Astrocytoma is the most common primary neoplasm of

Transcription

1 Overexpression of Cyclin A and Cyclin B1 Proteins in Astrocytomas Katherine Allan; Richard C. K. Jordan, DDS, PhD; Lee-Cyn Ang, MD, FRCPC; Michael Taylor, MD; Beverley Young, MLT Background. Cyclins are proteins that are expressed during the progression of a normal cell through the cell cycle. In a number of cancers, overexpression of cyclin A and cyclin B1 proteins has been reported, and in some instances the levels of expression correlated well with the grades of malignancy. The expression of cyclin A and cyclin B1 proteins in astrocytoma may be linked to the histologic grade or proliferative activities. Objective. To study the expression of cyclin A and cyclin B1 proteins in astrocytomas and correlate the labeling indices (LIs) of cyclin A and cyclin B1 with histologic grade and Ki-67 LI. Design. The surgical biopsy specimens from 65 adults with astrocytomas were reviewed and divided into grades based on the World Health Organization system. The paraffin sections were immunostained using primary antibodies against Ki-67, cyclin A, and cyclin B1. The LIs of these astrocytomas for the 3 different antibodies were determined by computerized image analysis. Results. The cyclin A LI showed good correlation with astrocytoma grade and Ki-67 LI. Both the nuclear and cytoplasmic cyclin B LIs correlated well with the tumor grade but showed poor correlation with Ki-67 LI. Conclusions. This study suggests that although both cyclin A and B protein expression are related to the grade of malignancy in astrocytomas, cyclin A levels more generally reflect the proliferative state of these tumors. We also provide indirect evidence that cyclin B1 is associated with the aberrant progression through the G2-M phase checkpoint in astrocytomas. (Arch Pathol Lab Med. 2000;124: ) Astrocytoma is the most common primary neoplasm of the central nervous system, representing between 30% and 40% of all tumors. Histologic grading has been regarded as one of the most important predictors of outcome in patients with astrocytomas, but this assessment is subjective and depends on the experience and training of individual pathologists. 1 In addition, there are different grading systems in use, making comparison of outcomes and treatment modalities in various centers difficult. Because of these factors, a number of markers of cellular proliferation have been used with the hope of providing more objectivity to the grading of astrocytomas. The best known among these markers are the proliferating cell nuclear antigen (PCNA), 2,3 Ki-67, 4 6 and the incorporation of bromodeoxyuridine. 4,6,7 Both PCNA 2,3 and Ki have been successfully applied to the study of neoplasms in a number of various tissues, including routinely processed archival material fixed in formalin and embedded in paraffin. Previous studies have demonstrated that PCNA may not be as reliable because the intensity of nuclear immunoreactivity is variable from cell to cell within the same tumor, ranging from very dark to hardly visible. 2,3 In addition, PCNA immunostaining is also adversely affected by an increase in fixation duration. 2,3 Comparing the prog- Accepted for publication July 9, From the Department of Pathology, Sunnybrook and Women s College Health Science Center, University of Toronto, Toronto, Ontario. Reprints: Lee-Cyn Ang, MD, FRCPC, Neuropathology Laboratory, Sunnybrook and Women s College Health Science Center, University of Toronto, 2075 Bayview Ave, Toronto Ontario M4N 3M5, Canada ( lee.ang@utoronto.ca). nostic power of PCNA LI, Ki-67 LI, and flow cytometry (S phase and S and G2-M fractions, respectively) in 50 astrocytomas, Sallinen et al 6 found that Ki-67 (MIB-1) LI correlated best with prognosis. Similarly, McKeever et al 4 concluded that Ki-67 (MIB-1) was superior to PCNA and bromodeoxyuridine as a proliferation marker in gliomas. Cyclins belong to a family of proteins that are sequentially produced and destroyed during the progression of a normal cell through the cell cycle. 