MUC4 is a novel prognostic factor of intrahepatic cholangiocarcinoma-mass forming type

Authors

  • Hiroaki Shibahara,

    1. Division of Human Pathology, Department of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    Search for more papers by this author
  • Shugo Tamada,

    1. Division of Human Pathology, Department of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    Search for more papers by this author
  • Michiyo Higashi,

    1. Division of Human Pathology, Department of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    Search for more papers by this author
  • Masamichi Goto,

    1. Division of Human Pathology, Department of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    Search for more papers by this author
  • Surinder K. Batra,

    1. Departments of Biochemistry and Molecular Biology, Eppley Institute for Research in Cancer and Allied Diseases, University of Nebraska Medical Center, Omaha, NE
    Search for more papers by this author
  • Michael A. Hollingsworth,

    1. Departments of Biochemistry and Molecular Biology, Eppley Institute for Research in Cancer and Allied Diseases, University of Nebraska Medical Center, Omaha, NE
    Search for more papers by this author
  • Kohzoh Imai,

    1. Department of Internal Medicine, Sapporo Medical College, Sapporo, Japan
    Search for more papers by this author
  • Suguru Yonezawa

    Corresponding author
    1. Division of Human Pathology, Department of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    • Division of Human Pathology, Department of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, 8-35-1 Sakuragaoka, Kagoshima 890-8520, Japan
    Search for more papers by this author
    • fax: +81-99-265-7235


Abstract

Complete surgical resection of the tumor is the sole approach to improve the cure rate of patients with intrahepatic cholangiocarcinoma-mass forming type (ICC-MF). Although patients are treated by curative resection, many of them show poor outcome. Mucin (MUC)4 expression has been implicated as a marker for diagnosis and progression of pancreatic adenocarcinomas, but there is no study of the relationship between MUC4 expression and patient's prognosis in ICC-MF. In the present study, we examined the expression profile of MUC4 in ICC-MF tissue from 27 patients using immunohistochemistry. MUC4 was expressed in the carcinoma tissues of 10 (37%) of the 27 ICC-MF tumors, whereas it was not expressed in normal liver tissue. Because MUC4 is an intramembrane ligand for receptor tyrosine kinase ErbB2 and is related with regulation of p27, we also compared the MUC4 expression with ErbB2 and p27 expressions in ICC-MFs. The patients with MUC4 and ErbB2 double positive expression showed a short survival period compared to non-expressing patients. MUC4 and p27 showed no relationship. The univariate analysis showed that tumor size, intrahepatic metastasis, lymph node metastasis, MUC4 expression, and MUC1 expression were statistically significant risk factors affecting the outcome of the patients with ICC-MF. The multivariate analysis demonstrated that MUC4 expression, as well as surgical margin, were statistically significant independent risk factors. In conclusion, the results suggest that expression of MUC4 in ICC-MF is a new independent factor for poor prognosis and is a useful marker to predict the outcome of the patients with ICC-MF. (HEPATOLOGY 2004;39:220–229.)

In intrahepatic cholangiocarcinoma (ICC), 3 types of tumors are classified by their growth pattern: intraductal growth type (ICC-IG), periductal infiltrating type (ICC-PI), and mass forming type (ICC-MF).1 Among ICCs, ICC-MF is the most common tumor type. Complete surgical resection of the tumor is the only way to improve the cure rate of patients with ICC-MF; however, many patients treated by curative resection have a poor outcome. Tumor size, surgical margin, intrahepatic metastasis, lymph node metastasis, vascular invasion, lymphatic invasion, and perineural invasion have been reported as prognostic factors of ICC-MF.2–7

Mucins (MUC) are high molecular weight glycoproteins with oligosaccharides attached to serine or threonine residues of the mucin core protein backbone by O-glycosidic linkages. During the past several years, core proteins for human mucins (MUC1-MUC9, MUC11-13, MUC15-17) have been identified.8–15 Our series of immunohistochemical studies for mucin expression in various human tumors have demonstrated that expression of MUC1 (membrane mucin) is related to invasive proliferation of the tumors and poor outcome of the patients, whereas expression of MUC2 (intestinal type secretory mucin) is related to noninvasive proliferation of the tumors and favorable outcome of the patients.8, 16–18 In our study of intrahepatic bile duct tumors, we demonstrated that patients with ICC-MF showed significantly poorer prognosis than those with the other 2 types (ICC-IG and ICC-PI), and also reported that ICC-MF showed MUC1 positive and MUC2 negative expression. ICC-IG and ICC-PI, as well as bile duct cystadenocarcinoma, showed MUC1 negative and MUC2 positive expression.19 In that study, however, we did not evaluate a relationship between the mucin expression pattern and the patient's outcome within ICC-MF cases.

