Introduction

Extrahepatic cholangiocarcinoma (EHCC) is a malignant epithelial neoplasm of the biliary tract, from the hepatic hilum to the distal bile duct, that accounts for 70–90 % of all cholangiocarcinomas [1]. EHCC is a relatively uncommon cancer in Western countries but is more prevalent in Eastern Asian countries, including Korea [1–3]. Surgical resection of the tumor is the only curative therapeutic modality for patients with EHCC but can be applied only in a limited number of patients with localized or locally advanced disease [4]. The 5-year survival rate of EHCC patients following surgical resection is approximately 20 % [5]. Several neoadjuvant therapies, including chemotherapy, radiation therapy, and photodynamic therapy, have been extensively studied, but none have yielded significant survival advantages for EHCC patients [1, 6]. Therefore, identification of new biomarkers for early detection and/or development of new therapeutic regimens based on a better understanding of the biological mechanisms of this deadly disease are essential for decreasing mortality among EHCC patients.

Epidermal growth factor receptors (EGFRs) are members of a membrane receptor protein tyrosine kinase family that includes HER1 (ErbB1/EGFR), HER2 (ErbB2), HER3 (ErbB3), and HER4 (ErbB4). After binding ligands, EGFRs are activated by receptor homo- or heterodimerization, resulting in the phosphorylation of various downstream substrates, which mediate intracellular signal transduction. EGFR signaling pathways play an important role in regulating cell division, proliferation, differentiation, and migration [7]. The EGFR family member HER3 lacks intrinsic protein tyrosine kinase activity [8]. However, binding of neuregulin family ligands enables HER3 to form functional heterodimers with EGFR or HER2 that act through the cytoplasmic domain of HER3 to exert tyrosine phosphorylation activity [9]. Thus, by heterodimerizing with other EGFR proteins, HER3 can participate in diverse signal transduction cascades and regulate various cellular processes [7, 9]. The development of tyrosine kinase inhibitors of EGFR or HER2 was initially considered a promising strategy for blocking EGFR or HER2 pathways in several solid cancers, including breast and lung cancers. However, evidence suggests that HER3 may be responsible for the resistance to these therapeutic regimens, which often develops [10]. Overexpression of HER3 has been reported in several malignant neoplasms in gastrointestinal tract organs, including gastric [11–13], colorectal [14–19], and ampulla of Vater [18] cancers. Although a few studies have demonstrated HER2 expression in cholangiocarcinoma, showing that it is relatively infrequent in these tumors [20–23], there have been no comprehensive studies of HER2 and HER3 expression in EHCCs. In the current study, we analyzed HER2 and HER3 protein expression in 230 EHCC cases using tissue microarray immunohistochemistry. We demonstrate that HER3 expression in EHCC constitutes a prognostic factor.

Materials and methods

Patients and tumor samples

The study population consisted of 230 patients with surgically resected EHCCs treated at Asan Medical Center, University of Ulsan College of Medicine, in Seoul, South Korea, between 1991 and 2005. Carcinomas with an epicenter in the extrahepatic bile duct were included in this study, whereas carcinomas from the ampulla of Vater or pancreas were not. Carcinomas arising in the gallbladder or intrahepatic bile duct with extension to the extrahepatic bile duct were also excluded. Information regarding the age and gender of patients, surgical procedure, survival time, and survival status was obtained by reviewing EHCC patients’ medical records. Data pertinent to location, size, and growth pattern of the tumor were obtained from surgical pathology reports. Materials were obtained with appropriate human protection approval from the Institutional Review Board of the Asan Medical Center (Project Number 2011–0734). Information regarding post-operative radiation, chemotherapy, and performance status of patients was not analyzed in this study.

Tissue microarray construction

Tissue microarrays were constructed from archival formalin-fixed, paraffin-embedded tissue blocks as previously described [24]. Briefly, 230 extrahepatic cholangiocarcinoma cases and 12 normal extrahepatic bile ducts were included. A representative area was carefully selected for each tumor or normal biliary epithelia from a hematoxylin-and-eosin-stained section of a donor block. Each case was represented by two to four 1.5-mm-diameter cores.

Immunohistochemistry and scoring

Immunohistochemistry was performed on 4-μm-thick tissue microarray sections as previously described [25]. Briefly, tissue sections were deparaffinized and hydrated in xylene and serially diluted ethanol, respectively. Endogenous peroxidase was blocked by incubation in 3 % H2O2 for 10 min. Antigen retrieval was performed in a steam pressure cooker using preheated antigen retrieval buffer, pH 6 (Dako, Glostrup, Denmark), at 95 °C for 10 min. Non-specific binding of antibodies was minimized by incubating sections with Protein Block (Dako) for 15 min. Microarrays were incubated at room temperature for 30 min with rabbit polyclonal anti-HER2 (c-erbB2, A0485; 1:750 dilution; Dako) or overnight at 4 °C with mouse monoclonal anti-HER3 (RTJ.2, 1:500 dilution; Santa Cruz, CA, USA). Antigen–antibody reactions were detected with an LSAB + peroxidase kit (Dako) and 3,3´-diaminodbenzidine (Dako). Negative controls were composed of identically treated histologic sections with the omission of primary antibodies. Immunostained sections were lightly counterstained with hematoxylin, dehydrated in ethanol, and cleared in xylene. We did not perform immunohistochemistry for Her2 and Her3 on full sections. Tissue core representativity might be a concern, but certainly for 1.5-mm cores previous studies have demonstrated that two to four tissue cores are representative with 95–97 % concordance rate [62].

HER2 protein expression was scored on a scale of 0 to 3 using the following gastric cancer staging system [26]: 0, no reactivity or membranous reactivity in less than 10 % of cells; 1+, faint/barely perceptible membranous reactivity in ≥10 % of cells or reactivity in only part of the cell membrane; 2+, weak to moderate complete or basolateral membranous reactivity in ≥10 % of tumor cells; 3+, strong complete or basolateral membranous reactivity in ≥10 % of tumor cells. Cases receiving a HER2 score of 3+ were considered positive for HER2 expression.

The results of immunohistochemical staining for HER3 were scored based on the intensity of staining as 0 (negative), 1 (weak), or 2 (strong), and the percentage of positive epithelial cells as 0 (<5 %), 1 (6–25 %), 2 (26–50 %), 3 (51–75 %), or 4 (>76 %). A Histo-score was generated as the product of intensity and area. The Histo-score was then dichotomized into no/lower expression (Histo-score, 0–6) and overexpression (Histo-score, 8).

Statistical analysis

Statistical analyses were performed using SPSS version 17 (SPSS Inc., Chicago, IL, USA). Associations between categorical variables were examined using Pearson’s chi-square and Fisher’s exact tests. Survival curves were calculated by the Kaplan–Meier method, and statistical significance was evaluated using the log-rank test and the Cox proportional hazards regression model. P < 0.05 was considered as statistically significant.

Results

Clinicopathological characteristics of cases

The clinicopathological characteristics of cases are summarized in Table 1. Patient age ranged from 30 to 84 years (mean, 60.9 years). Of the 230 patients, 164 were men and 66 were women. Tumor size ranged from 0.5 to 6 cm (mean, 2.5 cm). Thirty-one cases were pT1 tumors, 84 were pT2, 91 were pT3, and 24 were pT4. The length of patient follow-up time ranged from 1 to 127 months, and median survival time at last follow-up was 22 months.

Table 1 Clinicopathologic characteristics of patients with EHCC
Full size table

HER2 and HER3 expression

HER2 protein expression patterns were analyzed in tumors from a total of 224 EHCC patients using tissue microarray immunohistochemistry. Information regarding six patients was excluded from the HER2 analysis because their tissue microarray cores were either detached or folded during the HER2 immunohistochemical staining procedure. Representative images of the expression of HER2 and HER3 in EHCC tumors and normal biliary epithelial cells are shown in Fig. 1. All 12 samples of normal biliary epithelial cells did not show immunoreactivity for either HER2 or HER3. HER3 was predominantly distributed in a cytoplasmic pattern, whereas HER2 was mainly localized to the cell membrane in EHCC cases. HER2 and HER3 overexpression was observed in 6 % (13/224 cases) and 39 % (90/230 cases) of EHCCs, respectively. Only the histological subtype was significantly associated with HER2 overexpression (P = 0.02). Specifically, intraductal papillary neoplasms with associated invasive carcinoma cases showed significantly more HER2 overexpression than other histological subtypes, including tubular adenocarcinoma, adenosquamous carcinoma, intestinal type adenocarcinoma, and mucinous carcinoma, which did not express HER2 (Table 2). HER3 protein was more commonly overexpressed in nodular and infiltrative tumors than in papillary tumors (P = 0.03). HER3 overexpression was more frequent in cases without than in cases with invasion of the pancreas (P = 0.02).

Fig. 1
figure 1

Representative immunohistochemical staining of HER2 and HER3 in EHCC. Normal bile duct epithelia show negativity for HER2 (a) and HER3 (b). Cholangiocarcinoma show strong and complete membranous expression of HER2 (c) and diffuse and strong cytoplasmic HER3 immunopositivity (d)

Full size image
Table 2 Comparison between HER2 and HER3 immunohistochemical staining results and clinicopathologic factors in EHCC
Full size table

Among 13 HER2-overexpressing EHCCs, seven (54 %) showed HER3 overexpression. There was a significant correlation between cases with HER2 overexpression and those with HER3 overexpression (R = 0.24, P < 0.001) but not between HER2 and HER3 co-overexpression and clinicopathological parameters (Table 3).

Table 3 Comparison of results of immunohistochemical staining for HER2 and HER3 co-expression with clinicopathologic factors in EHCC
Full size table

Survival analysis

Median survival of patients with HER3 overexpression (24.0 months) was significantly worse than that of patients with no/lower HER3 expression (31.9 months; P = 0.01, log-rank test; Fig. 2a). The 1-, 3-, and 5-year survival rates in the HER3-overexpression group were 82.2, 22.6, and 13.9 %, respectively, whereas the corresponding rates in the no/lower HER3-expression group were 82.6, 40.5, and 31.2 %.

Fig. 2
figure 2

Survival curves stratified by HER2 and HER3 expression in EHCC patients. a Median survival in patients in the HER3-overexpression group (24.0 months) was significantly shorter than that for patients in the no/lower HER3-expression group (31.9 months; P = 0.01, log-rank test). b Survival was not significantly different between patients in the HER2-overexpression group (25.9 months) and the no/lower HER2-expression group (27.8 months; P = 0.17, log-rank test). c Median survival in EHCC patients with both HER2 and HER3 co-overexpression (24.8 months) was marginally different from that in patients without HER2 and HER3 co-overexpression (28.0 months; P = 0.08)

Full size image

There was no significant difference in survival between patients with HER2 overexpression (median survival, 25.9 months) and those without HER2 expression (median survival, 27.8 months; P = 0.17, log-rank test; Fig. 2b).

Median survival of EHCC patients with both HER2 and HER3 overexpression (24.8 months) tended to be lower than that in patients without HER2 and HER3 co-overexpression (28.0 months; P = 0.08; Fig. 2c), but this did not reach statistical significance.

Association between survival and other clinicopathological factors

The relationships between other clinicopathological variables and survival are summarized in Table 4. Of these additional clinicopathological variables, tumor differentiation status (P < 0.001), pT classification (P = 0.01), lymph node metastasis (P < 0.001), liver invasion (P = 0.01), and vascular invasion (P = 0.02) were also significantly associated with survival. In contrast, survival was not associated with gender (P = 0.38), tumor location (P = 0.47), growth pattern (P = 0.09), pancreatic invasion (P = 0.24), duodenal invasion (P = 0.07), perineural invasion (P = 0.36), or resection margin status (P = 0.10).

Table 4 Univariate analysis of pathologic features affecting survival in EHCC patients
Full size table

Multivariate analysis of clinicopathological factors

The independent prognostic significance of HER3 expression as well as of other clinicopathological parameters was determined by applying the Cox proportional hazards model (Table 5). HER3 overexpression (P = 0.007), differentiation (P = 0.002), liver invasion (P = 0.03), and lymph node metastasis (P = 0.002) remained as prognostic factors.

