sites. The results of the average labelling indices of p53, Ki-67, MMP7, MUC1, and E-cadherin at the different sites are summarized in Figure 2. There were significant differences between RI and ME in MMP7, MUC1, and E-cadherin, indicating the malignancy of RI. There were significant differences between RI and D in all p53, Ki-67, MMP7, MUC1, and E-cadherin. Significant difference between RI and C was seen in MMP7.
Average labelling indices at superficial spread. The average SS scores for p53, Ki-67, MMP7, MUC1 and E-cadherin were 8.4%±5.6%, 46.6%±12.4%, 10.2%±5.5%, 17.9%±6.3%, and 51.6%±10.6%, respectively. Sato et al. classified intraepithelial neoplasms into two categories, biliary intraepithelial neoplasia-3 (BilIN-3) and intraepithelial spread (IES), and they advocated the importance of distinguishing these two categories (26). In this study, BilIN-3 was seen in 28 patients, and IES was seen in seven. The average scores for p53, Ki-67, MMP7, MUC1, and E-cadherin in BilIN-3 were 7.1%±3.7%, 42.4%±5.7%, 8.8%±3.7%, 15.4%±3.5%, and 49.3%±5.5%, respectively, and the average scores for p53, Ki-67, MMP7, MUC1, and E-cadherin in IES were 15.3%±6.3%, 47.3%±9.7%, 9.0%±5.1%, 24.2%±10.2%, and 34.6%±12.9%, respectively. The p53 score was higher in IES than BilIN-3, but there was no significant difference.
Average marker scores in the recurrence and non-recurrence groups. Recurrence was found in 24 patients in the R0 resection group and 10 patients in the R1 resection group. Ten patients had remaining cancer cells at the RI site, and three patients had cancer cells remaining at the ME site (Table II).
There were associations between the markers and recurrence. After R0 resection, E-cadherin expression at RI sites was significantly lower in the recurrence group than in the no recurrence group (p=0.008), and expression of MMP7 and Ki-67 at RI was significantly higher in the recurrence group than in the no recurrence group (p=0.04 and p=0.03, respectively). Patients in the group with abdominal dissemination had significantly lower E-cadherin and higher MMP7 and Ki-67 expression at RI sites (Table III).
In the present study, p53, Ki-67, MUC1, E-cadherin, and MMP7 were not significantly associated with overall survival.
Discussion
ME/SS cells had high proliferative potential associated with high expression of Ki-67, but little invasive potential. RI cells had a higher degree of malignancy than ME/SS cells, demonstrated by reduced cell–cell adhesion associated with low expression of E-cadherin, invasive potential associated with high expression of MMP7, and strong malignancy associated with high expression of MUC1 and p53. The recurrence group had reduced expression of E-cadherin and high expression of MMP7 and Ki-67 at RI. Thus, RI cells should be resected first, followed by as many ME cells as possible.
The cells of the invasive front have been demonstrated to bear molecular changes associated with malignancy, such as high expression of membrane-type 1 (MT1)-MMP in breast cancer (27), MUC1 in colorectal cancer (28), ectopic laminin in gastric cancer (29), alpha smooth muscle actin in oral squamous cell carcinoma (30), and tumour-associated macrophages in HCCA (31). However, the biological difference between RI and ME in HCCA has not been sufficiently clarified.
The average E-cadherin score was significantly lower in RI cells than in ME cells (12.5% vs. 46.4%; p<0.001) in the present study. Reduced expression of E-cadherin has been reported to be associated with low survival, lymph node metastasis, high tumour stage, and strong blood vessel invasion in extrahepatic cholangiocarcinoma (12); it is an independent and significant prognostic factor along with N-cadherin and S100A4 (13), as well as an early event in tumourigenesis in both BilIN and intraductal papillary neoplasm of the bile duct (IPNB) lineages (14).
