Expression of UDP-N-acetyl-α-D-galactosamine–polypeptide galNAc N-acetylgalactosaminyl transferase-3 in relation to differentiation and prognosis in patients with colorectal carcinoma
Version of Record online: 28 MAR 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 7, pages 1939–1946, 1 April 2002
How to Cite
Shibao, K., Izumi, H., Nakayama, Y., Ohta, R., Nagata, N., Nomoto, M., Matsuo, K.-i., Yamada, Y., Kitazato, K., Itoh, H. and Kohno, K. (2002), Expression of UDP-N-acetyl-α-D-galactosamine–polypeptide galNAc N-acetylgalactosaminyl transferase-3 in relation to differentiation and prognosis in patients with colorectal carcinoma. Cancer, 94: 1939–1946. doi: 10.1002/cncr.10423
- Issue online: 28 MAR 2002
- Version of Record online: 28 MAR 2002
- Manuscript Accepted: 16 NOV 2001
- Manuscript Revised: 12 NOV 2001
- Manuscript Received: 25 MAY 2001
- Ministry of Education, Science, Sports, and Culture of Japan
- Princess Takamatsu Cancer Research Fund
- University of Occupational and Environmental Health
- UDP-N-acetyl-α-D-galactosamine–polypeptide GalNAc N-acetylgalactosaminyl transferase-3;
- colon carcinoma;
- rectal carcinoma;
- colorectal carcinoma;
- cell differentiation;
Tumor development usually is accompanied by alterations of O-glycosylation. Initial glycosylation of mucin-type, O-linked proteins is catalyzed by one of the UDP-GalNAc–polypeptide N-acetyl-galactosaminyl transferases, such as GalNAc-T3, which is expressed in adenocarcinoma cells. The authors investigated whether such expression influenced tumor differentiation or prognosis in patients with colorectal carcinoma.
The expression of GalNAc-T3 was evaluated immunohistochemically in 106 paraffin embedded samples from surgically resected colorectal carcinomas and was related to patient and tumor characteristics. Western blot analysis was performed on seven samples of frozen tissue.
Strong tumor expression of GalNAc-T3 predicted 5-year survival in patients with colorectal carcinoma (67.2% vs. 43.6% for weak expression; P = 0.017). GalNAc-T3 expression was not associated with age, gender, tumor size, tumor location, or disease stage but was related to histologic differentiation (P = 0.049) and depth of invasion (P = 0.031). Univariate analysis showed that strong GalNAc-T3 expression significantly enhanced the likelihood of survival. Multivariate Cox survival analysis identified enzyme expression as an independent prognostic factor that was second only to TNM stage.
GalNAc-T3 expression is a novel and useful indicator of tumor differentiation, disease aggressiveness, and prognosis in patients with colorectal carcinoma. Cancer 2002;94:1939–46. © 2002 American Cancer Society.
Tumor development usually is associated with alterations in cell-surface carbohydrates. Recently, it has been shown that carbohydrate antigens, such as carcinoembryonic antigen, CA19-9, sialyl Tn, and Tn, are useful for monitoring tumor status in patients with malignant disease.1–3 Mucin-type, O-linked carbohydrates may constitute up to 80% of the total amount of these glycoproteins. O-glycosylation has been implicated in a variety of cellular functions in tumors, such as the binding of cell-adhesion molecules, invasion, metastasis, and cell differentiation.4–8 Initial glycosylation of mucin-type, O-linked protein is catalyzed by one of the UDP-GalNAc–polypeptide N-acetyl-galactosaminyl transferases (GalNAc-T).9–11 To date, seven distinct human GalNAc-T genes have been cloned and characterized. O-glycosylation is regulated in part by differential expression of GalNAc-T in tumor cells.9, 12–15 Among these, GalNAc-T3 is expressed in tissues that contain epithelial glands.10–13
Using a polyclonal antibody to GalNAc-T3, we previously showed that GalNAc-T3 is expressed in adenocarcinoma cell lines but not in other carcinoma cell lines.16 Although our previous immunohistochemical results suggested that an anti-GalNAc-T3 antibody may be useful diagnostically in patients with breast carcinoma, no report has considered GalNAc-T3 expression in association with long-term patient survival in patients with any type of malignant disease. In the current study, the expression of GalNAc-T3 was examined in colorectal carcinoma specimens, and the loss of GalNAc-T3 expression was associated with a poor outcome.
MATERIALS AND METHODS
The specimens studied were obtained from 106 patients who underwent surgery for primary colorectal carcinoma between 1980 and 1985 in the Department of Surgery I at the School of Medicine of the University of Occupational and Environmental Health, Kitakyushu, Japan. Patients included 63 men and 43 women who ranged in age from 29 years to 82 years (median, 59 years). Survival data were available for a median of 108 months (range, 0.3–220.0 months), and the follow-up rate was 100%. The TNM classification system was used to define clinical stage and depth of tumor invasion. Histologic typing followed the World Health Organization classification system.17
Polyclonal antibody against GalNAc-T3 was generated by multiple immunizations of a New Zealand white rabbit using synthetic peptides, as described previously.16 Antibody was diluted in phosphate-buffered saline (PBS) containing 2% bovine serum albumin. A dilution of 1:3000 was used to immunostain paraffin embedded sections, and a 1:5000 dilution was used for Western blot analysis.
All resected specimens were processed for routine surgical pathologic evaluation by the same pathologist who was blinded to the GalNAc-T3 staining. A section of normal colorectal epithelium taken from the proximal or distal resection margin, at least 5 cm distant from macroscopically evident tumor, served as a control. Formalin fixed, paraffin embedded materials were cut at a thickness of 2 μm and stained. Immunohistochemistry was performed by a streptavidin-biotin-peroxidase complex method.18 Sections were pretreated twice for 5 minutes per treatment with 0.01 mol/L citrate buffer, pH 6.0, at 100 °C in a microwave oven. Endogenous peroxidase activity was blocked by incubation in 3% H202 in absolute methanol for 5 minutes. Each section was preincubated in sheep serum for 10 minutes. Sections then were incubated in polyclonal GalNAc-T3 antibody diluted 1:3000 for 60 minutes, followed by antirabbit immunoglobulin for 10 minutes, and then streptavidin-biotinylated horseradish peroxidase complex (Nichirei, Tokyo, Japan) for 5 minutes. Finally, diaminobenzidine was used as a chromogen, and sections were counterstained with hematoxylin. PBS was substituted for the primary antibody in a negative control. Paraffin embedded sections of breast tumor were used as positive controls for GalNAc-T3.16
Each slide was assessed independently by two investigators who were unaware of clinical outcomes. Interobserver agreement with regard to the scoring of immunostaining exceeded 95%. In each slide, the staining was evaluated by counting the frequency of labeled cells in 5 high-power fields containing roughly 100 cells each. This count was performed using the following criteria: 0, no positive staining; +, staining in < 25% of tumor cells; ++, staining in 25–50% of tumor cells (0, +, and ++ indicated weak GalNAc-T3 expression); +++, staining in 50–75% of tumor cells; and ++++, staining in 75–100% of tumor cells (+++ and ++++ indicated strong GalNAc-T3 expression). The intensity of staining in all tumors was compared with staining in histologically normal colonic mucosa from the same specimens.
Western Blot Analysis
All tissue samples were kept frozen at −80 °C. After total protein was extracted from available frozen tumor samples and normal mucosal samples with mercaptoethanol treatment, Western blot analysis was performed, as described previously.19, 20 Equal amounts of protein were subjected to 7.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis; then, proteins were transferred electrophoretically to a 0.45-μm polyvinylene difluoride membrane (Immobilon; Millipore, Bedford, MA). Detection of specific proteins was performed according to a previously described procedure using an enhanced chemiluminescence detection kit (ECL Plus; Amersham International, Buckinghamshire, United Kingdom).
Chi-square tests and Mann–Whitney U tests were used for comparison of clinicopathologic features of tumors with strong and weak GalNAc-T3 expression. Survival rates were estimated by using the method of Kaplan and Meier21 and were compared using the log-rank test. Various parameters were analyzed for prognostic significance by univariate and multivariate Cox proportional hazards methods22 using an SAS software package (SAS Institute, Cary, NC). Significance was set at the P = 5% level.
Clinicopathologic Features and GalNAc-T3 Expression
To investigate associations between GalNAc-T3 expression and clinicopathologic features, tumors were divided into two groups: tumors with strong GalNAc-T3 expression and tumors with weak GalNAc-T3 expression. Table 1 summarizes the clinicopathologic data for all 106 tumors, and immunohistochemical data are summarized together with histologic types in Table 2.
|Variable||No. of patients (%) (n = 106)|
|Age (yrs) (mean ± SD)||60.6 ± 11.0|
|Well differentiated||53 (50.0)|
|Moderately differentiated||44 (41.5)|
|Poorly differentiated||2 (1.9)|
|Depth of invasiona||4 (3.8)|
|Histologic type||Strong GalNAc-T3 expression|
|No. of patients||%|
Intense, granular, cytoplasmic GalNAc-T3 staining was observed in glandular epithelial cells of normal colorectal mucosa (Fig. 1A) in all 106 samples. The typical appearance of staining in tumors with immunohistochemically strong and weak GalNAc-T3 expression is shown in Figure 1B and Figure 1C, respectively. Tumors with strong GalNAc-T3 expression were found in 67 of 106 patients (63.2%).
We examined how closely GalNAc-T3 expression was related to various clinicopathologic characteristics (Tables 2, 3). From the standpoint of histologic type, tumors with strong GalNAc-T3 expression were seen in 38 of 53 patients with well-differentiated carcinoma (71.7%), in 23 of 44 patients with moderately differentiated carcinoma (52.3%), in 0 of 2 patients with poorly differentiated adenocarcinoma, and in 6 of 7 patients with mucinous carcinoma (85.7%). With a strong GalNac-T3 expression rate of 71.7% in well-differentiated tumors, compared with 52.3% in moderately differentiated tumors, the difference was statistically significant (chi-square test; P = 0.049). Thus, GalNAc-T3 expression in tumors showed a correlation with histologic tumor differentiation. Although the expression of GalNAc-T3 was not associated with age, gender, tumor size, tumor location, or TNM stage, its expression was related to the depth of invasion (Mann–Whitney U test; P = 0.031; Table 3). Although the difference did not attain statistical significance, hepatic metastasis was more frequent among patients in the weak GalNAc-T3 expression group (12.8%) compared with patients in the strong GalNAc-T3 expression group (7.6%). Conversely, no significant difference in the rate of lymph node metastasis was evident between patients in the strong GalNAc-T3 expression group (51.4%) and patients in the weak GalNAc-T3 expression group (50%).
|Variable||No. of patients (%)||P value|
|Strong GalNAc-T3 expression (n = 67 patients)||Weak GalNAc-T3 expression (n = 39 patients)|
|Age (yrs) (mean ± SD)||61.8 ± 10.7||58.5 ± 11.3||0.12|
|Male||42 (62.7)||21 (53.8)||0.371|
|Female||25 (37.3)||18 (46.2)||—|
|Colon||34 (50.7)||25 (64.1)||0.4|
|Rectum||33 (49.3)||14 (35.9)||—|
|Depth of invasiona||3 (4.5)||1 (2.6)||0.031|
|T1||2 (3.0)||1 (2.6)||—|
|T2||10 (14.9)||4 (10.3)||—|
|T3||34 (50.7)||13 (33.3)||—|
|T4||17 (25.4)||20 (51.2)||—|
|Lymph node metastasis|
|Absent||31 (50)||18 (48.6)||0.897|
|Present||31 (50)||19 (51.4)||—|
|Absent||61 (92.4)||34 (87.2)||0.376|
|Present||5 (7.6)||5 (12.8)||—|
|0||3 (4.5)||1 (2.6)||0.216|
|I||8 (11.9)||2 (5.1)||—|
|II||22 (32.8)||15 (38.5)||—|
|III||28 (41.8)||12 (30.8)||—|
|IV||6 (9.0)||9 (23.1)||—|
GalNAc-T3 Expression and Prognosis
Survival analysis was performed for the 106 patients in terms of several individual variables. Patients who had tumors with weak GalNAc-T3 expression showed a lower long-term survival rate compared with patients who had tumors with strong GalNAc-T3 expression (5-year survival rates, 43.6% vs. 67.2%, respectively; 10-year survival rates, 38.5% vs. 59.7%, respectively; P = 0.017; log-rank test) (Fig. 2A). GalNAc-T3 expression had a strong association with survival among the variables tested (Cox univariate hazard model; P = 0.019) that was second only to TNM stage (P = 0.002). The relative hazard ratio for overall survival in patients who had tumors with strong GalNAc-T3 expression was 0.535 relative to patients who had tumors with weak GalNAc-T3 expression (95% confidence interval, 0.317–0.903). In a Cox multivariate analysis for survival, GalNAc-T3 was a significant variable (P = 0.009) that was second only to TNM stage (P = 0.005). These results indicate that GalNAc-T3 expression is an independent, statistically significant prognostic indicator (Table 4).