8 In a number of cancers, aberrant expression of cyclin proteins has been reported. Moreover studies have shown that cyclins and cyclin-dependent kinase are linked to malignant progression of astrocytoma In the present study, we examined the expression of two cell cycle proteins, cyclin A and cyclin B1, in diffuse astrocytomas in adults and correlated their immunoreactivity with the histologic grades and Ki-67 LI. MATERIALS AND METHODS Cases of adult supratentorial astrocytoma during a 9-year period ( ) were retrieved from the surgical pathology files of the Department of Pathology, Sunnybrook and Women s College Health Science Center, Toronto, Ontario. Juvenile pilocytic astrocytoma, optic glioma, subependymal giant cell astrocytoma, pleomorphic xanthoastrocytoma, ganglioma, and mixed glioma with a prominent oligodendrogliomatous component were excluded. Sixty-five cases were selected after the diagnoses of astrocytoma was confirmed. The histologic grades of these tumors were reassessed using the World Health Organization Histological Typing of Tumors of the Central Nervous System. Based on this system, there were 19 grade II, 18 grade III, and 28 grade IV astrocytomas. 13 All selected cases had sufficient material for study and had been fixed in 4% buffered formaldehyde for approximately 24 hours and embedded in paraffin. 216 Arch Pathol Lab Med Vol 124, February 2000 Cyclin A and Cyclin B1 in Astrocytomas Allen et al

2 Five-micrometer serial sections were cut and mounted on adherent glass slides. These sections were then dewaxed in xylene and rehydrated in graded ethanols. Endogeneous peroxidase activity was blocked by immersion in 0.3% methanolic peroxide for 15 minutes. Pressurized heat antigen retrieval (microwave pressure cooking) was used to enhance the immunoreactivity of the target antigens. The sections were placed in a pressure cooker containing 10 mmol/l sodium citrate buffer and heated in a microwave oven for 20 minutes. After removal, the sections were cooled in deionized water then rinsed in phosphate-buffered saline. Sections were incubated for 10 minutes with horse serum (Vector Laboratories, Burlington, Ontario) that was diluted 1:15 in 1 phosphate-buffered saline to block nonspecific antibody binding. The monoclonal antibodies used were cyclin A (NCL- Cyclin A; Novocastra Laboratories Ltd, Newcastle-upon-Tyne, England) diluted at 1:500 in antibody-diluting buffer (Dimension Laboratories, Burlington, Ontario), cyclin B1 (NCL-Cyclin B1; Novocastra) diluted at 1:500, and Ki-67 (NCL-Ki-67-MM1; Novocastra) diluted at 1:1000. All incubations were carried out overnight at 4 C. For cyclin A and Ki-67, the sections were processed using standard avidin-biotin complex (ABC) immunohistochemistry technique according to manufacturer s recommendations (Vector). Cyclin B1 protein was detected using the Elite Vectastain ABC Kit (Vector). All sections were lightly counterstained in hematoxylin. For positive controls, specimens of surgically resected tonsils and adenocarcinoma were used. Normal cerebral cortex and omission of the primary antibody were used as negative control. To determine the percentage of cells showing nuclear staining for cyclin A, Ki-67, or cyclin B1 as a proportion of the total tumor cell population in astrocytomas, a computer image analysis system (IBAS System, Kontron Electronik GMBH, Dusseldorf, Germany) was used. 2 Images captured at 200 magnification were altered using a gray scale distribution, with 0 as the darkest level and 225 as the lightest. Gray levels from 0 to 50 were used to identify cells in the field of view and lower gray levels from 0 to 30 to enhance only positively stained cells. These cells were subsequently discriminated and counted. The fields chosen for image analysis were those with the highest density of positive cellular immunostaining in each case. The labeling index of a tumor expressed as a percentage was determined by dividing the number