MUC4 was first reported as tracheobronchial mucin,20 and it is known that MUC4 is expressed in various normal tissues.21–24 Recently, MUC4 was proved to be a novel intramembrane ligand for receptor tyrosine kinase ErbB2,25–27 which is a transmembrane glycoprotein encoded by c-ErbB-2 protooncogene with a tyrosine kinase domain that is highly homologous with the epidermal growth factor receptor.28, 29 Furthermore, MUC4 is related to regulation of p27,30 which is a cyclin-dependent kinase inhibitor involved in the control of G1 and S phases of the cell cycle.31

MUC4 is a membrane mucin,27 like MUC1, which is a poor prognostic factor in various human tumors.8, 17, 19, 32–34 Recently, MUC4 has been reported to be expressed in pancreatic adenocarcinoma,35, 36 whereas MUC4 was not expressed in the normal pancreatic tissue.35–38 Thus, MUC4 is considered to be a tumor marker for pancreatic adenocarcinoma. In biliary epithelial cells, Vandenhaute et al.39 reported that MUC4 was not expressed in the normal epithelium. Lee and Liu40 reported that MUC4 was expressed in 10% of normal controls, whereas it was expressed in 67% of cholagiocarcinomas. However, there is no study of the relationship between MUC4 expression and the patient's prognosis in ICC-MF. In the present study, we expected that MUC4 is also a candidate for one of the poor prognostic factors in ICC-MF, and examined the expression profile of MUC4 in ICC-MF tissues. We found that MUC4 expression is a very useful predictor of poor prognosis in patients with ICC-MF. In addition, ErbB2 and p27 expression were examined to investigate the relationship of their expressions with MUC4 expression in ICC-MF. For the comparison, we also examined MUC1 expression, which has been previously reported as a useful prognostic factor in ICC-MF.19, 33

Abbreviations:

ICC, intrahepatic cholangiocarcinoma; ICC-IG, ICC intraductal growth type; ICC-PI, ICC periductal infiltrating type; ICC-MF, ICC mass forming type; MUC, mucin; Mab, monoclonal antibody; IgG, immunoglobulin G; ABC, avidin-biotinylated horseradish peroxidase; PBS, phosphate buffered saline; PanIN, pancreatic intraepithelial neoplasms.

Patients and Methods

Tissue Samples

Twenty-seven consecutive cases of surgically resected ICC-MF (16 men, 11 women) were retrieved from the files of the Department of Pathology, Faculty of Medicine, Kagoshima University during the period from 1986 to 1999. Seven of the 27 patients were overlapped with the patients studied in our previous study,19 but the other 20 patients were not overlapped. The mean age of the patients was 65.3 years (range: 45–79 years). The study was approved by the Kagoshima University Faculty of Medicine Human Investigation Committee.

In histologic grading,41 5 cases (19%) were well differentiated, 17 cases (62%) were moderately differentiated, and 5 cases (19%) were poorly differentiated. Surgical procedures were as follows: 20 cases (74%) were treated by several types of hepatectomy alone and 7 cases (26%) were treated by hepatectomy with extrahepatic bile duct resection. Lymph node dissection was performed in the 22 cases; lymph node metastasis was positive in 11 cases but negative in the other 11 cases. In the other 5 cases, lymph node dissection was not performed in 4 cases because no swollen lymph node was observed in the surgical areas, and in one case because of peritoneal dissemination and apparent lymph node metastasis. Consequently, in the 27 cases, lymph node metastasis was positive in 12 cases but negative in the other 15 cases. Clinical outcome data were available in these 27 patients with ICC-MF, and overall survival was analyzed in the current study. In the 27 patients, 21 died of carcinoma progression, but no patient died of other diseases. There was no perioperative death. All specimens were fixed in formalin, embedded in paraffin, and cut into 4-μm-thick sections for immunohistochemistry, in addition to the usual hematoxylin and eosin (HE) staining.

Immunohistochemistry

Antibodies.

Immunohistochemistry was performed using the following antibodies. MUC4 was detected by mouse monoclonal antibody (Mab)36 (generated by S.K.B.). ErbB2 was detected by rabbit polyclonal antibody (Dako Cytomation, Glostrup, Denmark). p27 was detected by Mab NCL-p27 (Novocastra, Newcastle, UK). MUC1 expression was examined by Mab DF3 (mouse immunoglobulin G [IgG], Toray-Fuji Bionics, Tokyo, Japan).

Biotinylated affinity-purified horse anti-mouse IgG, goat anti-rabbit IgG, and avidin-biotinylated horseradish peroxidase (ABC) complex were purchased from Vector Laboratories (Burlingame, CA) as the Vectastain Elite ABC Kit.

Staining Procedure.

Immunohistochemical staining was performed with an immunoperoxidase method using the ABC complex as described previously.18, 19, 34 Each section was deparaffinized with xylene. Endogenous peroxidase was blocked by incubating the sections in 0.3% hydrogen peroxidase in absolute methanol at room temperature for 30 min. After hydration in decreasing concentrations of ethanol in water, the sections were washed in 0.01 mol/L phosphate buffered saline (PBS, pH 7.4).

For epitope retrieval of MUC4, a waterbath pretreatment at 80°C for 20 min in 0.01 M citrate buffer (pH 6.0) was performed; for ErbB2, a waterbath pretreatment at 98°C for 40 min was performed, and for p27, an autoclave pretreatment at 120°C for 5 min was performed.

The sections were washed 2 times with PBS. Then, 2% horse or goat serum in PBS was applied for 30 min at room temperature to prevent non-specific staining. In the staining using each antibody, the sections were incubated with dilutions of the primary antibodies (MUC4, 1:3,000; ErbB2, 1:100; p27, 1:40; MUC1, 1:10) in PBS with 1% bovine serum albumin for 16 h at 4°C. The sections were washed 3 times with PBS and incubated with the biotinylated secondary antibodies, and then washed 3 times with PBS. All the sections then received ABC complex for 30 min. After washing with PBS 3 times, the sections were finally reacted with diaminobenzidine substrate for 10 min for visualization, rinsed with tap water, counterstained with hematoxylin, and mounted. Reaction products were not present when non-immune serum or PBS was used instead of the primary antibodies.

Evaluation of the Results by Scoring.

The results of the immunohistochemical staining were evaluated by the percentage of positively stained neoplastic cells. Expression of MUC4 in the cytoplasm of the cells was evaluated. In ErbB2, membranous immunoreactivity was evaluated. In p27, nuclear immunoreactivity was evaluated. For MUC1 expression, expression in the membrane and/or cytoplasm was evaluated. According to our previous study of extrahepatic bile duct carcinoma,34 the percentage of positively stained neoplastic cells were graded as follows: -, less than 5% of neoplastic cells stained; +, more than 5% to less than 20% of neoplastic cells stained; ++, more than 20% to less than 50% of neoplastic cells stained; and +++, more than 50% of neoplastic cells stained.

Statistical Analysis

Statistical analysis was performed using the chi-square test, Fisher exact test, or Spearmann rank-correlation test where appropriate. Survival of the patients was compared between the group with positive MUC4, ErbB2, or MUC1 expression and the group with negative MUC4, ErbB2, or MUC1 expression according the Kaplan-Meier method, and differences between the survival curves were tested using the log-rank test. Survival of the patients was compared between the group with high p27 expression (more than 50%) and the group with low p27 expression (under 50%). Univariate and multivariate survival analysis were performed using the Cox proportional hazards regression model. For the multivariate model, we used .20 as the cutoff P-value to select the analyzed factors from the univariate analysis data. Backward stepwise multivariate analysis was also used to find independent prognostic factors. A probability of P less than .05 was considered statistically significant.

Results

MUC4, ErbB2, p27, or MUC1 Expression in Normal Epithelium of Intrahepatic Bile Duct

MUC4 was not expressed in normal epithelium of the intrahepatic bile duct (Fig. 1A and B). ErbB2 was not expressed in the normal epithelium. p27 was expressed in the nuclei in most of the normal epithelium, and MUC1 was not expressed in the normal epithelium.

Figure 1.

Expression of MUC4, ErbB2, and p27 in ICC-MF (A, C). MUC4 is expressed in the carcinoma, but not expressed in the normal epithelium of intrahepatic bile duct (B). In the carcinoma, MUC4 is expressed in the cytoplasm (D). ErbB2 is expressed in membrane (E), and p27 is expressed in the nuclei (F).

