- Email: [email protected]
Increased expression of cyclin D1 is an early event in multistage colorectal carcinogenesis
GASTROENTEROLOGY 1996;110:669–674
Increased Expression of Cyclin D1 Is an Early Event in Multistage Colorectal Carcinogenesis NADIR ARBER,*,‡ HANINA HIBSHOOSH,§ STEVEN F. MOSS,* THOMAS SUTTER,‡ YEN ZHANG,§ MELISSA BEGG,㛳 SHAOBAI WANG,§ I. BERNARD WEINSTEIN,‡ and PETER R. HOLT* *Division of Gastroenterology, Department of Medicine, St. Luke’s/Roosevelt Hospital Center, New York; ‡Columbia–Presbyterian Cancer Center, New York; and §Department of Pathology and 㛳Division of Biostatistics, Columbia University, New York, New York
Background & Aims: Cyclin D1 gene amplification and/ or overexpression occurs in several human cancers. The level of expression of cyclin D1 protein during the multistage process of human colon carcinogenesis was determined. Methods: Cyclin D1 protein abundance was determined by immunostaining samples of normal colonic mucosa (n Å 23), transitional normal mucosa adjacent to adenomas or adenocarcinomas (n Å 41), hyperplastic polyps (n Å 8), adenomatous polyps (n Å 35), and adenocarcinomas (n Å 27), using a polyclonal anti-human cyclin D1 antibody. Results: Cyclin D1 nuclear staining occurred in 30% of adenocarcinomas and 34% of adenomatous polyps but not in hyperplastic polyps or normal or transitional mucosa. Nuclear staining did not correlate with sex, age, size, or dysplasia of the adenomatous polyps or with differentiation and Dukes’ staging of the adenocarcinomas. Left-sided colon neoplasms showed nuclear staining more frequently than those right-sided lesions. Diffuse or supranuclear cytoplasmic staining occurred in about one third of hyperplastic polyps, adenomas, and adenocarcinomas and in transitional mucosa adjacent to adenocarcinoma. Conclusions: Increased nuclear expression of cyclin D1 occurs in around one third of colonic tumors as an early event during multistage process of colon carcinogenesis. Increased expression of cyclin D1 may perturb cell-cycle control in benign adenomas and thereby enhance tumor progression.
P
rogression of cells through the cell cycle is governed by the sequential formation and degradation of a series of cyclins that complex with and activate several cyclin-dependent kinases. These cyclin and cyclin-dependent kinase CDK complexes play a critical role in cell proliferation and differentiation. There are at least 11 distinct cyclin genes in the human genome that fall into three categories: G1-phase cyclins (C, D1–3, E, G, and H), S-phase cyclins (A and F), and G2/M-phase cyclins (A and B1–2) (for review, see Pines and Hunter1 and Sherr2). The human cyclin D1 gene was first cloned through / 5e0a$$0001
02-09-96 11:25:28
gasa
its ability to complement a Saccharomyces cerevisiae strain that had mutations in all three of the known yeast G1 cyclins.3,4 Immunoneutralization with cyclin D1 antibodies resulted in cell-cycle arrest in the G1 phase,5 implying that cyclin D1 plays an important role in normal cell proliferation. This also raises the possibility that mutations and/or altered expression of cyclin D1 may be involved in states of altered cell proliferation, such as neoplasia. Furthermore, cyclin D1 overexpression shortens the G1 phase of the cell cycle of cultured cells, decreases cell size, and makes cells less dependent on exogenous growth factors.1,6,7 Evidence for the oncogenic properties of deregulated cyclin D1 expression is provided by the analysis of several model systems in which increased expression of cyclin D1 enhances malignant cell transformation.1,6,8 – 10 Rearrangements and increased expression of the cyclin D1 gene are observed in parathyroid adenomas, a subset of B-cell lymphomas,1 and amplification and increased expression of the cyclin D1 gene are observed in esophageal, head and neck, hepatic, and breast cancers.1,8,11 – 16 In addition, immunohistochemical detection of cyclin D1 in human breast cancers has identified a subset of carcinomas in which the cyclin D1 gene is overexpressed in the absence of gene amplification.17 This is associated with increased cyclin D1 messenger RNA levels in more than one third of these tumors,18 a greater frequency than that predicted from DNA amplification studies. Although cyclin D1 gene amplification is an important event in many tumor types, it is rarely found in colon cancer cell lines.19 Because mechanisms other than gene amplification may lead to deregulation and accumulation of this proto-oncogenic cyclin in cancer, this prompted us to examine the level of expression of cyclin D1 protein at different stages along the multistage process of colon carcinogenesis from normal colon mucosa to adenoma and to carcinoma. 䉷 1996 by the American Gastroenterological Association 0016-5085/96/$3.00
WBS-Gastro
670 ARBER ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
nology), representing approximately 100-fold excess of peptide over the antibody.
Materials and Methods Tissue Samples Formalin-fixed, paraffin-embedded human colonic specimens were obtained from the Pathology Department of St. Luke’s/Roosevelt Hospital Center (New York, NY). The tissues had been removed endoscopically or obtained at surgery between 1991 and 1994 and had been processed by routine clinical histopathologic methods. The samples included normal colon mucosa (n Å 23), which was divided into the following two distinct categories: normal mucosa obtained from the resection margin of colorectal adenocarcinomas (n Å 13) and biopsy specimens obtained during normal colonoscopy from patients without tumors (n Å 10); transitional mucosa (n Å 41), i.e., normal mucosa adjacent to an adenoma or adenocarcinoma on the same histological slide; hyperplastic polyps (n Å 8); adenomatous polyps (n Å 35); and adenocarcinomas (n Å 27).