|Variable||Univariate analysis||Multivariate analysis|
|HR||95% CI||P value||HR||95% CI||P value|
|Histologic type (differentiation)|
|0, I, II||1.0||—||—||1.0||—||—|
Within the group of patients with well-differentiated carcinoma, patients who had tumors with weak GalNAc-T3 expression showed a lower survival rate compared with patients who had tumors with strong GalNAc-T3 expression (Fig. 2B). This difference was statistically significant (5-year survival rate, 40% vs. 65.8%, respectively; 10-year survival rate, 40% vs. 63.2%, respectively; P = 0.035; log-rank test). Within the group of patients with moderately differentiated carcinoma, patients who had tumors with weak GalNAc-T3 expression had a lower survival rate compared with patients who had tumors with strong GalNAc-T3 expression (5-year survival rate, 52.4% vs. 69.6%, respectively; 10-year survival rate, 42.9% vs. 60.9%, respectively; P = 0.425 by log-rank test; Fig. 2C). In the group of patients with mucinous carcinoma, patients who had tumors with weak GalNAc-T3 expression also showed a significantly lower survival rate compared with patients who had tumors with strong GalNAc-T3 expression (5-year survival rate, 0% vs. 66.7%, respectively; 10-year survival rate, 0% vs. 33.3%, respectively; P = 0.014; log-rank test). In all subgroups, patients who had tumors with weak GalNAc-T3 expression showed a lower survival rate compared with patients who had tumors with strong GalNAc-T3 expression. No poorly differentiated tumors had strong GalNAc-T3 expression.
Western Blot Analysis
To determine whether immunologic quantitation of GalNAc-T3 protein also was associated with tumor differentiation, we examined seven frozen human colorectal carcinoma samples and normal tissue from the same specimens by Western blot analysis (Fig. 3). Except for one mucinous tumor, tumors contained less GalNAc-T3 than normal mucosa, although GalNAc-T3 expression was confirmed in all tumors (Fig. 3). The greatest tumor expression of GalNAc-T3 was observed in Sample 5: the mucinous tumor. High levels of GalNAc-T3 expression were observed in Samples 3, 6, and 7, which were well-differentiated tumors. Medium-to-low levels of GalNAc-T3 expression were observed in Samples 1, 2, and 4, which were moderately differentiated adenocarcinomas. GalNAc-T3 expression was related to tumor differentiation, which also was true in the immunohistochemical analysis.
In the current study, we evaluated how GalNAc-T3 expression may be associated with characteristics of 106 colorectal tumors using an antihuman polyclonal GalNAc-T3 antibody. We found that immunohistochemical assessment of GalNAc-T3 expression in surgically resected specimens provided valuable prognostic information for patients with colorectal carcinoma. The outcome of patients who had tumors with weak GalNAc-T3 expression was worse compared with the outcome of patients who had tumors with strong GalNAc-T3 expression. Furthermore, we identified clear associations of GalNAc-T3 expression with differentiation-based histologic subgroups. Also, within each subgroup, patients who had tumors with weak GalNAc-T3 expression had a lower survival rate compared with the survival rate of patients who had tumors with strong GalNAc-T3 expression.
To our knowledge, this is the first analysis of GalNAc-T3 as a prognostic factor in patients with malignant disease. Like any tumor marker, results vary, depending on the threshold chosen for determining whether a tumor is positive or negative for the marker. To address this question, we classified 106 tumors semiquantitatively with either 0–25%, 25–50%, 50–75% or 75–100% GalNAc-T3 expression. Then, we examined the three different cut-off values: 25%, 50%, and 75%. Using 25% as the cut-off value, GalNAc-T3 still predicted 5-year survival but no longer was an independent predictor according to the multivariate Cox survival analysis. This likely reflects the fact that GalNAc-T3 is correlated with TNM stage in patients who have tumors that express GalNAc-T3 in < 25% of cells. Using 75% as the cut-off value, GalNAc-T3 no longer predicted 5-year survival in patients with colorectal carcinoma (data not shown). Thus, GalNAc-T3 is a useful and independent prognostic marker in patients with colorectal carcinoma only if the GalNAc-T3 cut-off value is set at 50%.