MUC4, ErbB2, p27, or MUC1 Expression Rate in Intrahepatic Cholangiocarcinoma-Mass Forming Type

Positive staining of MUC4 was seen in the cytoplasm of the carcinoma cells in 10 (37%) of 27 ICC-MFs (Fig. 1C and D).

In the 27 ICC-MFs, ErbB2 was expressed in the membrane of the carcinoma cells in 9 (33%) (Fig. 1E), p27 was expressed in the nuclei of the carcinoma cells in 22 (81%) (Fig. 1F), and MUC1 was expressed in the membrane and/or cytoplasm of the carcinoma cells in 22 (81%).

Relationship Between MUC4, ErbB2, p27, or MUC1 Expression and Clinicopathologic Features

The relationship between MUC4, ErbB2, p27, or MUC1 expression and clinicopathologic features is summarized in Table 1. MUC4, ErbB2, and MUC1 expressions were divided into “positive (more than 5%)” and “negative (less than 5%)” according to our previous studies for cholangiocarcinoma and extrahepatic bile duct carcinoma using 5% as the cutoff in MUC protein expression.19, 34 p27 expression was divided into “high (more than 50%)” and “low (less than 50%)” because p27 positive expression rate (more than 5%) was high (81%) in our data; other previous reports42, 43 used 50% as the cutoff in p27 expression for the cholangiocarcinoma.

Table 1. Summary of the Data on the Expression of MUC4, ErbB2, p27 and MUC1 in Clinicopathological Features of 27 Cases of Intrahepatic Cholangiocarcinoma-Mass Forming Type
CategoryNo. patients (%)MUC4ErbB2p27MUC1
Negative <5%Positive ≥5%P ValueNegative <5%Positive ≥5%P valueLow <50%High ≥50%P ValueNegative <5%Positive ≥5%P Value
Age (yrs)             
 ≥6512 (44.4)8 (66.7)4 (33.3)>.99992 (16.7)10 (83.3).21725 (41.7)7 (58.3).70632 (16.7)10 (83.3)>.9999
 <6515 (55.6)9 (60.0)6 (40.0) 7 (46.7)8 (53.3) 5 (33.3)10 (66.7) 3 (20.0)12 (80.0) 
Gender             
 Men16 (59.3)10 (62.5)6 (37.5)>.99994 (25.0)12 (75.0).41056 (37.5)10 (62.5)>.99993 (18.7)13 (81.3)>.9999
 Women11 (40.7)7 (63.6)4 (36.4) 5 (45.5)6 (54.5) 4 (36.4)7 (63.6) 2 (18.2)9 (81.8) 
Tumor size (cm)             
 <49 (33.3)8 (88.9)1 (11.1).09123 (33.3)6 (66.7)>.99994 (44.4)5 (55.6).68313 (33.3)6 (66.7).2950
 ≥418 (66.7)9 (50.0)9 (50.0) 6 (33.3)12 (66.7) 6 (33.3)12 (66.7) 2 (11.1)16 (88.9) 
Surgical margin             
 Negative11 (40.7)7 (63.6)4 (36.4)>.99994 (36.4)7 (63.6)>.99993 (27.3)8 (72.7).44753 (27.3)8 (72.7).3705
 Positive16 (59.3)10 (62.5)6 (37.5) 5 (31.3)11 (68.7) 7 (43.8)9 (56.2) 2 (12.5)14 (87.5) 
Intrahepatic metastasis             
 Negative15 (55.6)11 (73.3)4 (26.7).25666 (40.0)9 (60.0).68286 (40.0)9 (60.0)>.99995 (33.3)10 (66.7).0470
 Positive12 (44.4)6 (50.0)6 (50.0) 3 (25.0)9 (75.0) 4 (33.3)8 (66.7) 0 (0.0)12 (100.0) 
Lymph node metastasis             
 Negative15 (55.6)11 (73.3)4 (26.7).25664 (26.7)11 (73.3).44796 (40.0)9 (60.0)>.99995 (33.3)10 (66.7).0470
 Positive12 (44.4)6 (50.0)6 (50.0) 5 (41.7)7 (58.3) 4 (33.3)8 (66.7) 0 (0.0)12 (100.0) 
Histological grading             
 Well5 (18.5)3 (60.0)2 (40.0).97051 (20.0)4 (80.0).04893 (60.0)2 (40.0).41171 (20.04 (80.0).9885
 Moderate17 (63.0)11 (64.7)6 (35.3) 4 (23.5)13 (76.5) 6 (35.3)11 (64.7) 3 (17.6)14 (82.4) 
 Poorly5 (18.5)3 (60.0)2 (40.0) 4 (80.0)1 (20.0) 1 (20.0)4 (80.0) 1 (20.0)4 (80.0) 
Lymphatic invasion             
 Negative15 (55.6)10 (66.7)5 (33.3).70634 (26.7)11 (73.3).44796 (40.0)9 (60.0)>.99994 (26.7)11 (73.3).3419
 Positive12 (44.4)7 (58.3)5 (41.7) 5 (41.7)7 (58.3) 4 (33.3)8 (66.7) 1 (8.3)11 (91.7) 
Venous invasion             
 Negative11 (40.7)8 (72.7)3 (27.3).44751 (9.1)10 (90.9).04175 (45.5)6 (54.5).68684 (36.4)7 (63.6).1252
 Positive16 (59.3)9 (56.3)7 (43.7) 8 (50.0)8 (50.0) 5 (31.3)11 (68.7) 1 (6.2)15 (93.8) 
Perineural invasion             
 Negative16 (59.3)9 (56.3)7 (43.7).44755 (31.3)11 (68.7)>.99998 (50.0)8 (50.0).12415 (31.2)11 (68.8).0598
 Positive11 (40.7)8 (72.7)3 (27.3) 4 (36.4)7 (63.6) 2 (18.2)9 (81.8) 0 (0.0)11 (100.0) 
Survival (mo.)             
 <1215 (55.6)6 (40.0)9 (60.0).01406 (40.0)9 (60.0).68284 (26.7)11 (73.3).25660 (0.0)15 (100.0).0098
 ≥1212 (44.4)11 (91.7)1 (8.3) 3 (25.0)9 (75.0) 6 (50.0)6 (50.0) 5 (41.7)7 (58.3) 

MUC4 expression in the ICC-MF was not related to most of the clinicopathologic features listed in Table 1. ErbB2 expression was barely related with the differentiated histologic grade and venous invasion. MUC1 expression was barely related to the intrahepatic metastasis and lymph node metastasis. When the patients were divided into 2 groups based on the length of survival (short-term survivors [survival period <12 months] and long-term survivors [survival period ≥12 months]), MUC4 and MUC1 positive expressions were seen in 60.0% and 100%, respectively, of the short-term survivors and 8.3% and 58.3%, respectively, of the long-term survivors. These differences were statistically significant (P = .0140 and .0098, respectively; Table 1).

Relationship Between MUC4, ErbB2, p27, and MUC1 Expression

There was no significant relationship between MUC4 expression and expression of ErbB2, p27, or MUC1 (Table 2).

Table 2. Correlation Between MUC4, ErbB2, p27 and MUC1 Expression in 27 Cases of Intrahepatic Cholangiocarcinoma-Mass Forming Type
 MUC4 Expression
++++++P Value
ErbB2 expression    .3301
 −4221 
 +3010 
 ++7003 
 +++3100 
p27 expression    .7451
 −3020 
 +0000 
 ++4001 
 +++10313 
MUC1 expression    .0654
 −5000 
 +0000 
 ++3001 
 +++9333 

Relationship Between MUC4, ErbB2, p27, or MUC1 Expression and Cumulative Survival Rate

Among the 27 patients examined, 21 patients died during the follow-up period. Median and mean length of survival for patients with ICC-MF who underwent hepatectomy was 10.0 and 20.9 months, respectively. The longest survivor died at 56 months from the surgery, and the longest survivor was alive at 51 months from the surgery in the 6 patients who were alive at the last follow-up. The 1-, 3-, and 5-year survival rates of patients with ICC-MF were 44.4%, 22.6%, and 0%, respectively. The 1- and 3-year survival rates for patients with positive MUC4 expression were 10% and 0%, respectively, and for MUC4 negative expression were 64.7% and 37.0%, respectively.

The survival rate of patients with MUC4 positive expression (10 patients) was significantly poorer than those with MUC4 negative expression (17 patients) (P = .0003, log-rank test; Fig. 2). The survival rate of patients with MUC1 positive expression (22 patients) was significantly poorer than those with MUC1 negative expression (5 patients) (P = .0044, log-rank test; Fig. 3). ErbB2 and p27 expression were not related with survival of 27 patients with ICC-MF.

Figure 2.