The average MMP7 scores were significantly higher in RI cells than in ME cells (p<0.001). High MMP7 expression has been reported to be significantly associated with advanced TNM stage, especially with T and N factors, and perineural invasion, a poor prognostic factor (9). It as an independent and significant prognostic factor, along with intrahepatic metastasis (15), and is an indicator of invasive growth and poor outcome in cholangiocarcinoma (16).
The average MUC1 scores were significantly higher in RI cells than in ME cells (p<0.001). High MUC1 expression is associated with poor survival in patients with cholangiocarcinoma (10, 17), is predictive of overall survival along with fascin, epidermal growth factor receptor (EGFR), MUC4, and p27 in resected extrahepatic cholangiocarcinoma (18), and is associated with poorer survival with the presence of tumour invasion and margin-positive resection in a study of resected IPNBs (19).
The average Ki-67 score in the present study was higher, although not significantly, in ME cells than in RI cells (p=0.12) (Figure 2B). High Ki-67 (MIB-1) expression in resected and biopsy specimens of cholangiocarcinoma has been associated with lymphatic invasion, obstruction of stent patency, and poor survival (20, 21) and is a useful predictor of lymph node metastasis (8). This suggests that ME cells do not always lack aggressiveness, but they may have strong proliferative potential associated with Ki-67.
Clinically, many reports support resection of RI cells. Combined hepatectomy and resection of hilar vessels for en bloc resection of HCCA has now become standard. Portal vein resection has come to contribute to long-term survival in HCCA (32). However, hepatic artery resection has not yet been adequately justified (33, 34).
Cancer cells remaining at ME sites after resection of cholangiocarcinoma have been shown to be associated with late local recurrence (3,7). We would consider that it is the high Ki-67 expression at ME sites that is associated with late recurrence, and that, as the present results suggest, ME where Ki-67 is highly expressed should be removed as much as possible considering the patient's status.
Recurrence was apparently associated with reduced E-cadherin expression and high MMP7 and Ki-67 expression at RI sites in the R0 resection group. The average E-cadherin score at RI sites was significantly lower in the recurrence than in the no recurrence group (p=0.008). The average scores for MMP7 and Ki-67 at RI and those of Ki-67 at ME were higher in the recurrence group than in the no recurrence group (p=0.04, p=0.03, and p=0.053, respectively). In the R1 resection group, E-cadherin expression was lower in the recurrence than in the no recurrence group. Reduced E-cadherin expression may be associated with strong malignancy and may be a cause of recurrence, especially abdominal dissemination, even after R0 resection. Ki-67 was highly expressed at both RI and ME in the recurrence group.
Long ME sites, or SS sites, were also assessed. They are occasionally found, and additional resection becomes necessary to accomplish R0 resection, but deciding the resection line may be difficult (35). The present results show that SS cells have almost the same immunohistochemical expression as ME. The average p53, Ki-67, MMP7, MUC1, and E-cadherin scores at SS sites were 8.4%, 46.6%, 10.2%, 17.9%, and 51.6%, respectively. There were no significant differences between SS and ME sites in the expressions of these markers. The p53 score was higher at IES than at BilIN-3 (p=0.64), but there was no significant difference in the present study, consistent with Sato et al.'s report (26).
There are, however, some uncertain elements in this study. Staining does not actually show uniformly, and the labelling index is a semi-quantitative evaluation. To exclude irrelevant factors, unnecessary tissues were carefully excluded by haematoxylin and eosin staining, and evaluations were performed by several staff members. Postoperative therapies given were diverse, and HCCA recurrence was evaluated with long-term follow-up. Despite these difficulties, the difference in malignancy between RI and ME/SS has been clarified.
In conclusion, ME has high Ki-67 expression, indicating proliferative potential. Removal of RI cells is a priority in cases where complete resection cannot be carried out in HCCA, and as many ME cells as possible should also be removed.
Acknowledgements
The Authors would like to thank Arata Ogihara for providing statistical support, Harumi Sakurada for providing technical support, and Yukari Nakamura for providing support.
- Received July 26, 2017.
- Revision received August 9, 2017.
- Accepted August 11, 2017.
- Copyright© 2017, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved