Main

Extrahepatic bile duct carcinoma (EBDC) is a relatively uncommon tumor worldwide, but is more common in Japan than in Western countries.1 Despite recent progress in diagnostic and therapeutic procedures, its prognostic outcome remains unfavorable.2, 3 A number of EBDCs arise in bile ducts with an abnormal pancreatico-biliary duct junction, and their tumorigenesis is considered to be associated with inflow of pancreatic juice into the bile duct.4, 5, 6, 7 However, more common mechanisms for the tumorigenesis and the reasons for the biological aggressiveness of EBDCs have not yet been sufficiently elucidated, and further investigations are required.

In recent years, O-linked glycoproteins, such as mucin-type glycoproteins, have been reported to be associated with the differentiation and malignant potential, such as altered cell adhesion, invasion, metastasis, recurrence and prognosis, of adenocarcinomas.8 O-glycosylation of these glycoproteins has been considered to play a key role in determining the tumor characteristics. The initial glycosylation of mucin-type O-linked glycoproteins is catalyzed by a family of uridine diphosphate (UDP) N-acetyl-alpha-D-galactosamine:polypeptide N-acetylgalactosaminyl transferases (GalNAc-Ts).9, 10, 11, 12, 13, 14, 15 A total of 15 human GalNAc-Ts have been identified to date,16 and observed differences in the O-linked glycoprotein expression patterns among tumors, which may affect tumor differentiation and progression, can be explained by differing expression levels or activities of these GalNAc-T subtypes.8, 11 Among the GalNAc-T subtypes, the expression of GalNAc-T3 is highly specific for glandular epithelia, and it is also expressed at high levels in human adenocarcinomas.17, 18 Furthermore, recent studies have suggested a relationship between GalNAc-T3 expression and the tumor differentiation or progression, including prognosis, of human adenocarcinomas in the colon, lung, stomach, pancreas and gallbladder.24, 25, 26, 27, 28, 29, 30 Although the GalNAc-T subtypes are usually localized in the Golgi apparatus, their subcellular localization alters following malignant transformation due to reorganization of the Golgi apparatus elements.8, 19, 20 Therefore, not only the levels but also the patterns of GalNAc-T3 expression, such as the subcellular distribution, may be associated with the malignant transformation and differentiation of adenocarcinoma cells.

In the present study, the clinicopathological findings of 61 surgically resected EBDCs were evaluated in detail using whole stepwise sections, and their GalNAc-T3 expression levels and patterns were studied by immunohistochemistry using formalin-fixed paraffin-embedded sections. The correlations between the immunohistochemical expression patterns of GalNAc-T3 and the clinicopathological factors, mode of progression and prognosis were investigated to clarify the aggressiveness of EBDCs in association with GalNAc-T3 expression.

Materials and methods

Tissue Specimens

Surgical specimens were obtained from 61 patients who had undergone surgery for primary EBDCs. For pathological examination, the surgically resected specimens were fixed in 10% formalin at room temperature. After recording of the gross features of the tumors, such as the tumor size or growth pattern, each whole specimen was cut into slices at intervals of 0.5 cm, and all the slices were routinely processed, embedded in paraffin, sectioned and microscopically evaluated for the histopathological characteristics. All the specimens in this study were diagnosed and filed as primary EBDCs at the Department of Anatomic Pathology, Graduate School of Medicine, Kyushu University during the period from June 1989 to September 2001.

The clinicopathological findings were evaluated according to the General Rules for Surgical and Pathological Studies on Cancer of the Biliary Tract21 with reference to the WHO classification,22, 23 and staged according to the TNM classification of the UICC.23

Immunohistochemical Staining

Immunohistochemistry was performed on paraffin-embedded tissue sections using the streptavidin-biotin-complex method. Sections (3-μm thick) of the EBDC specimens were deparaffinized in xylene and rehydrated through a graded ethanol series. After blocking of endogenous peroxidase activity by incubation in methanol containing 0.3% H2O2 for 30 min, antigen retrieval was performed by heating in 10 mM citrate buffer (pH 6.0) at 99°C for 20 min in a microwave oven. Next, nonspecific binding was blocked by a 10-min incubation with 10% goat serum in phosphate-buffered saline, and the slides were incubated with a rabbit polyclonal anti-GalNAc-T3 antibody (1:3000 dilution) at 4°C overnight. The antigens labeled by the primary antibody were detected with a Histofine kit (Nichirei Pharmaceutical, Tokyo, Japan) and visualized using 3,3′-diaminobenzidine tetrahydrochloride as a chromogen, followed by counterstaining with hematoxylin. Some breast carcinoma specimens were stained in parallel as positive controls, while nonimmunized rabbit IgG was used instead of the primary antibody for negative controls. All the slides were evaluated by two observers (TI and TE) who had no prior knowledge of the clinicopathological information.

Statistical Analysis

Statistical analyses were performed using StatView (Abacus Concepts Inc., Berkeley, CA, USA). Fisher’s exact test and Mann–Whitney’s U-test were used to analyze the correlations between GalNAc-T3 expression and clinicopathologic features. Kaplan–Meier methods with the log-rank test were used to calculate the overall survival rate and differences in the survival curves. The results were considered to be statistically significant when the P-values were less than 0.05.

Results

Clinicopathological Features of the EBDCs

The study patients were 46 males and 15 females ranging in age from 45 to 85 years (mean age, 66.1 years). The clinicopathological findings based on detailed examination of whole stepwise sections of each EBDC according to the General Rules for Surgical and Pathological Studies on Cancer of the Biliary Tract and the TNM classification are summarized in Table 1. Overall, 57 (93%) of the 61 EBDCs were massively invasive, while only three (5%) and one (2%) EBDCs were noninvasive and minimally invasive (invading the fibromuscular layer), respectively. In addition, two of the three noninvasive EBDCs showed a pedunculated gross appearance. Overall, the EDBCs exhibited marked lymphatic permeation (47/61 cases, 77%), venous invasion (38/61 cases, 62%), perineural invasion (49/61 cases, 80%) and frequent lymph node metastases (28/61 cases, 46%). Regarding tumor differentiation, 42 of the total 61 EBDCs (69%) were well differentiated (papillary or well differentiated tubular) at the luminal surface and 42 of the 58 minimally or massively invasive EBDCs (72%) were poorly differentiated (moderately to poorly differentiated tubular and undifferentiated) at the invasive front. The gross and histopathological appearances of representative cases of massively invasive and noninvasive EBDCs are shown in Figures 1 and 2, respectively.

Table 1 Clinicopathological factors of the EBDCs (n=61)
Full size table
Figure 1
figure 1

Representative case of a massively invasive EBDC. (a) Gross appearance and (b) mapping of the carcinoma based on histological examination of stepwise sections (Ca: carcinoma; CBD: common bile duct; Pap Vater: papilla of Vater; Panc head: pancreas head; Du: duodenum). (c) Cut surface and (d) histological features of the carcinoma. The carcinoma involves the whole thickness of the bile duct wall, and invades the surrounding fibroadipose tissue. Histologically, a well to poorly differentiated tubular adenocarcinoma is recognized.

Full size image
Figure 2
figure 2

Representative case of a noninvasive EBDC. (a) Gross appearance and (b) mapping of the carcinoma based on histological examination of stepwise sections (Ca: carcinoma; CBD: common bile duct; MPD: main pancreatic duct; Panc head: pancreas head; Du: duodenum). (c) Cut surface and (d) histological features of the carcinoma. Grossly, the tumor is pedunculated. Histologically, a well-differentiated papillary or tubular adenocarcinoma is recognized. The carcinoma is confined to the mucosa of the bile duct, and stromal invasion is not recognized.

Full size image

Immunohistochemical Expression and Localization of GalNAc-T3 in EBDCs

In the current study, the immunohistochemical localizations and patterns of GalNAc-T3 expression in EBDCs were investigated. The immunohistochemical patterns of GalNAc-T3 were classified into two types, namely granular and diffuse types, based on the predominant expression pattern and subcellular distribution. The granular type revealed GalNAc-T3 staining that was strong, granular and mostly restricted to the perinuclear areas, whereas the diffuse type revealed GalNAc-T3 staining that was diffusely but weakly positive in the cytoplasm. Figures 3 and 4 show representative examples of the granular and diffuse types of GalNAc-T3 staining pattern in EBDCs, respectively. The GalNAc-T3 expression patterns at the luminal surface and invasive front were evaluated separately if the carcinoma was invasive, and classified as the granular or diffuse type when over 50% of the cancerous epithelia in each region of the section showed the corresponding expression pattern.

Figure 3
figure 3

Representative expression pattern of GalNAc-T3 in an EBDC case. In this case, the granular-type expression at the luminal surface (top) remains the granular type at the invasive front (bottom).

Full size image
Figure 4
figure 4

Representative expression pattern of GalNAc-T3 in an EBDC case. In this case, the granular-type expression at the luminal surface (top) alters to the diffuse type at the invasive front (bottom).