Immunohistochemistry All immunohistochemical analyses were performed with an avidin-biotin complex immunoperoxidase technique. Five-micrometer tissue sections were mounted on poly-L-lysine–coated slides. After deparaffinization in Americlear (Baxter, McGaw Park, IL) and absolute ethanol, sections were hydrated through a series of graded alcohols, distilled water, and phosphate-buffered saline at pH 7.4. Slides were then immersed in 10 mmol/L citrate buffer (pH Å 6) and microwaved (to enhance antigen exposure) for a total of 10 minutes at 750 W. After blocking with goat serum for 20 minutes, the primary antibody polyclonal immunoglobulin G–rabbit anti-human cyclin D1 (Upstate Biotechnology, Lake Placid, NY) was applied and incubated overnight at 4⬚C in a high-humidity chamber. Although all concentrations of the primary antibody provided good nuclear staining, the ideal concentration with minimal background was 5 mg/mL. This antibody was generated using a specific amino acid sequence contained within the C-terminal domain of the human cyclin D1 protein20 that reacts with cyclin D1 and D2 but not with D3.21 As a negative control, a duplicate section of each tissue sample was immunostained in the absence of the primary antibody. A breast carcinoma with known cyclin D1 overexpression served as a positive control. Subsequent steps used the Vectastain Rabbit Elite ABC Kit (Vector Laboratories, Burlingame, CA) according to the manufacturer’s instructions. Color development was accomplished with a 0.375 mg/dL solution of a 3,3ⴕ-diaminobenzidine tetrahydrochloride (Sigma Chemical Co., St. Louis, MO) containing 0.003% hydrogen peroxide. Slides were counterstained with hematoxylin and dehydrated, and coverslips were applied using Acrytol mounting medium (Surgipath Medical Industries, Richmond, IL). The specificity of the antibodies has been shown (data not shown) by the inhibition of immunohistochemical staining in positive controls by preincubating the antibody with 1 mg of cyclin D1–immunizing peptide (molecular weight, 1413 daltons; amino acids, 285– 295) for 1 hour at 4⬚C (catalog no. 12–167; Upstate Biotech-
/ 5e0a$$0001
02-09-96 11:25:28
gasa
Interpretation of Immunohistochemical Staining All slides were interpreted by one of the investigators (H.H.), who is an experienced pathologist. Nuclear staining was considered positive if the chromogen was detected in at least 5% of the nuclei within a microscopic field. Staining intensity included the following four scales: no staining (scale level, 0; Figure 1B); weak staining, comparable with adjacent nonneoplastic epithelium (scale 1; Figure 1C); moderately positive (scale 2; Figure 2B); and strongly positive (scale 3; Figure 2A and C). Scales 0 and 1 were regarded as negative, and scales 2 and 3 were regarded as positive. Cytoplasmic staining as well as its intracellular distribution (diffuse or localized) were recorded separately on a similar scale as the nuclear stain. Nonneoplastic epithelial cells and stromal and inflammatory cells were also evaluated for cyclin D1 expression. Positive and negative controls were included within each batch of slides. To confirm reproducibility, 25% of the slides were chosen randomly and scored twice in the same batch, and all batches were coded and scored as blind at least twice. All duplicate slides were interpreted similarly.
Statistical Analysis The proportions of samples overexpressing the cyclin D1 gene from different histological categories were computed and then compared across categories for selected factors, including sex, age, location of the lesion, histology, differentiation, Dukes’ staging, and presence of dysplasia. Some subjects contributed more than one specimen to the analysis; thus, statistical methods that adjust for intrasubject dependence (i.e., correlation between multiple samples taken from a single individual) were used. When comparing proportions positive for staining across histological categories, Brier’s correction for the ordinary x2 test was used.22 All other comparisons were made via logistic regression for binary outcomes,23 with corrections for intrasubject correlation performed by the method of Liang and Zeger,24 assuming constant correlation structure.
Results Immunostaining for cyclin D1 was performed on 134 paraffin-embedded tissue specimens taken from 69 subjects (age range, 18–93 years; mean, 63.8 years). Clinical characteristics of patients and tissue samples are summarized in Table 1. Positive nuclear staining for cyclin D1 was observed in 8 of the 27 adenocarcinoma (30%) and 12 of the 35 adenomatous polyps (34%) (Figure 2). No nuclear staining was detected in eight hyperplastic polyps, 23 normal-appearing mucosa, and 41 transitional mucosa samples. Positive nuclear staining was not related to the sex or age of the patient, to the size and degree of dysplasia in adenomatous polyps, or to the differentiation and Dukes’ stage of adenocarcinomas. WBS-Gastro
March 1996
CYCLIN D1 IN COLON NEOPLASIA 671
Thus, it was observed with about equal frequency in Dukes’ A – D lesions. The number of tumors with positive staining was significantly greater in lesions from the left-sided colon, i.e., rectum, sigmoid colon, and descending colon (47%), than from the right-sided colon, i.e., transverse colon, ascending colon, and cecum (12%) (P õ 0.05). This was true even after controlling for diagnosis (adenomatous polyps vs. adeno-
Figure 1. Cyclin D1–positive supranuclear cytoplasmic staining (A ) in normal mucosa adjacent to an adenocarcinoma and (B ) in an adenomatous polyp and (C ) diffuse cytoplasmic staining in a hyperplastic polyp.
/ 5e0a$$0001
02-09-96 11:25:28
gasa
carcinomas). Eleven of 22 (50%) of the polyps and 6 of 14 (43%) of the adenocarcinomas in the left-sided colon were positive, compared with only 1 of 13 (8%) of the polyps and 2 of 13 (15%) of the adenocarcinomas in the right-sided colon (Table 2). The percentage of cyclin D1 – positive staining nuclei were similar in adenomatous polyps and adenocarcinomas. In about 60% of cyclin D1 – positive tumors, 25% – 50% of the nuclei showed immunohistological staining, and in 30% of the cases, ú50% of nuclei were positive. Adenomatous polyps showed positive nuclei almost exclusively near
Figure 2. Cyclin D1–positive nuclear staining in an (A ) adenomatous polyp and (B and C ) adenocarcinoma.
WBS-Gastro
672 ARBER ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
Table 1. Clinical Characteristics of Patients Showing Cyclin D1 Immunoreactivity Normal mucosa No. of samples Sex (M/F) (%) Mean age (yr) Left-sided colon samples (%) Right-sided colon samples (%) Positive nuclear stain (%) Positive cytoplasmic stain (in the lesion)b (%) Positive cytoplasmic stain (in adjacent normal mucosa)c (%)
Transitional mucosa
Hyperplastic polyps
23 59/41 60.4 55 45 0 22
41 40/60 66.1 59 41 0 59
8 63/37 66.4 88 12 0 38
NA
NA
0
a
Adenomatous polyps
Adenocarcinoma
35 49/51 67.2 63 37 34 34
27 65/35 64.5 52 48 30 26
33 (7 of 21)
85 (17 of 20)
NA, Not applicable. a Includes both resection margin and normal colonoscopy samples. b Refers to percent of samples in each histological category that showed positive cytoplasmic staining in the lesion itself. c Refers to percent of positive cytoplasmic staining of normal mucosa adjacent to each histological category.
the luminal surface, whereas adenocarcinomas showed positive cells diffusely. Positive cyclin D1 cytoplasmic staining was detected in many of these samples. It was identified in both neoplastic and nonneoplastic tissues. The incidence of cytoplasmic staining for cyclin D1 in the various histological categories is summarized in Table 3. Positive cytoplasmic staining was noted in about 30% of adenomatous polyps and adenocarcinomas, similar to the frequency of nuclear staining. The frequency of positive cytoplasmic cyclin D1 staining was not related to the sex or age of the patient or to the site of the tissues, the size or degree of dysplasia in polyps, or Dukes’ stage and differentiation in the adenocarcinomas (Table 2). Only 9% of adenomatous polyps and 4% of adenocarcinomas had both positive cyclin D1 nuclear and cytoplasmic staining, whereas 26% of adenomatous polyps and 22% of adenocarcinomas showed positive cytoTable 2. Cyclin D1 Nuclear and Cytoplasmic Immunoreactivity in Adenomas and Adenocarcinomas
Sex (M/F) Left-sided colon Adenomatous polyps Adenocarcinomas Both lesions Right-sided colon Adenomatous polyps Adenocarcinomas Both lesions Dysplasia in adenomas Absent Present
No. of samples
Samples positive for nuclear staining
Samples positive for cytoplasmic staining
27/34
9 (33)/11 (32)
9 (33)/10 (29)
22 14 36
11 (50) 6 (43) 17 (47)
6 (27) 5 (36) 11 (31)
13 13 26
1 (8) 2 (15) 3 (12)
6 (46) 2 (15) 8 (31)
12 23
4 (33) 8 (35)
6 (50) 9 (39)
02-09-96 11:25:28
Table 3. Nuclear and Cytoplasmic Cyclin D1 Immunoreactivity in Different Colonic Lesions
Normal mucosa Transitional mucosa Hyperplastic polyp Adenomatous polyps Adenocarcinomas
NOTE. Values in parentheses indicate the percentage of positive samples of tumors.
/ 5e0a$$0001
plasmic staining in the absence of nuclear staining. Cytoplasmic staining also was observed in 38% of the hyperplastic polyps and in 22% of the normal mucosa samples (Table 3). It is of interest that, in the latter group, all of the positive cases were apparently normal mucosa samples taken from the resection margins of tumors. Thus, 5 of 13 (38%) of the latter samples were positive, whereas none of the 10 samples obtained from patients without tumors was positive. Positive cytoplasmic staining was observed in 59% of transitional mucosa samples. It was associated with adenocarcinomas in 85% of the cases compared with only 33% with the adenomatous polyps (P õ 0.05). We noted two different patterns of cytoplasmic staining for cyclin D1. Some samples showed a diffuse pattern of staining, and other samples showed punctate discrete intense staining with a supranuclear location (Figure 1). Supranuclear localization was observed in 6 of the 7 (86%) adenocarcinomas and 10 of the 12 (83%) adenomatous polyps showing cytoplasmic staining. In most of
gasa
No. of samples
Samples positive for nuclear staining (%)
23a 41 8 35 27
0 0 0 12 (34) 8 (30)
Samples positive for cytoplasmic staining (%) 5 24 3 12 7
(22)b (59) (38) (34) (26)
NOTE. Values in parentheses designate the percentage of samples that were positive. a Includes both resection margin and normal colonoscopic biopsy samples. b All of the positive samples came from resection margins of tumors.
WBS-Gastro
March 1996
CYCLIN D1 IN COLON NEOPLASIA 673
these samples, 80% of the adenomatous polyps and 83% of the adenocarcinomas, this cytoplasmic staining occurred in the absence of positive nuclear staining. Thus, cytoplasmic staining for cyclin D1 is observed quite frequently in both normal colonic mucosa and colon tumors. Our data suggest that it occurs with increased frequency in transitional mucosa, polyps, and adenocarcinomas, but the differences between these categories did not reach statistical significance.
Discussion The major finding in this study is that about 30% of adenomatous polyps and colorectal adenocarcinomas showed increased nuclear expression of the cyclin D1 protein, whereas hyperplastic polyps and normal colonic mucosa showed no nuclear expression of this protein (Table 1). These findings are intriguing because there have been no reports of translocation or amplification involving the 11q13 chromosome region (the site of the cyclin D1 gene) in human colon cancer25 and no amplification of the cyclin D1 gene has been found in any of 47 colorectal cancer cell lines.19 Similarly, we found no evidence for cyclin D1 gene amplification in the 18 primary human colorectal cancers and six human colorectal cancer cell lines that were examined, despite high levels of cyclin D1 messenger RNA and protein in around half of these cases (Sutter et al., unpublished data, February 1995). Therefore, it is unlikely that gene amplification accounts for the high frequency of increased expression of the cyclin D1 protein that we have observed in this histochemical study. Cyclin D1 protein accumulation was detected in about one third of both dysplastic and nondysplastic adenomatous polyps and in both small and large lesions and was also observed with a similar frequency in colon adenocarcinomas, regardless of their differentiation or Dukes’ stage. Therefore, increased expression of this gene seems to be an early event in colorectal tumorigenesis. Recently, K-ras and APC mutations were reported in aberrant crypt foci, which may represent the earliest lesion in colorectal tumorigenesis.26 It will be interesting to examine whether or not cyclin D1 is overexpressed in these foci. Increased cyclin D1 expression was significantly more prevalent in left- than right-sided neoplastic lesions, providing further evidence of differences between left- and right-sided colonic tumors.27 The distribution of cyclin D1–positive nuclei in adenomas was restricted primarily to the superficial aspects of the lesion. This contrasts with the diffuse distribution of positive nuclei in adenocarcinomas. It also is of interest to note that nonneoplastic crypts showed no detectable nuclear staining. This suggests that physiological levels / 5e0a$$0001
02-09-96 11:25:28
gasa
of cyclin D1 in colonic crypts are too low for detection. This observation supports the importance of abnormal levels of expression of this protein in colonic neoplasia. We should emphasize that the polyclonal antibodies used in the present study recognize both cyclin D1 and cyclin D2 proteins. However, Bartkova et al.,28 using monoclonal antibodies specific to cyclin D1, showed nuclear staining in colonic adenocarcinoma similar to that in the present study. In addition, we have found that 9 of 18 primary colorectal adenocarcinomas and 3 of 6 human colon cancer cell lines showed a twofold to threefold increase in the expression of cyclin D1 but not cyclin D2, messenger RNA, and protein (Sutter et al., unpublished data, February 1995). Therefore, it is likely that the cyclin D1 overexpression detected in the present study is caused by increased cyclin D1 protein. The demonstration of cyclin D1 expression in the cytoplasm of colonic epithelial cells is novel. We found that this occurred in a relatively high proportion of our specimens, including hyperplastic polyps, transitional mucosa, and normal mucosa obtained from the resection margins of tumors (Tables 2 and 3). Cytoplasmic staining is not caused by leakage of this protein from the nucleus because it was observed frequently in the absence of nuclear staining. Moreover, cytoplasmic staining frequently had a supranuclear distribution in the region of the Golgi apparatus. This pattern was observed more often in the transitional mucosa of adenocarcinomas (85%) than in the transitional mucosa of adenomatous polyps (33%) (P õ 0.05). Cytoplasmic immunostaining for cyclin D1 also has been described in some lymphomas,29 and the presence of cyclin D1 outside the nucleus suggests an additional presently undefined physiological role. Its occurrence in the transitional colonic mucosa may reflect reactivity of the normal mucosa to the adjoining tumor or may be caused by the inherent biology of colonic neoplasms. Further studies are necessary to understand this phenomenon more fully.
References 1. Pines J, Hunter T. Cyclins and cancer II: cyclin D1 and CDK inhibitors come of age. Cell 1994;79:573–582. 2. Sherr CJ. Mammalian G1 cyclins. Cell 1993;73:1059–1065. 3. Lew DJ, Dulic V, Reed SI. Isolation of three novel human cyclins by rescue of G1 cyclin function in yeast. Cell 1991;66:1197– 1206. 4. Xiong Y, Connolly T, Futcher B, Beach D. Human D-type cyclin. Cell 1991;65:691–699. 5. Baldin V, Lukas J, Marcote MJ, Pagano M, Draetta G. Cyclin D1 is a nuclear protein required for cell cycle progression in G1. Genes Dev 1993;7:812–821. 6. Jiang W, Kahn SM, Zhou P, Zhang YJ, Cacace AM, Infante AS, Doi S, Santella RM, Weinstein IB. Overexpression of cyclin D1 in rat fibroblasts causes abnormalities in growth control, cell
WBS-Gastro
674 ARBER ET AL.
7.
8.
9.
10.
11.
12.
13. 14. 15.
16. 17.
18.
19.
GASTROENTEROLOGY Vol. 110, No. 3
cycle progression and gene expression. Oncogene 1993; 8:3447–3457. Quelle DE, Ashmun RA, Shurtleff SA, Kato J-Y, Bar-Sagi D, Roussel HF, Sherr CJ. Overexpression of mouse D-type cyclins accelerates G1 phase in rodent fibroblasts. Genes Dev 1993;7:1559– 1571. Wang TC, Cardiff RD, Zukerberg L, Lees E, Arnold A, Schmidt EV. Mammary hyperplasia and carcinoma in MMTV-cyclin D1 transgenic mice. Nature 1994;369:669–671. Hinds PW, Dowdy SF, Ng Eaton E, Arnold A, Weinberg RA. Function of a human cyclin gene as an oncogene. Proc Natl Acad Sci USA 1994;91:709–713. Lovec H, Sewing A, Lucibello FC, Muller R, Moroy T. Oncogenic activity of cyclin D1 revealed through cooperation with Ha-ras: link between cell cycle control and malignant transformation. Oncogene 1994;9:323–326. Jiang W, Zhang Y-J, Kahn SM, Hollstein MC, Santella RM, Lu SH, Harris CC, Montesano R, Weinstein IB. Altered expression of the cyclin D1 and retinoblastoma genes in human esophageal cancer. Proc Natl Acad Sci USA 1993;90:9026–9030. Zhang YJ, Jiang W, Chen CJ, Lee CS, Kahn SM, Santella RM, Weinstein IB. Amplification and overexpression of cyclin D1 in human hepatocellular carcinoma. Biochem Biophys Res Commun 1993;196:1010–1016. Hunter T, Pines J. Cyclins and cancer. Cell 1991;66:1071– 1074. Marx J. How cells cycle toward cancer. Science 94:263:319– 321. Keyomarsi K, Pardee AB. Redundant cyclin overexpression and gene amplification in breast cancer cells. Proc Natl Acad Sci USA 1993;90:1112–1116. Lammie GA, Peters G. Chromosome 11q13 abnormalities in human cancer. Cancer Cells 1991;3:413–420. Gillett C, Fantl V, Smith R, Fisher C, Bartek J, Dickson C, Barnes D, Peters G. Amplification and overexpression of cyclin D1 in breast cancer detected by immunohistochemical staining. Cancer Res 1994;54:1812–1817. Buckley MF, Sweeney KJE, Hamilton JA, Sini RL, Manning DL, Niccholson RI, DeFazio A, Watts CKW, Musgrove EA, Sutherland RL. Expression and amplification of cyclin genes in human breast cancer. Oncogene 1993;8:2127–2133. Leach SF, Elledge SJ, Sherr CJ, Wilson JKV, Markowitz S, Kinzler
/ 5e0a$$0001
02-09-96 11:25:28
gasa
20. 21.
22. 23.
24. 25. 26.
27.
28.
29.
KW, Vogelstein B. Amplification of cyclin genes in colorectal carcinomas. Cancer Res 1993;53:1986–1989. Pines J. Cyclins: wheels within wheels. Cell Growth Differ 1991;2:305–310. Hall FL, Williams RT, Wu L, Wu F, Carbonaro-Hall DA, Harper JW, Warburton D. Two potentially oncogenic cyclins, cyclin A and cyclin D1, share common properties of subunit configuration, tyrosine phosphorylation and physical association with the Rb protein. Oncogene 1993;8:1377–1384. Brier SS. Analysis of contingency tables under cluster sampling. Biometrika 1980;67:591–596. Breslow NE, Day NE. Statistical methods in cancer research, volume 1: the analysis of case control studies. Lyon, France: International Agency for Research on Cancer, 1980. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73:13–22. Muleris M, Salmon RJ, Dutrillaux B. Cytogenetics of colorectal adenocarcinomas. Cancer Genet Cytogenet 1990;46:143–156. Pretlow TS. Aberrant crypt foci and K-ras mutations: earliest recognized players or innocent bystanders in colon carcinogenesis? Gastroenterology 1995;108:600–603. Liu T, Distler P, Mokoulu AD, Reddy BS, Zang E, Holt PR. Regional differences in carcinogene induced colon neoplasia—effects of chemoprevention. Gastroenterology (in press). Bartkova J, Lukas J, Strauss M, Bartek J. The prad-1 cyclin D1 oncogene product accumulates aberrantly in a subset of colorectal carcinomas. Int J Cancer 1994;58:568–573. Banno S, Yoshikawa K, Nakamura S, Yamamoto K, Seito T, Nitta M, Takahashi T, Ueda R, Seto M. Monoclonal antibody against PRAD1/cyclin D1 stains nuclei of tumor cells with translocation or amplification at BCL-1 Locus. Jpn J Cancer Res 1994;85:918– 926.
Received June 22, 1995. Accepted October 6, 1995. Address requests for reprints to: Peter R. Holt, M.D., Division of Gastroenterology, Department of Medicine, St. Luke’s Hospital Center, 1111 Amsterdam Avenue, New York, New York 10025. Fax: (212) 523-3683. Supported by grants from the Dorot Foundation (to N.A.), the National Dairy Council (to P.R.H.), the National Foundation for Cancer Research (to I.B.W.), and the Markey Charitable Trust (to I.B.W.). The authors thank the Department of Pathology at St. Luke’s/ Roosevelt Hospital for valuable assistance in these studies.
WBS-Gastro
Increased Expression of Cyclin D1 Is an Early Event in Multistage Colorectal Carcinogenesis NADIR ARBER,*,‡ HANINA HIBSHOOSH,§ STEVEN F. MOSS,* THOMAS SUTTER,‡ YEN ZHANG,§ MELISSA BEGG,㛳 SHAOBAI WANG,§ I. BERNARD WEINSTEIN,‡ and PETER R. HOLT* *Division of Gastroenterology, Department of Medicine, St. Luke’s/Roosevelt Hospital Center, New York; ‡Columbia–Presbyterian Cancer Center, New York; and §Department of Pathology and 㛳Division of Biostatistics, Columbia University, New York, New York
Background & Aims: Cyclin D1 gene amplification and/ or overexpression occurs in several human cancers. The level of expression of cyclin D1 protein during the multistage process of human colon carcinogenesis was determined. Methods: Cyclin D1 protein abundance was determined by immunostaining samples of normal colonic mucosa (n Å 23), transitional normal mucosa adjacent to adenomas or adenocarcinomas (n Å 41), hyperplastic polyps (n Å 8), adenomatous polyps (n Å 35), and adenocarcinomas (n Å 27), using a polyclonal anti-human cyclin D1 antibody. Results: Cyclin D1 nuclear staining occurred in 30% of adenocarcinomas and 34% of adenomatous polyps but not in hyperplastic polyps or normal or transitional mucosa. Nuclear staining did not correlate with sex, age, size, or dysplasia of the adenomatous polyps or with differentiation and Dukes’ staging of the adenocarcinomas. Left-sided colon neoplasms showed nuclear staining more frequently than those right-sided lesions. Diffuse or supranuclear cytoplasmic staining occurred in about one third of hyperplastic polyps, adenomas, and adenocarcinomas and in transitional mucosa adjacent to adenocarcinoma. Conclusions: Increased nuclear expression of cyclin D1 occurs in around one third of colonic tumors as an early event during multistage process of colon carcinogenesis. Increased expression of cyclin D1 may perturb cell-cycle control in benign adenomas and thereby enhance tumor progression.
P
rogression of cells through the cell cycle is governed by the sequential formation and degradation of a series of cyclins that complex with and activate several cyclin-dependent kinases. These cyclin and cyclin-dependent kinase CDK complexes play a critical role in cell proliferation and differentiation. There are at least 11 distinct cyclin genes in the human genome that fall into three categories: G1-phase cyclins (C, D1–3, E, G, and H), S-phase cyclins (A and F), and G2/M-phase cyclins (A and B1–2) (for review, see Pines and Hunter1 and Sherr2). The human cyclin D1 gene was first cloned through / 5e0a$$0001
02-09-96 11:25:28
gasa
its ability to complement a Saccharomyces cerevisiae strain that had mutations in all three of the known yeast G1 cyclins.3,4 Immunoneutralization with cyclin D1 antibodies resulted in cell-cycle arrest in the G1 phase,5 implying that cyclin D1 plays an important role in normal cell proliferation. This also raises the possibility that mutations and/or altered expression of cyclin D1 may be involved in states of altered cell proliferation, such as neoplasia. Furthermore, cyclin D1 overexpression shortens the G1 phase of the cell cycle of cultured cells, decreases cell size, and makes cells less dependent on exogenous growth factors.1,6,7 Evidence for the oncogenic properties of deregulated cyclin D1 expression is provided by the analysis of several model systems in which increased expression of cyclin D1 enhances malignant cell transformation.1,6,8 – 10 Rearrangements and increased expression of the cyclin D1 gene are observed in parathyroid adenomas, a subset of B-cell lymphomas,1 and amplification and increased expression of the cyclin D1 gene are observed in esophageal, head and neck, hepatic, and breast cancers.1,8,11 – 16 In addition, immunohistochemical detection of cyclin D1 in human breast cancers has identified a subset of carcinomas in which the cyclin D1 gene is overexpressed in the absence of gene amplification.17 This is associated with increased cyclin D1 messenger RNA levels in more than one third of these tumors,18 a greater frequency than that predicted from DNA amplification studies. Although cyclin D1 gene amplification is an important event in many tumor types, it is rarely found in colon cancer cell lines.19 Because mechanisms other than gene amplification may lead to deregulation and accumulation of this proto-oncogenic cyclin in cancer, this prompted us to examine the level of expression of cyclin D1 protein at different stages along the multistage process of colon carcinogenesis from normal colon mucosa to adenoma and to carcinoma. 䉷 1996 by the American Gastroenterological Association 0016-5085/96/$3.00
WBS-Gastro
670 ARBER ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
nology), representing approximately 100-fold excess of peptide over the antibody.
Materials and Methods Tissue Samples Formalin-fixed, paraffin-embedded human colonic specimens were obtained from the Pathology Department of St. Luke’s/Roosevelt Hospital Center (New York, NY). The tissues had been removed endoscopically or obtained at surgery between 1991 and 1994 and had been processed by routine clinical histopathologic methods. The samples included normal colon mucosa (n Å 23), which was divided into the following two distinct categories: normal mucosa obtained from the resection margin of colorectal adenocarcinomas (n Å 13) and biopsy specimens obtained during normal colonoscopy from patients without tumors (n Å 10); transitional mucosa (n Å 41), i.e., normal mucosa adjacent to an adenoma or adenocarcinoma on the same histological slide; hyperplastic polyps (n Å 8); adenomatous polyps (n Å 35); and adenocarcinomas (n Å 27).
Immunohistochemistry All immunohistochemical analyses were performed with an avidin-biotin complex immunoperoxidase technique. Five-micrometer tissue sections were mounted on poly-L-lysine–coated slides. After deparaffinization in Americlear (Baxter, McGaw Park, IL) and absolute ethanol, sections were hydrated through a series of graded alcohols, distilled water, and phosphate-buffered saline at pH 7.4. Slides were then immersed in 10 mmol/L citrate buffer (pH Å 6) and microwaved (to enhance antigen exposure) for a total of 10 minutes at 750 W. After blocking with goat serum for 20 minutes, the primary antibody polyclonal immunoglobulin G–rabbit anti-human cyclin D1 (Upstate Biotechnology, Lake Placid, NY) was applied and incubated overnight at 4⬚C in a high-humidity chamber. Although all concentrations of the primary antibody provided good nuclear staining, the ideal concentration with minimal background was 5 mg/mL. This antibody was generated using a specific amino acid sequence contained within the C-terminal domain of the human cyclin D1 protein20 that reacts with cyclin D1 and D2 but not with D3.21 As a negative control, a duplicate section of each tissue sample was immunostained in the absence of the primary antibody. A breast carcinoma with known cyclin D1 overexpression served as a positive control. Subsequent steps used the Vectastain Rabbit Elite ABC Kit (Vector Laboratories, Burlingame, CA) according to the manufacturer’s instructions. Color development was accomplished with a 0.375 mg/dL solution of a 3,3ⴕ-diaminobenzidine tetrahydrochloride (Sigma Chemical Co., St. Louis, MO) containing 0.003% hydrogen peroxide. Slides were counterstained with hematoxylin and dehydrated, and coverslips were applied using Acrytol mounting medium (Surgipath Medical Industries, Richmond, IL). The specificity of the antibodies has been shown (data not shown) by the inhibition of immunohistochemical staining in positive controls by preincubating the antibody with 1 mg of cyclin D1–immunizing peptide (molecular weight, 1413 daltons; amino acids, 285– 295) for 1 hour at 4⬚C (catalog no. 12–167; Upstate Biotech-
/ 5e0a$$0001
02-09-96 11:25:28
gasa
Interpretation of Immunohistochemical Staining All slides were interpreted by one of the investigators (H.H.), who is an experienced pathologist. Nuclear staining was considered positive if the chromogen was detected in at least 5% of the nuclei within a microscopic field. Staining intensity included the following four scales: no staining (scale level, 0; Figure 1B); weak staining, comparable with adjacent nonneoplastic epithelium (scale 1; Figure 1C); moderately positive (scale 2; Figure 2B); and strongly positive (scale 3; Figure 2A and C). Scales 0 and 1 were regarded as negative, and scales 2 and 3 were regarded as positive. Cytoplasmic staining as well as its intracellular distribution (diffuse or localized) were recorded separately on a similar scale as the nuclear stain. Nonneoplastic epithelial cells and stromal and inflammatory cells were also evaluated for cyclin D1 expression. Positive and negative controls were included within each batch of slides. To confirm reproducibility, 25% of the slides were chosen randomly and scored twice in the same batch, and all batches were coded and scored as blind at least twice. All duplicate slides were interpreted similarly.
Statistical Analysis The proportions of samples overexpressing the cyclin D1 gene from different histological categories were computed and then compared across categories for selected factors, including sex, age, location of the lesion, histology, differentiation, Dukes’ staging, and presence of dysplasia. Some subjects contributed more than one specimen to the analysis; thus, statistical methods that adjust for intrasubject dependence (i.e., correlation between multiple samples taken from a single individual) were used. When comparing proportions positive for staining across histological categories, Brier’s correction for the ordinary x2 test was used.22 All other comparisons were made via logistic regression for binary outcomes,23 with corrections for intrasubject correlation performed by the method of Liang and Zeger,24 assuming constant correlation structure.
Results Immunostaining for cyclin D1 was performed on 134 paraffin-embedded tissue specimens taken from 69 subjects (age range, 18–93 years; mean, 63.8 years). Clinical characteristics of patients and tissue samples are summarized in Table 1. Positive nuclear staining for cyclin D1 was observed in 8 of the 27 adenocarcinoma (30%) and 12 of the 35 adenomatous polyps (34%) (Figure 2). No nuclear staining was detected in eight hyperplastic polyps, 23 normal-appearing mucosa, and 41 transitional mucosa samples. Positive nuclear staining was not related to the sex or age of the patient, to the size and degree of dysplasia in adenomatous polyps, or to the differentiation and Dukes’ stage of adenocarcinomas. WBS-Gastro
March 1996
CYCLIN D1 IN COLON NEOPLASIA 671
Thus, it was observed with about equal frequency in Dukes’ A – D lesions. The number of tumors with positive staining was significantly greater in lesions from the left-sided colon, i.e., rectum, sigmoid colon, and descending colon (47%), than from the right-sided colon, i.e., transverse colon, ascending colon, and cecum (12%) (P õ 0.05). This was true even after controlling for diagnosis (adenomatous polyps vs. adeno-
Figure 1. Cyclin D1–positive supranuclear cytoplasmic staining (A ) in normal mucosa adjacent to an adenocarcinoma and (B ) in an adenomatous polyp and (C ) diffuse cytoplasmic staining in a hyperplastic polyp.
/ 5e0a$$0001
02-09-96 11:25:28
gasa
carcinomas). Eleven of 22 (50%) of the polyps and 6 of 14 (43%) of the adenocarcinomas in the left-sided colon were positive, compared with only 1 of 13 (8%) of the polyps and 2 of 13 (15%) of the adenocarcinomas in the right-sided colon (Table 2). The percentage of cyclin D1 – positive staining nuclei were similar in adenomatous polyps and adenocarcinomas. In about 60% of cyclin D1 – positive tumors, 25% – 50% of the nuclei showed immunohistological staining, and in 30% of the cases, ú50% of nuclei were positive. Adenomatous polyps showed positive nuclei almost exclusively near
Figure 2. Cyclin D1–positive nuclear staining in an (A ) adenomatous polyp and (B and C ) adenocarcinoma.
WBS-Gastro
672 ARBER ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
Table 1. Clinical Characteristics of Patients Showing Cyclin D1 Immunoreactivity Normal mucosa No. of samples Sex (M/F) (%) Mean age (yr) Left-sided colon samples (%) Right-sided colon samples (%) Positive nuclear stain (%) Positive cytoplasmic stain (in the lesion)b (%) Positive cytoplasmic stain (in adjacent normal mucosa)c (%)
Transitional mucosa
Hyperplastic polyps
23 59/41 60.4 55 45 0 22
41 40/60 66.1 59 41 0 59
8 63/37 66.4 88 12 0 38
NA
NA
0
a
Adenomatous polyps
Adenocarcinoma
35 49/51 67.2 63 37 34 34
27 65/35 64.5 52 48 30 26
33 (7 of 21)
85 (17 of 20)
NA, Not applicable. a Includes both resection margin and normal colonoscopy samples. b Refers to percent of samples in each histological category that showed positive cytoplasmic staining in the lesion itself. c Refers to percent of positive cytoplasmic staining of normal mucosa adjacent to each histological category.