Correlation between MUC4 expression and cumulative survival rate in patients with ICC-MF (Kaplan-Meier method). The survival rate of patients with MUC4 positive was poorer than that of patients with MUC4 negative. ○, patients who were alive at the last follow-up.

Figure 3.

Correlation between MUC1 expression and cumulative survival rate in patients with ICC-MF (Kaplan-Meier method). The survival rate of patients with MUC1 positive was poorer than that of patients with MUC1 negative. ○, patients who were alive at the last follow-up.

Survival Related to the Combined Status of MUC4 and ErbB2 Expression

The results of the combined evaluation of MUC4 and ErbB2 expression are shown in Fig. 4. The patients with MUC4 positive and ErbB2 positive expression showed the worst outcome. In contrast, the patients with MUC4 negative and ErbB2 negative expression showed the best outcome. There was a significant difference in the survival between these 2 groups (P = .0279). In addition, the MUC4 and ErbB2 double positive group showed poorer outcome than the MUC4 positive and ErbB2 negative group (P = .0446).

Figure 4.

Combined evaluation of MUC4 and ErbB2. Patients with MUC4 positive and ErbB2 positive expression showed the worst outcome. In contrast, patients with MUC4 negative and ErbB2 negative expression showed the best outcome. There was a significant difference in the survival between the 2 groups (P = .0279). In addition, the MUC4 and ErbB2 double positive group showed poorer outcome than the MUC4 positive and ErbB2 negative group (P = .0446). ○, patients who were alive at the last follow-up.

Survival Related to the Combined Status of MUC4 and p27 Expression

In the combined evaluation of MUC4 and p27 expression, survival of the patients with MUC4 positive and p27 low expression tended to be worse than the patients with MUC4 negative and p27 high expression, although there was no significant difference in the survival between the 2 groups (P = .0545; data not shown).

Survival Related to the Combined Status of MUC4 and MUC1 Expression

The results of the combined evaluation of MUC4 and MUC1 expression are shown in Fig. 5. The patients with MUC4 and MUC1 positive expression showed the worst outcome. In contrast, the patients with MUC4 and MUC1 negative expression showed the best outcome. There was a significant difference in the survival between the 2 groups (P = .0004).

Figure 5.

Combined evaluation of MUC4 and MUC1. Patients with MUC4 positive and MUC1 positive expression showed the worst outcome. In contrast, patients with MUC4 negative and MUC1 negative expression showed the best outcome. There was a significant difference in the survival between the 2 groups (P = .0004). ○, patients who were alive at the last follow-up.

Univariate Analysis of Prognostic Factors

The univariate analysis of prognostic factors of ICC-MF is summarized in Table 3. Tumor size (P = .0318), intrahepatic metastasis (P = .0488), lymph node metastasis (P = .0060), MUC4 expression (P = .0017), and MUC1 expression (P = .0207) were statistically significant risk factors affecting the outcome of the patients with ICC-MF.

Table 3. Univariate Analysis of Prognostic Factors
VariableHazard Ratio95% Confidence LimitsP Value
Age (yrs)   
 ≤65 (n = 12)1  
 >65 (n = 15)0.6780.279–1.644.3897
Gender   
 Men (n = 16)1  
 Women (n = 11)0.8140.331–2.001.6534
Tumor size (cm)   
 <4 (n = 9)1  
 ≥4 (n = 18)3.1221.104–8.827.0318
Surgical margin   
 Negative (n = 11)1  
 Positive (n = 16)2.6610.951–7.446.0623
Intrahepatic metastasis   
 Negative (n = 15)1  
 Positive (n = 12)2.6111.005–6.783.0488
Lymph node metastasis   
 Negative (n = 15)1  
 Positive (n = 12)5.1521.600–16.595.0060
Histological grading   
 Well (n = 5)1  
 Moderate (n = 17)1.3330.427–4.160.6205
 Poorly (n = 5)0.9570.210–4.360.9547
Lymphatic invasion   
 Negative (n = 15)1  
 Positive (n = 12)2.0010.811–4.941.1324
Venous invasion   
 Negative (n = 11)1  
 Positive (n = 16)1.5730.629–3.931.3327
Perineural invasion   
 Negative (n = 16)1  
 Positive (n = 11)1.4930.602–3.701.3867
MUC4 expression   
 Negative (<5%) (n = 17)1  
 Positive (≥5%) (n = 10)5.2721.863–14.924.0017
ErbB2 expression   
 Negative (<5%) (n = 9)1  
 Positive (≥5%) (n = 18)0.9130.364–2.290.8454
p27 expression   
 Low (<50%) (n = 10)1  
 High (≥50%) (n = 17)1.3720.520–3.619.5232
MUC1 expression   
 Negative (<5%) (n = 5)1  
 Positive (≥5%) (n = 22)11.6841.455–93.858.0207

Multivariate Analysis of Prognostic Factors

The multivariate analysis of prognostic factors of ICC-MF is summarized in Table 4. MUC4 expression (P = .0269), as well as surgical margin (P = .0267), were found to be statistically significant independent risk factors.