of positive cells by the total number of cells (at least 1000 cells counted). Necrotic tissue, blood vessels, and leukocytes were excluded from the quantification process. Since cyclin B1 has both a nuclear (M phase) and cytoplasmic location (G2 phase), the cytoplasmic immunostaining was scored separately to give the cytoplasmic cyclin B1 LI. For the statistical analysis, the cyclin A, nuclear cyclin B1, and cytoplasmic cyclin B1 LIs were each correlated separately with the histologic grade using the nonparametric Kruskal-Wallis test. The results were considered statistically significant if P equaled.05 or less. The Pearson correlation coefficient was used to determine the association among cyclin A, nuclear cyclin B1, cytoplasmic cyclin B1, and Ki-67 LIs. RESULTS The mean LIs for Ki-67, cyclin A, and B1 protein expression in the 3 grades of astrocytomas are summarized in the Table. Normal brain tissues showed no immunostaining with the monoclonal antibodies for Ki-67, cyclin A, and cyclin B1 proteins. The immunostaining for cyclin A was confined to the nuclei, and the intensity of staining was consistent Figure 1. Cyclin A immunoreactivity in grade II astrocytoma (A), grade III astrocytoma (B), and grade IV astrocytoma (C), demonstrating an increase in immunolabeling with tumor grade (cyclin A immunostaining, original magnification 400). Arch Pathol Lab Med Vol 124, February 2000 Cyclin A and Cyclin B1 in Astrocytomas Allen et al 217

3 Labeling Indices* of Cyclins in Different Grades of Astrocytomas Tumor Grade Ki-67, % Cyclin A, % Nuclear Cyclin B1, % Cytoplasmic Cyclin B1, % Grade II Grade III Grave IV * Values are means standard error throughout each tumor. Generally, there was a corresponding increase in number of tumor cell nuclei stained with the increase in the grade of the astrocytoma (Figure 1, A through C). The regions of maximal cyclin A nuclear immunoreactivity corresponded topographically with areas of maximal Ki-67 immunoreactivity. The mean cyclin A LI among the 3 grades of astrocytomas was significantly different (P.0005) (Figure 2, A). Although there was some overlap of data points among the 3 grades of tumors, there was no overlap of the 95% confidence intervals of the grade II and IV tumors. There was significant correlation between the Ki-67 and cyclin A LIs (Pearson correlation coefficient r 0.71) for individual tumors (Figure 2, B). Positive immunostaining for cyclin B1 was noted in both the cytoplasm and the nuclei of the tumor cells. In all cases, there were mixtures of cells with nuclear staining only, cells with cytoplasmic staining only, and cells with both nuclear and cytoplasmic staining. The number of cells immunostained (including nuclear and/or cytoplasmic staining) with cyclin B1 in the grade III and IV astrocytomas was much lower when compared with cyclin A immunostaining for the same tumors. There was also a similar trend as in the cyclin A immunoreactivity, with tumors having an increased cyclin B1 labeling as the grade increased (Figure 3, A through C). The nuclear and cytoplasmic B1 staining was discriminated and counted separately to produce nuclear LI and cytoplasmic LI. The mean of the nuclear cyclin B1 LI was significantly different among the 3 grades of astrocytomas (P.005) (Figure 4, A). A more significant difference between the means was also noted for the cytoplasmic LI (P.0005) and tumor grade (Figure 4, B). Lower-grade astrocytomas (grades II and III) showed overlapping data points for nuclear and cytoplasmic cyclin B1, but there was no overlap of the 95% confidence intervals between grades II and IV tumors. When comparing the Ki-67 LIs, there was a moderate correlation for nuclear cyclin B1 LI (r 0.45) but almost none with cytoplasmic LI (r 0.04). Similarly, there was good correlation between nuclear cyclin B1 LI (r 0.45) and cyclin A but poor correlation between cytoplasmic cyclin B1 LI (r 0.17) and cyclin A LI. COMMENT Cyclins regulate the progression of cells through the cell cycle by exerting their influence on their kinase partners. 