the luminal surface, whereas adenocarcinomas showed positive cells diffusely. Positive cyclin D1 cytoplasmic staining was detected in many of these samples. It was identified in both neoplastic and nonneoplastic tissues. The incidence of cytoplasmic staining for cyclin D1 in the various histological categories is summarized in Table 3. Positive cytoplasmic staining was noted in about 30% of adenomatous polyps and adenocarcinomas, similar to the frequency of nuclear staining. The frequency of positive cytoplasmic cyclin D1 staining was not related to the sex or age of the patient or to the site of the tissues, the size or degree of dysplasia in polyps, or Dukes’ stage and differentiation in the adenocarcinomas (Table 2). Only 9% of adenomatous polyps and 4% of adenocarcinomas had both positive cyclin D1 nuclear and cytoplasmic staining, whereas 26% of adenomatous polyps and 22% of adenocarcinomas showed positive cytoTable 2. Cyclin D1 Nuclear and Cytoplasmic Immunoreactivity in Adenomas and Adenocarcinomas
Sex (M/F) Left-sided colon Adenomatous polyps Adenocarcinomas Both lesions Right-sided colon Adenomatous polyps Adenocarcinomas Both lesions Dysplasia in adenomas Absent Present
No. of samples
Samples positive for nuclear staining
Samples positive for cytoplasmic staining
27/34
9 (33)/11 (32)
9 (33)/10 (29)
22 14 36
11 (50) 6 (43) 17 (47)
6 (27) 5 (36) 11 (31)
13 13 26
1 (8) 2 (15) 3 (12)
6 (46) 2 (15) 8 (31)
12 23
4 (33) 8 (35)
6 (50) 9 (39)
02-09-96 11:25:28
Table 3. Nuclear and Cytoplasmic Cyclin D1 Immunoreactivity in Different Colonic Lesions
Normal mucosa Transitional mucosa Hyperplastic polyp Adenomatous polyps Adenocarcinomas
NOTE. Values in parentheses indicate the percentage of positive samples of tumors.
/ 5e0a$$0001
plasmic staining in the absence of nuclear staining. Cytoplasmic staining also was observed in 38% of the hyperplastic polyps and in 22% of the normal mucosa samples (Table 3). It is of interest that, in the latter group, all of the positive cases were apparently normal mucosa samples taken from the resection margins of tumors. Thus, 5 of 13 (38%) of the latter samples were positive, whereas none of the 10 samples obtained from patients without tumors was positive. Positive cytoplasmic staining was observed in 59% of transitional mucosa samples. It was associated with adenocarcinomas in 85% of the cases compared with only 33% with the adenomatous polyps (P õ 0.05). We noted two different patterns of cytoplasmic staining for cyclin D1. Some samples showed a diffuse pattern of staining, and other samples showed punctate discrete intense staining with a supranuclear location (Figure 1). Supranuclear localization was observed in 6 of the 7 (86%) adenocarcinomas and 10 of the 12 (83%) adenomatous polyps showing cytoplasmic staining. In most of
gasa
No. of samples
Samples positive for nuclear staining (%)
23a 41 8 35 27
0 0 0 12 (34) 8 (30)
Samples positive for cytoplasmic staining (%) 5 24 3 12 7
(22)b (59) (38) (34) (26)
NOTE. Values in parentheses designate the percentage of samples that were positive. a Includes both resection margin and normal colonoscopic biopsy samples. b All of the positive samples came from resection margins of tumors.
WBS-Gastro
March 1996
CYCLIN D1 IN COLON NEOPLASIA 673
these samples, 80% of the adenomatous polyps and 83% of the adenocarcinomas, this cytoplasmic staining occurred in the absence of positive nuclear staining. Thus, cytoplasmic staining for cyclin D1 is observed quite frequently in both normal colonic mucosa and colon tumors. Our data suggest that it occurs with increased frequency in transitional mucosa, polyps, and adenocarcinomas, but the differences between these categories did not reach statistical significance.
Discussion The major finding in this study is that about 30% of adenomatous polyps and colorectal adenocarcinomas showed increased nuclear expression of the cyclin D1 protein, whereas hyperplastic polyps and normal colonic mucosa showed no nuclear expression of this protein (Table 1). These findings are intriguing because there have been no reports of translocation or amplification involving the 11q13 chromosome region (the site of the cyclin D1 gene) in human colon cancer25 and no amplification of the cyclin D1 gene has been found in any of 47 colorectal cancer cell lines.19 Similarly, we found no evidence for cyclin D1 gene amplification in the 18 primary human colorectal cancers and six human colorectal cancer cell lines that were examined, despite high levels of cyclin D1 messenger RNA and protein in around half of these cases (Sutter et al., unpublished data, February 1995). Therefore, it is unlikely that gene amplification accounts for the high frequency of increased expression of the cyclin D1 protein that we have observed in this histochemical study. Cyclin D1 protein accumulation was detected in about one third of both dysplastic and nondysplastic adenomatous polyps and in both small and large lesions and was also observed with a similar frequency in colon adenocarcinomas, regardless of their differentiation or Dukes’ stage. Therefore, increased expression of this gene seems to be an early event in colorectal tumorigenesis. Recently, K-ras and APC mutations were reported in aberrant crypt foci, which may represent the earliest lesion in colorectal tumorigenesis.26 It will be interesting to examine whether or not cyclin D1 is overexpressed in these foci. Increased cyclin D1 expression was significantly more prevalent in left- than right-sided neoplastic lesions, providing further evidence of differences between left- and right-sided colonic tumors.27 The distribution of cyclin D1–positive nuclei in adenomas was restricted primarily to the superficial aspects of the lesion. This contrasts with the diffuse distribution of positive nuclei in adenocarcinomas. It also is of interest to note that nonneoplastic crypts showed no detectable nuclear staining. This suggests that physiological levels / 5e0a$$0001
02-09-96 11:25:28
gasa
of cyclin D1 in colonic crypts are too low for detection. This observation supports the importance of abnormal levels of expression of this protein in colonic neoplasia. We should emphasize that the polyclonal antibodies used in the present study recognize both cyclin D1 and cyclin D2 proteins. However, Bartkova et al.,28 using monoclonal antibodies specific to cyclin D1, showed nuclear staining in colonic adenocarcinoma similar to that in the present study. In addition, we have found that 9 of 18 primary colorectal adenocarcinomas and 3 of 6 human colon cancer cell lines showed a twofold to threefold increase in the expression of cyclin D1 but not cyclin D2, messenger RNA, and protein (Sutter et al., unpublished data, February 1995). Therefore, it is likely that the cyclin D1 overexpression detected in the present study is caused by increased cyclin D1 protein. The demonstration of cyclin D1 expression in the cytoplasm of colonic epithelial cells is novel. We found that this occurred in a relatively high proportion of our specimens, including hyperplastic polyps, transitional mucosa, and normal mucosa obtained from the resection margins of tumors (Tables 2 and 3). Cytoplasmic staining is not caused by leakage of this protein from the nucleus because it was observed frequently in the absence of nuclear staining. Moreover, cytoplasmic staining frequently had a supranuclear distribution in the region of the Golgi apparatus. This pattern was observed more often in the transitional mucosa of adenocarcinomas (85%) than in the transitional mucosa of adenomatous polyps (33%) (P õ 0.05). Cytoplasmic immunostaining for cyclin D1 also has been described in some lymphomas,29 and the presence of cyclin D1 outside the nucleus suggests an additional presently undefined physiological role. Its occurrence in the transitional colonic mucosa may reflect reactivity of the normal mucosa to the adjoining tumor or may be caused by the inherent biology of colonic neoplasms. Further studies are necessary to understand this phenomenon more fully.