Table 4. Multivariate Analysis of Prognostic Factors
VariableHazard Ratio95% Confidence LimitsP Value
Tumor size (cm)   
 <4 (n = 9)1  
 ≥4 (n = 18)2.037.334–12.416.4405
Surgical margin   
 Negative (n = 11)1  
 Positive (n = 16)6.0701.232–29.920.0267
Intrahepatic metastasis   
 Negative (n = 15)1  
 Positive (n = 12)2.655.776–9.081.1196
Lymph node metastasis   
 Negative (n = 15)1  
 Positive (n = 12)1.092.253–4.717.9060
Lymphatic invasion   
 Negative (n = 15)1  
 Positive (n = 12)2.310.570–9.360.2409
MUC4 expression   
 Negative (n = 17)1  
 Positive (n = 10)4.5601.190–17.478.0269
MUC1 expression   
 Negative (n = 5)1  
 Positive (n = 22)4.536.292–70.336.2797

Backward Stepwise Multivariate Analysis of Prognostic Factors

By the backward stepwise multivariate analysis of 7 prognostic factors listed in Table 4, surgical margin (P = .0027), intrahepatic metastasis (P = .0149), lymphatic invasion (P = .0290), and MUC4 expression (P = .0008) were statistically significant independent risk factors (Table 5).

Table 5. Backward Stepwise Multivariate Analysis of Prognostic Factors
VariableHazard Ratio95% Confidence LimitsP Value
Surgical margin   
 Negative (n = 11)1  
 Positive (n = 16)9.2192.161–39.339.0027
Intrahepatic metastasis   
 Negative (n = 15)1  
 Positive (n = 12)3.7651.295–10.945.0149
Lymphatic invasion   
 Negative (n = 15)1  
 Positive (n = 12)3.6211.141–11.492.0290
MUC4 expression   
 Negative (n = 17)1  
 Positive (n = 10)7.2722.274–23.251.0008

Discussion

The expression of MUC4 has been reported to be a candidate for a tumor marker of pancreatic adenocarcinoma.35, 36 However, there is no study of the relationship between MUC4 expression and the patient's prognosis in ICC-MF. The present study demonstrated that MUC4 was expressed in several cases of ICC-MF, although it was not expressed in the normal liver tissue, including normal intrahepatic bile duct. We found that MUC4 expression was significantly related to poor survival of the patients with ICC-MF using the Kaplan-Meier analysis, and that the MUC4 expression is a statistically significant independent poor prognostic factor in patients with ICC-MF, that is, the outcomes after surgery were significantly poorer in the patients with MUC4 positive expression than in those with MUC4 negative expression.

Tumor size, surgical margin, intrahepatic metastasis, lymph node metastasis, vascular invasion, lymphatic invasion, and perineural invasion have been reported as prognostic factors in patients with ICC-MF.2–7 The univariate analysis for prognostic factors in the present study showed that tumor size, intrahepatic metastasis, lymph node metastasis, MUC4 expression, and MUC1 expression were significant prognostic factors. In the multivariate analysis, MUC4 expression, as well as surgical margin, were significant poor prognostic factors. Also, by the backward stepwise multivariate analysis, MUC4 expression, as well as surgical margin, intrahepatic metastasis, and lymphatic invasion, were statistically significant independent risk factors. Among these factors, MUC4 expression was the most significant factor.