8,14 These proteins are homologous in a 100 amino acid domain termed the cyclin box, which is essential for cyclin-dependent kinase binding and activation. Different cyclins appear at different phases of the cell cycle and are then rapidly destroyed. A region of the protein, termed the destruction box, is required for the elimination of cyclin. Loss of the destruction box results in persistence of the protein. In healthy cells, cyclin A is most abundantly expressed in S phase, coinciding with the onset of DNA synthesis. 15 It is also required for S-phase progression and for passage from G2 phase into mitosis. 16 In hepatocellular carcinoma, integration of cyclin A and hepatitis B virus genes occurs, forming a fusion chimeric protein with the translated cyclin A protein, lacking the destruction box, and leading to persistence of the protein. 17 Therefore, the aberrant expression of cyclin A has been implicated in the pathogenesis of hepatocellular carcinoma. Normally, cyclin B is localized to the cytoplasm, and the translocation of cyclin B cyclin-dependent kinase inhibitor complex into the nucleus signals the onset of mitosis. 18 For the cell to exit mitosis, cyclin B1 destruction is accomplished through the ubiquitin-dependent pathway. 19 Cyclin B1, with its associated cyclin-dependent kinase, controls the cell progression from G2 to M phase. Cyclin overexpression has been observed in many human tumors. For example, cyclin D1, a G1 protein, is overexpressed in breast, 20,21 oesophageal, 22,23 rectal, 24 and head and neck carcinoma. 25 In addition, cyclin D1 protein expression had been shown to increase with histologic grade malignancy in astrocytomas, although one study has shown that the correlation between MIB-1 and cyclin Figure 2. Graphs of cyclin A labeling index versus astrocytoma grade (A) and cyclin A labeling index versus Ki-67 labeling index (B). 218 Arch Pathol Lab Med Vol 124, February 2000 Cyclin A and Cyclin B1 in Astrocytomas Allen et al

4 D1 LIs was not statistically significant. 9 In an earlier study, however, there was some correlation between the cyclin D1 and Ki-67 LIs in diffuse astrocytomas but not in pilocytic astrocytoma and glioblastomas. 10 Another cell cycle protein, cyclin E, which marks the G1 and early S phase, has been noted to increase expression with the grade of breast carcinoma. 26 The overexpression of cyclin A protein has also been reported in breast, prostatic, esophageal, and oral cancers Overexpression of cyclin B1 protein has been noted in malignant breast, kidney, prostatic, and oral neoplasms. 26,27,29 Dirks et al 12 have previously shown higher cyclin expression, including cyclin A and cyclin B1, in human astrocytoma cell lines by northern blot, immunoblot, and immunohistochemistry. In the present study, cyclin A protein expression was consistently present in all the grades (II IV) of astrocytoma but was absent in the controls of normal cerebral cortex. The protein was localized to the nuclei of tumor cells and the intensity of immunostaining was consistent, permitting easy quantitative image analysis. The cyclin A LI also showed a significant correlation with the World Health Organization grades and the LI of the proliferative marker, Ki-67. Our findings support those of Chakrabarty and Bridges, 11 who analyzed 41 astrocytomas (including 9 pilocytic astrocytomas) by immunohistochemistry and found a good correlation between the cyclin A and MIB- 1 LIs. The overexpression of cyclin A in astrocytoma is consistent with findings in carcinomas outside the central nervous system, such as those in the breast, prostate, and oral cavity, where cyclin A levels also correlate strongly with the Ki-67 levels. 26,27,29 The good correlation between cyclin A and Ki-67 levels suggests that increases in cyclin A protein expression reflect the increasing proliferative activity in astrocytoma with the higher tumor grades. Alternatively, the increased cyclin A levels in astrocytomas could be the result of some undetermined genetic alterations in neoplasia or protein up-regulation in response to other events in tumoriogenesis such as dysregulation of apoptosis. 