References 1. Pines J, Hunter T. Cyclins and cancer II: cyclin D1 and CDK inhibitors come of age. Cell 1994;79:573–582. 2. Sherr CJ. Mammalian G1 cyclins. Cell 1993;73:1059–1065. 3. Lew DJ, Dulic V, Reed SI. Isolation of three novel human cyclins by rescue of G1 cyclin function in yeast. Cell 1991;66:1197– 1206. 4. Xiong Y, Connolly T, Futcher B, Beach D. Human D-type cyclin. Cell 1991;65:691–699. 5. Baldin V, Lukas J, Marcote MJ, Pagano M, Draetta G. Cyclin D1 is a nuclear protein required for cell cycle progression in G1. Genes Dev 1993;7:812–821. 6. Jiang W, Kahn SM, Zhou P, Zhang YJ, Cacace AM, Infante AS, Doi S, Santella RM, Weinstein IB. Overexpression of cyclin D1 in rat fibroblasts causes abnormalities in growth control, cell
WBS-Gastro
674 ARBER ET AL.
7.
8.
9.
10.
11.
12.
13. 14. 15.
16. 17.
18.
19.
GASTROENTEROLOGY Vol. 110, No. 3
cycle progression and gene expression. Oncogene 1993; 8:3447–3457. Quelle DE, Ashmun RA, Shurtleff SA, Kato J-Y, Bar-Sagi D, Roussel HF, Sherr CJ. Overexpression of mouse D-type cyclins accelerates G1 phase in rodent fibroblasts. Genes Dev 1993;7:1559– 1571. Wang TC, Cardiff RD, Zukerberg L, Lees E, Arnold A, Schmidt EV. Mammary hyperplasia and carcinoma in MMTV-cyclin D1 transgenic mice. Nature 1994;369:669–671. Hinds PW, Dowdy SF, Ng Eaton E, Arnold A, Weinberg RA. Function of a human cyclin gene as an oncogene. Proc Natl Acad Sci USA 1994;91:709–713. Lovec H, Sewing A, Lucibello FC, Muller R, Moroy T. Oncogenic activity of cyclin D1 revealed through cooperation with Ha-ras: link between cell cycle control and malignant transformation. Oncogene 1994;9:323–326. Jiang W, Zhang Y-J, Kahn SM, Hollstein MC, Santella RM, Lu SH, Harris CC, Montesano R, Weinstein IB. Altered expression of the cyclin D1 and retinoblastoma genes in human esophageal cancer. Proc Natl Acad Sci USA 1993;90:9026–9030. Zhang YJ, Jiang W, Chen CJ, Lee CS, Kahn SM, Santella RM, Weinstein IB. Amplification and overexpression of cyclin D1 in human hepatocellular carcinoma. Biochem Biophys Res Commun 1993;196:1010–1016. Hunter T, Pines J. Cyclins and cancer. Cell 1991;66:1071– 1074. Marx J. How cells cycle toward cancer. Science 94:263:319– 321. Keyomarsi K, Pardee AB. Redundant cyclin overexpression and gene amplification in breast cancer cells. Proc Natl Acad Sci USA 1993;90:1112–1116. Lammie GA, Peters G. Chromosome 11q13 abnormalities in human cancer. Cancer Cells 1991;3:413–420. Gillett C, Fantl V, Smith R, Fisher C, Bartek J, Dickson C, Barnes D, Peters G. Amplification and overexpression of cyclin D1 in breast cancer detected by immunohistochemical staining. Cancer Res 1994;54:1812–1817. Buckley MF, Sweeney KJE, Hamilton JA, Sini RL, Manning DL, Niccholson RI, DeFazio A, Watts CKW, Musgrove EA, Sutherland RL. Expression and amplification of cyclin genes in human breast cancer. Oncogene 1993;8:2127–2133. Leach SF, Elledge SJ, Sherr CJ, Wilson JKV, Markowitz S, Kinzler
/ 5e0a$$0001
02-09-96 11:25:28
gasa
20. 21.
22. 23.
24. 25. 26.
27.
28.
29.
KW, Vogelstein B. Amplification of cyclin genes in colorectal carcinomas. Cancer Res 1993;53:1986–1989. Pines J. Cyclins: wheels within wheels. Cell Growth Differ 1991;2:305–310. Hall FL, Williams RT, Wu L, Wu F, Carbonaro-Hall DA, Harper JW, Warburton D. Two potentially oncogenic cyclins, cyclin A and cyclin D1, share common properties of subunit configuration, tyrosine phosphorylation and physical association with the Rb protein. Oncogene 1993;8:1377–1384. Brier SS. Analysis of contingency tables under cluster sampling. Biometrika 1980;67:591–596. Breslow NE, Day NE. Statistical methods in cancer research, volume 1: the analysis of case control studies. Lyon, France: International Agency for Research on Cancer, 1980. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73:13–22. Muleris M, Salmon RJ, Dutrillaux B. Cytogenetics of colorectal adenocarcinomas. Cancer Genet Cytogenet 1990;46:143–156. Pretlow TS. Aberrant crypt foci and K-ras mutations: earliest recognized players or innocent bystanders in colon carcinogenesis? Gastroenterology 1995;108:600–603. Liu T, Distler P, Mokoulu AD, Reddy BS, Zang E, Holt PR. Regional differences in carcinogene induced colon neoplasia—effects of chemoprevention. Gastroenterology (in press). Bartkova J, Lukas J, Strauss M, Bartek J. The prad-1 cyclin D1 oncogene product accumulates aberrantly in a subset of colorectal carcinomas. Int J Cancer 1994;58:568–573. Banno S, Yoshikawa K, Nakamura S, Yamamoto K, Seito T, Nitta M, Takahashi T, Ueda R, Seto M. Monoclonal antibody against PRAD1/cyclin D1 stains nuclei of tumor cells with translocation or amplification at BCL-1 Locus. Jpn J Cancer Res 1994;85:918– 926.
Received June 22, 1995. Accepted October 6, 1995. Address requests for reprints to: Peter R. Holt, M.D., Division of Gastroenterology, Department of Medicine, St. Luke’s Hospital Center, 1111 Amsterdam Avenue, New York, New York 10025. Fax: (212) 523-3683. Supported by grants from the Dorot Foundation (to N.A.), the National Dairy Council (to P.R.H.), the National Foundation for Cancer Research (to I.B.W.), and the Markey Charitable Trust (to I.B.W.). The authors thank the Department of Pathology at St. Luke’s/ Roosevelt Hospital for valuable assistance in these studies.
WBS-Gastro