MUC4 is normally expressed in normal tissues of the respiratory tract, stomach, small intestine, colon, and endocervix.21–24 MUC4 has been recently proved to be a novel intramembrane ligand for receptor tyrosine kinase ErbB2.25–27 Ascites sialoglycoprotein-2, the transmembrane subunit of MUC4, showed specific binding to ErbB2, and MUC4 binding to ErbB2 induced tyrosine phosphorylation of the receptor.26, 27 MUC4 is the only ligand that has been characterized for ErbB2, which was stated to be an orphan receptor with no known ligand.25, 27, 44 Furthermore, although MUC4 alone failed to activate mitogen-activated kinase or protein kinase B/Akt of the phosphatidyl inositol 3-kinase pathway, MUC4 expression induced up-regulation of p27. However, a combination of MUC4 and neuregulin down-regulated p27 and activated protein kinase B/Akt.27, 30 In our present immunohistochemical study, MUC4 expression was not significantly related with ErbB2 or p27 expression in ICC-MF. From the comparison of the 3 factors in the present immunohistochemical study, these 3 proteins might be controlled and regulated under unknown mechanisms in ICC-MF.

In ICC, ErbB2 expression has been reported.45–48 ErbB2 is not expressed in normal biliary epithelium.46, 47 The combined evaluation of MUC4 and ErbB2 expression revealed that the patients with MUC4 and ErbB2 double positive expression had a significantly worse outcome than those with MUC4 and ErbB2 double negative expression. Moreover, the MUC4 and ErbB2 double positive group had a poorer outcome than the MUC4 positive and ErbB2 negative group. Aishima et al.47 reported that there was no significant difference in survival between ErbB2 positive and negative patients. However, our results indicate that ErbB2 expression influences the survival in the condition of MUC4 positive expression.

p27 expression in ICC also has been reported.42, 43, 49 p27 is expressed in normal biliary epithelium.42, 43, 49 Fiorentino et al.42 and Taguchi et al.43 reported that p27 expression was a prognostic factor in ICC. In the present study, however, p27 expression was not a significant prognostic factor. The combined evaluation of MUC4 and p27 expression showed no significant result.

Carraway et al.27 proposed a model for the participation of ErbB2 and MUC4 as a switch in epithelial differentiation and proliferation in a relatively complex manner. Their model of proliferation by the combination of MUC4, ErbB2, and neuregulin was partly supported by our study, which demonstrated that MUC4 and ErbB2 double negative cases showed better survival than MUC4 and ErbB2 double positive cases. In addition, their differentiation model triggered by the combination of MUC4 and ErbB2 (in the absence of neuregulin) was unlikely in ICC-MF because there was no relationship between MUC4 and p27 expression in our study. The role of neuregulin for the proliferation of ICC-MF would be an interesting future area of study.

Recently, Swartz et al.36 reported that MUC4 is detected in pancreatic adenocarcinoma and pancreatic intraepithelial neoplasms (PanIN), and the expression of MUC4 increases with increasing grade of PanIN. They considered that the expression pattern of MUC4 in PanIN supports the progression model for pancreatic adenocarcinoma. However, we could not find lesions like PanIN around the nests of ICC-MF.

Matsumura et al.33 reported that MUC1 expression in ICC-MF is a poor prognostic factor. We confirmed that in the univariate analysis in our present study, although MUC4 expression in ICC-MF is a more significant poor prognostic factor compared with MUC1 expression in the multivariate analysis. On the other hand, combined evaluation of MUC4 and MUC1 expression may be useful clinically to predict the outcome in patients with ICC-MF. In the report by Matsumura et al.33 and in our observation, MUC1 was expressed in the luminal surface membrane and/or in the cytoplasm. In contrast, MUC4 expression showed cytoplasmic pattern only but did not show membrane pattern. Also in invasive pancreatic adenocarcinoma, the expression pattern of MUC4 showed cytoplasmic pattern.36 We confirmed the cytoplasmic expression pattern in our study of invasive pancreatic adenocarcinoma (unpublished data). Like MUC1, MUC4 is a membrane mucin; however, MUC4 acts as a different mechanism from MUC1. For signaling molecules, MUC1 acts as a docking protein, whereas MUC4 acts as a receptor ligand.27 Differences of the expression patterns might suggest the possibility of a different mechanism between MUC1 and MUC4, both of which are membrane mucins possessing cell signaling function.

In conclusion, we report that expression of MUC4 in ICC-MF is an independent poor prognostic factor and is a useful marker to predict the outcome of patients with ICC-MF who had surgical resection of the tumor. Patients with ICC-MF showing positive MUC4 expression should be followed-up carefully.

Acknowledgements

The authors are grateful to Dr. Tomofumi Hamada for his suggestions for statistical analysis, and Mr. Yoshiharu Atsuji, Ms. Yoshiko Arimura, and Ms. Yukari Nishimura for their excellent technical assistance.

Ancillary