11,28,30 In our study, cyclin B1 protein was also overexpressed in astrocytomas but was not expressed in the normal brain. There was a significant trend of increasing nuclear and cytoplasmic LIs with higher-grade tumors. The correlation with both Ki-67 and cyclin A LIs was moderate for nuclear cyclin B1 LI and poor for cytoplasmic cyclin B1 LI. A recent study of breast cancers found a poor correlation between cyclin B1 LI and Ki-67 LI, and this has been attributed to difficulties measuring the small fraction of cells expressing cyclin B1. 26 Kushner et al, 29 however, using computer image analysis to accurately quantify the nuclear expression of cyclin B1 protein in oral cancers, suggested that the lack of correlation between cyclin B1 and Ki-67 could be due to aberrant cell cycle progression at the G2-M checkpoint in such tumors. Our findings in adult astrocytomas indicate that although there is no correlation between cyclin B1 with either cyclin A or Ki-67 levels, there is a significant increase in cyclin B1 protein expression with higher grades, supporting the notion that Figure 3. Cyclin B1 immunoreactivity in grade II astrocytoma (A), grade III astrocytoma (B), and grade IV astrocytoma (C), demonstrating the increase in immunolabeling with tumor grade (cyclin B1 immunostaining, original magnification 400). Arch Pathol Lab Med Vol 124, February 2000 Cyclin A and Cyclin B1 in Astrocytomas Allen et al 219

5 Figure 4. Graphs of nuclear cyclin B1 labeling index versus astrocytoma grade (A) and cytoplasmic cyclin B1 labeling index versus astrocytoma grade (B). its expression reflects aberrant progression at the G2-M checkpoint rather than the tumor proliferative activities. As such, it would be of clinical interest to determine the relationship between tumor cyclin B1 levels and their response to radiotherapy, since tumor cells appear to be most sensitive to the effects of radiation in the G2-M phase. 31 In summary, the present study has demonstrated that overexpression of both cyclin A and cyclin B1 proteins in adult astrocytomas correlates with the grade of malignancy. Although the expression of cyclin A protein has been previously analyzed in astrocytic tumors in vivo, 11 that of cyclin B1 has not, to our knowledge, been studied before. The strong correlation of cyclin A and Ki-67 suggests that cyclin A can be used as another tool for assessing the proliferation of astrocytomas and therefore their biological behavior. Although the overexpression of cyclin B1 protein did not correlate well with Ki-67 LI, our results provide indirect evidence of aberrant cell cycle progression at the G2-M checkpoint and indicate the need for further studies examining the relationship of this event with response to radiotherapy. We thank Norma Blythe and Cindy Fedell for their help in the preparation of the manuscript. Katherine Allan was supported by the Sunnybrook Trust Undergraduate Summer Student Program consecutively for 1997 and References 1. Kim TS, Halliday AL, Hedley-White ET, Convery K. Correlates of survival and Daumas-Duport grading system for astrocytomas. J Neurosurg. 1991;74: Ang LC, Plewes M, Tan L, Begley H, Aggranovich A, Shul D. Proliferating cell nuclear antigen expression in survival of astrocytoma patients. Can J Neurol Sci. 1994;21: Theunissen P, Blaaw G. Proliferating cell nuclear antigen immunostaining and survival in cerebral astrocytoma. Histopathology. 1993;23: McKeever PE, Ross DA, Strawderman MS, Brunberg JA, Greenberg HS, Junck L. A comparison of predictive power for survival in gliomas provided by MIB-1, bromodeoxyuridine and proliferating cell nuclear antigen with histopathologic and clinical parameters. J Neuropathol Exp Neurol. 1997;56: Montine T, Vandersteenhoven J, Aguzzi A, et al. Prognostic significance of Ki-67 proliferation index in supratentorial fibrillary astrocytic neoplasms. Neurosurgery. 