Table 5 Multivariate analysis for prognosis in EHCC
Full size table

Discussion

In this study, we observed overexpression of HER2 and HER3 protein in a subset of EHCCs, with about 6 % of EHCC cases overexpressing HER2 (13/224) and 39 % (90/230) overexpressing HER3. In addition, we found HER3 overexpression to be an independent prognostic factor in EHCC patients. HER3 is unique among members of the HER receptor tyrosine kinase family in that it lacks intrinsic tyrosine kinase activity but it does contain six consensus phosphotyrosine sites, which bind the SH2 domain of the three regulatory subunits of phosphoinositide-3-kinase (PI3K) [27, 28]. Binding of HER3 can activate the PI3K/Akt signaling pathway, which is a critical regulator of many cellular processes. HER2-mediated transformation of epithelial cells is associated with the activation of the PI3K/Akt signaling pathway in breast cancer. Several targeted therapeutics, including trastuzumab, a monoclonal antibody against the extracellular domain of HER2, and lapatinib, an EGFR/HER2 tyrosine kinase inhibitor, have been approved for use in the treatment of HER2-overexpressing breast cancers. It has been proposed that inhibition of HER3 phosphorylation and blockade of the PI3K/Akt signaling pathway are required in order to obtain a tumor suppression effect despite differences in the operating mechanisms of these therapeutic agents [29]. HER2–HER3 heterodimerization plays a key role in HER2-mediated transformation, tumor progression, and drug resistance [29]. Previous studies on loss of HER3 expression in HER2-dependent cells have demonstrated reduced PI3K signaling and decreased cell proliferation, suggesting that HER3 is essential for HER2-driven carcinogenesis [30, 31]. Treatment of HER2-amplified tumors with tyrosine kinase inhibitors results in a compensatory increase in HER3 expression, membranous HER3 localization, and decreased HER3 dephosphorylation [32]. MET-dependent phosphorylation of HER3 has been identified as one mechanism of resistance to EGFR inhibitors in lung cancer [33]. A recent study demonstrated that blocking both HER2 and HER3 inhibits the PI3K/Akt pathway more effectively than blocking HER2 or HER3 alone, suggesting that also HER3 should be inhibited in patients with HER2- and PI3K-dependent cancers in order to completely block the PI3K/Akt pathway [29]. Our group previously demonstrated that the Akt pathway is activated in a large proportion of EHCCs (84 %) [24]. In the present study, we demonstrate that HER3 is overexpressed in 39 % of EHCC cases. Considering the current study and the results of previous studies of breast cancers and EHCCs, we hypothesize that application of a therapeutic regimen for blocking HER3, such as AMG888, together with HER2 inhibitors and PI3K inhibitors might be effective in a subset of EHCC patients with activated Akt and HER3.

HER3 overexpression has been identified in malignant neoplasms from several organs, including melanoma, and prostate, colorectal, lung, ovarian, and gastric cancers [17, 30, 34–38]. In addition, HER3 overexpression has been reported to be associated with a worse outcome in melanomas and lung and gastric cancers [34, 38, 39]. Although HER3 is specifically required for HER2-driven tumorigenesis and has been extensively studied in the HER2-amplified subtype of breast cancers, other types of cancer also show higher HER3 expression without accompanying HER2 overexpression [30, 40]. In these tumors, HER3 may function as an allosteric activator of other members of the HER receptor tyrosine kinase family, such as EGFR [41–43]. EGFR expression has been reported in 8–57 % of cholangiocarcinomas, and patients with EGFR-positive tumors showed worse clinical outcome than those with EGFR-negative cancers [20, 23, 44, 45]. The absence of a survival difference in patients with HER2 overexpression and the significantly shorter survival in patients with HER3 overexpression, as demonstrated in this study, suggest a possible link between shorter survival time in the HER3-overexpression group and EGFR–HER3 heterodimerization in a subset of these patients. These patients may benefit from a therapeutic regimen that includes other types of HER inhibitors, including erlotinib, cetuximab, and lapatinib, which have been investigated recently in cholangiocarcinomas [46–50].

HER2 overexpression in cholangiocarcinomas has been explored in various studies and has shown to vary widely (between 5 and 76 %) [20, 51–57]. These discrepancies might be explained by the lack of a standardized methodology, different standards of interpretation, or differences in tumor location. In recent reports, the HER2 expression rate in resected EHCCs was reported to be 5 to 10 % [20, 23, 45, 52]. The approximately 6 % HER2 expression rate identified in the current study is in agreement with these previous studies. Consistent with some previous reports, we found a higher HER2 overexpression rate in intraductal papillary neoplasm with an associated invasive carcinoma than in other histological types [20, 45, 54]. Although the HER2 overexpression rate is not exceptionally high in EHCCs, a subset of patients with HER2 overexpression might benefit from HER2-targeted monoclonal therapy.

The recently updated (7th edition) American Joint Committee on Cancer Staging Manual divided EHCCs into perihilar and distal bile duct cancers [58]. The term “distal bile duct cancers” was applied to cancers distal to the insertion of the cystic duct. This division reflects the growing evidence for distinct biological patterns in perihilar and distal EHCCs [59–61]. However, in our study, the proportions of HER2 and HER3 overexpression were similar in perihilar and distal bile duct cancer groups.

In conclusion, HER2 and HER3 proteins are overexpressed in subsets of EHCCs. HER3 overexpression is correlated with decreased patient survival and therefore constitutes a prognostic factor and a potential therapeutic target in patients with EHCCs.