We found that GalNAc-T3 expression was decreased in most tumors, in contrast to its strong expression in normal colorectal mucosa. These data were consistent with previous reports that tumor tissues contained less carbohydrate than normal mucosa,23, 24 with mucins isolated from colon carcinoma showing a reduction of carbohydrate chain number and length compared with mucin from normal mucosa.25 Vavasseur et al. reported that GalNAc-T3 activity in an intermediate premalignant cell line was twice as high as in an early premalignant cell line, whereas, in a carcinoma cell line, activity was decreased to 61% of that in the early premalignant cell line.26 These data indicate that glycosyltransferases are regulated in characteristic patterns along the adenoma-carcinoma sequence.
Mortality from colorectal carcinoma is related to the extent of metastasis and tumor invasion at the time of diagnosis. Certain cell-surface antigens have been associated with a poor prognosis in patients with colorectal carcinoma, such as the sialyl-Tn antigen.1 An O-glycan elongation inhibitor, arylglycoside benzyl-α-N-acetyl-galactosamine, inhibited liver metastasis by highly metastatic, mucin-productive colon carcinoma cells injected into nude mice.4 Bresalier and colleagues demonstrated that, compared with primary tumors, metastases in patients with colorectal carcinoma showed increased amounts of sialyl-Tn (NeuAcα6GalNAc), and sialyl-T (NeuAcαGalβ3GalNAc) associated reciprocally with a decrease in the mucin core antigens Tn (GalNAcαThr/Ser) and T (Galβ3GalNac).4–6 It is unclear how those findings relate to our current observation that there was a higher rate of liver metastasis in patients who had tumors with weak GalNAc-T3 expression (12.8%) compared with the liver metastasis rate in patients who had tumors with strong GalNAc-T3 expression (7.6%; Table 3). One possible explanation is that GalNAc-T3 activity is reduced relative to sialyl transferase, which may allow the sialylation of a greater proportion of oligosaccharide chains. Because GalNAc-T regulates O-glycosylation, GalNAc-T3 expression changes may induce changes in O-glycosylation and cell-surface carbohydrates in adenocarcinoma. Consequently, cellular functions in tumors, such as binding of cell adhesion molecules, invasion, and metastasis, may be changed.
We found that GalNAc-T3 expression varied according to the histologic tumor type. In adenocarcinoma cell lines, which are derived from epithelial cells, the expression of GalNAc-T3 has been associated with cellular differentiation.7 Furthermore, we found that the expression of this gene was induced in two colorectal carcinoma cell lines by a differentiation-promoting agent, sodium butyrate, using immunoblotting and transient transfection assay (data not shown), which also supports the hypothesis that GalNAc-T3 gene expression may serve as a marker for differentiation in adenocarcinoma. The current immunohistochemical analysis of colorectal carcinoma specimens identified tumors with strong and weak GalNAc-T3 expression, even among well-differentiated adenocarcinomas. Patients who had tumors with weak GalNAc-T3 expression had poorer outcomes, even in this favorable histologic subgroup. Similar correlations were observed in moderately differentiated and mucinous colorectal carcinomas. Multivariate Cox survival analyses identified GalNAc-T3 expression as an independent prognostic factor that was second only to TNM stage. These results suggest that GalNAc-T3 is a highly promising diagnostic marker for biologic tumor differentiation, exceeding morphologic differentiation in its predictive ability. Specifically, we found that GalNAc-T3 expression showed an inverse correlation with the depth of tumor invasion. The findings presented here suggest that GalNAc-T3 expression may be useful in evaluating potential for aggressiveness in colorectal adenocarcinomas. Disruption of differentiated tissue architecture associated with malignant transformation and ultimate invasiveness and metastatic behavior may be related to the reduced O-glycosylation of cell-surface molecules.26
Although many clinical, biologic, and histologic variables have been investigated as prognostic indicators, GalNAc-T3 may be one of the useful predictors. Molecular analyses as well as further correlative studies of GalNAc-T3 expression in patients with other types of adenocarcinoma need to be pursued.
The authors thank Dr. M. H. Nathanson and Dr. Y. Morimitsu for their valuable advice. They also appreciate Dr. K. Onitsuka and Ms. Y. Ueda for their excellent technical assistance.
- 15Cloning and characterization of a close homologue of human UDP-N-acetyl-alpha-D-galactosamine: polypeptide N-acetylgalactosaminyltransferase-T3, designated GalNAc-T6. Evidence for genetic but not functional redundancy. J Biol Chem. 1999;274: 25362–25370., , , et al.
- 22Regression models and life tables. J R Stat Soc B. 1972; 34: 187–120..