Full size image

All four noninvasive or minimally invasive EBDC cases showed the granular type (Figure 5). Among the 58 minimally or massively invasive EBDC cases, the GalNAc-T3 expression pattern at the luminal surface was the granular type in 38 cases (66%) and diffuse type in 20 cases (34%), while the expression pattern at the invasive front was the granular type in 26 cases (45%) and diffuse type in 32 cases (55%). The 38 cases with the granular type at the luminal surface were divided into 26 cases of the granular type (68%) and 12 cases of the diffuse type (32%) at the invasive front. All 20 cases of the diffuse type at the luminal surface showed diffuse-type expression at the invasive front, and none of them revealed the granular-type expression at the invasive front (Figure 6).

Figure 5
figure 5

Granular-type expression pattern of GalNAc-T3 in a representative case of a noninvasive EBDC.

Full size image
Figure 6
figure 6

Alterations in the GalNAc-T3 expression patterns at the luminal surfaces and invasive fronts of the EBDCs.

Full size image

Relationships between the GalNAc-T3 Expression Patterns and Clinicopathological Features of EBDCs

The correlations between the GalNAc-T3 expression patterns at the luminal surface and invasive front and the clinicopathological features in EBDCs are summarized in Table 2.

Table 2 Relationships between GalNAc-T3 expression and clinicopathologic features of the EBDCs
Full size table

All four noninvasive or minimally invasive carcinoma cases showed only the granular type, but the GalNAc-T3 expression pattern at the luminal surface did not affect the depth of invasion (T number of the TNM classification).

Cases with diffuse-type expression at the luminal surface tended to have a positive lymph node status, but the association did not reach statistical significance (P=0.055). No other apparent correlations between the different GalNAc-T3 expression patterns and any of the other clinicopathological factors, such as tumor differentiation, depth of tumor invasion and angio- or perineural invasion, were detected.

The difference in GalNAc-T3 expression pattern at the invasive front did have an effect on the lymph node status of the EBDC cases. The diffuse type of GalNAc-T3 expression was associated with positive lymph node metastasis (P=0.020), but no correlations with any of the other clinicopathological factors, such as tumor differentiation, depth of tumor invasion and angio- or perineural invasion, were detected.

Relationships between Overall Survival in EBDCs and the Different GalNAc-T3 Expression Patterns

The 5-year overall survival rate of all the EBDC patients was 37.7%. Among the invasive EBDC patients, no statistically significant relationships between the overall survival and the different GalNAc-T3 expression patterns were detected. There was a slight tendency that the overall survival of patients with diffuse-type expression at the invasive front was better than that of patients with the granular type in contrast to the relationships between the different GalNAc-T3 expression patterns and the lymph node status (Figure 7).

Figure 7
figure 7

Relationships between the survival outcome and GalNAc-T3 expression patterns at the invasive fronts of the EBDC cases evaluated by Kaplan–Meier’s survival curves.

Full size image

Discussion

The mechanisms for the tumorigenesis, progression and biological aggressiveness of EBDCs have not yet been clarified. However, recent studies have suggested that GalNAc-T3 expression is associated with the differentiation or biological behavior of human adenocarcinomas through the initial glycosylation of mucin-type O-linked glycoproteins. In the present study, the relationships between the GalNAc-T3 expression patterns and the tumor characteristics of EBDCs were investigated in order to clarify the biological aggressiveness of EBDCs.

The immunohistochemical expression patterns of GalNAc-T3 in our EBDC cases were classified into two types, namely the granular and diffuse types, based on the predominant expression pattern and subcellular distribution. Although most previous studies evaluating GalNAc-T3 expression by immunohistochemistry have divided the observed expression patterns into high-intensity (strong, positive) and low-intensity (weak, negative) groups,24, 25, 26, 27, 28, 30, 31 Miyahara et al29 divided the GalNAc-T3 expression patterns of gallbladder carcinomas into granular and diffuse types, mostly corresponding to high-intensity and low-intensity expression, respectively. We also found that the immunohistochemical staining of GalNAc-T3 in EBDCs showed granular-type and diffuse-type expression patterns, and these features of EBDCs were similar to those reported for gallbladder adenocarcinomas.29 Furthermore, our observations that all the noninvasive or minimally invasive carcinomas showed only the granular type and that the proportion of diffuse-type expression was increased at the invasive front of advanced EBDCs were also similar to the findings for gallbladder adenocarcinomas. The GalNAc-T family enzymes normally reside in the Golgi apparatus of cells,19 and alterations in the subcellular localization of GalNAc-T3 may be associated with the status of the morphological and functional differentiation of the carcinoma cells. The diffuse-type localization of GalNAc-T3 in carcinoma cells indicates a reorganization of the Golgi apparatus elements in the carcinoma cells as reported previously.8, 19, 20 Our data reveal further similarities of the GalNAc-T3 expression in EBDCs with that in gallbladder carcinomas, in that EBDCs alter their GalNAc-T3 expression from the granular type to the diffuse type during tumor growth. The present study also demonstrates that EBDCs are less differentiated and show diffuse-type GalNAc-T3 expression more frequently at their invasive front. However, there was no significant correlation between the GalNAc-T3 expression and tumor differentiation. The above facts indicate that altered subcellular localization of GalNAc-T3 cannot be explained solely by differences in the tumor differentiation of EBDCs.

In our investigation, GalNAc-T3 was weakly to moderately expressed with a granular pattern in noncancerous bile duct epithelium, compared with the strong positivity of noncarcinomatous colorectal epithelium and negativity of breast glandular tissue reported in previous studies.18, 24 Colorectal carcinomas showed a lower intensity of GalNAc-T3 expression than normal colorectal epithelium,24 while breast carcinomas showed a higher intensity than normal breast glandular tissue.18 Furthermore, in the current study, the diffuse type of GalNAc-T3 expression in EBDCs was associated with positive lymph node metastasis. In contrast, differences in GalNAc-T3 expression did not have any significant effect on the overall survival of EBDC patients, although there was a slight tendency that the overall survival of patients with diffuse-type expression was better than that of patients with granular-type expression. In the adenocarcinomas of some organs, including the pancreas and gallbladder, the presence of weak expression or diffuse-type localization of GalNAc-T3 was correlated with tumor aggressiveness.24, 25, 26, 27, 28, 29 On the other hand, early-stage gastric carcinomas with submucosal invasion tended to show positive (strong or granular) GalNAc-T3 expression, while early-stage undifferentiated gastric carcinomas showing positive GalNAc-T3 expression were associated with significantly higher frequencies of metastatic lymph node involvement.30 In addition, GalNAc-T3-positive esophageal squamous cell carcinomas exhibited more aggressive behavior.31 It has been reported that some O-linked carbohydrate or mucin antigens affect the metastatic processes of malignant tumors, such as escape from the primary tumor, vessel invasion and migration.8 The initial glycosylation of O-linked carbohydrate antigens is regulated by differential expression of GalNAc-T subtypes. Therefore, differential GalNAc-T expression may also play a key role in the metastatic processes of malignant tumors. However, since the functions of the GalNAc-T family, as well as their associations and mechanisms in the metastatic processes of malignant tumors, have not yet been completely elucidated, it is currently difficult to explain how the various correlations between different GalNAc-T expression patterns and the biological behavior of malignant tumors in various organs occur. Differences in the results from EBDCs and other types of carcinomas are partly, but not completely, explained by the dissimilar characters of the carcinomas themselves. Further investigations into the GalNAc-T3 expression patterns in the carcinomas of different organs as well as the functions of this enzyme are necessary.

Regarding the GalNAc-T3 expression in EBDCs from the clinical aspect, we consider that, if a preoperative biopsy is taken, evaluation of the GalNAc-T3 expression in sections of the biopsy is useful for predicting the lymph node status preoperatively. Our present results have demonstrated that cases with diffuse-type expression at the luminal surface tend to have a positive lymph node status, while those with diffuse-type expression at the invasive front are significantly correlated with positive lymph node metastasis. All the cases with diffuse-type expression at the luminal surface also showed diffuse-type expression at the invasive front. Therefore, when a preoperative EBDC biopsy specimen shows the diffuse type of GalNAc-T3 expression pattern, the patient is predicted to carry lymph node metastasis with a higher probability than patients with granular-type GalNAc-T3 expression, and it is necessary to plan extensive lymph node dissection during the surgical treatment. Therefore, preoperative evaluation of GalNAc-T3 expression in EBDCs is useful for predicting the lymph node status and planning surgical procedures.

In conclusion, expression of GalNac-T3 in EBDCs was evaluated and compared with the clinicopathologic features in the present study. The immunohistochemical expression patterns of GalNAc-T3 in EBDCs were classified into two types, namely granular and diffuse types. EBDCs altered their GalNAc-T3 expression during tumor growth, and this difference in the GalNAc-T3 expression pattern was considered to be associated with the mode of tumor progression, such as lymph node metastasis. Clinically, a preoperative evaluation of GalNAc-T3 expression is considered to be useful for planning operative procedures.