1994;34: Sallinen PK, Haapasalo HK, Visakorpi T, et al. Prognostication of astrocytoma patient survival by Ki-67 (MIB-1), PCNA, and S-phase fraction using archival paraffin-embedded samples. J Pathol. 1994;174: Labrousse F, Daumas-Duport C, Batorski L, Hoshino T. Histological grading and bromodeoxyuridine labeling index of astrocytomas. J Neurosurg. 1991;75: Pines J. Cyclins, CDKs and cancer. Semin Cancer Biol. 1995;6: Cavalla P, Dutto A, Piva R, Richardi P, Grosso R, Schiffer D. Cyclin Di expression in gliomas. Acta Neuropathol. 1998;95: Chakrabarty A, Bridges LR, Gray S. Cyclin D1 in astrocytic tumors: an immunohistochemical study. Neuropathol Appl Neurobiol. 1996;22: Chakrabarty A, Bridges LR. Immunohistochemical analysis of cyclin A in astrocytic tumors. Neuropathol Appl Neurobiol. 1998;24: Dirks PB, Hubbard SL, Murakami M, Rutka JT. Cyclin and cyclin-dependent kinase expression in human astrocytoma cell lines. J Neuropathol Exp Neurol. 1997;56: Kleihues P, Burger PC, Scheithauer BW. Histological Typing of Tumors of the Central Nervous System. 2nd ed. Berlin, Germany: Springer-Verlag; 1993: Norbury C, Nurse P. Animal cell cycles and their control. Annu Rev Biochem. 1993;61: Girard F, Strausfeld U, Fernandez A, Lamb NJ. Cyclin A is required for onset of DNA replication in mammalian fibroblasts. Cell. 1991;67: Henglein B, Chenivesse X, Wang J, Eick D, Brechot C. Structure and cell cycle-related transcription of the human cyclin A gene. Proc Natl Acad Sci U S A. 1994;91: Wang J, Zindy F, Chenivesse X, Lamas E, Henglein B, Brechot C. Modification of cyclin A expression by hepatitis B virus DNA integration in a hepatocellular carcinoma. Oncogene. 1992;7: Pines J, Hunter T. The differential localization of human cyclins A and B is due to a cytoplasmic retention signal in cyclin B. EMBO J. 1994;13: Holloway SL, Glotzer M, King RW, Murray AW. Anaphase is initiated by proteolysis rather than by the inactivation of maturation-promoting factor. Cell. 1993;73: Bartkova J, Lukas J, Muller H, Lutzhoft D, Strauss M, Bartek J. Cyclin D1 protein expression and function in human breast cancer. Int J Cancer. 1994;57: Gillet C, Fant V, Smith R, et al. Amplification and overexpression of cyclin D1 in breast cancer detected by immunohistochemical staining. Cancer Res. 1994;54: Jiang W, Kahn SM, Tomita N, Zhang Y, Lu S, Weinstein IB. Amplification and overexpression of the human cyclin D gene in esophageal cancer. Cancer Res. 1992;52: Jiang W, Zhang Y, Kahn SM, et al. Altered expression of cyclin D1 and retinoblastoma genes in human esophageal cancer. Proc Natl Acad Sci U S A. 1993;90: Bartkova J, Lukas J, Strauss M, Bartek J. The PRAD-1/cyclin D1 oncogene product accumulates aberrently in a subset of colorectal carcinomas. Int J Cancer. 1994;58: Bartkova J, Lukas J, Muller H, Strauss M, Gusterson B, Bartek J. Abnormal patterns of D-type cyclin expression and G1 regulation in human head and neck cancer. Cancer Res. 1995;55: Dutta A, Chandra R, Leiter LM, Lester S. Cyclins as markers of tumor proliferation: immunocytochemical studies in breast cancer. Proc Natl Acad Sci USA.1995;92: Marshal RD, Lester S, Corless C, et al. Expression of cycle-regulated proteins in prostate cancer. Cancer Res. 1996;56: Furihata M, Ishikawa T, Inoue A, et al. Determination of the prognostic significance of unscheduled cyclin A overexpression in patients with esophageal squamous cell carcinoma. Clin Cancer Res. 1996;2: Kushner J, Bradley G, Young B, Jordan RCK. Aberrant expression of cyclin A and B1 proteins in oral carcinoma. J Oral Pathol Med. 1999;88: Kamb A, Gruis NA, Weaver-Feldhaus J, et al. A cell cycle regulator potentially involved in the genesis of many tumor types. Science. 1994;264: Sinclair WK, Morton RA. X-ray sensitivity during the cell generation cycle of cultured Chinese hamster cells. Radiat Res. 1996;4: Arch Pathol Lab Med Vol 124, February 2000 Cyclin A and Cyclin B1 in Astrocytomas Allen et al