Int. J. Cancer: 70, 508–511 (1997) r 1997 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer RARE MUTATIONS AND NO HYPERMETHYLATION AT THE CDKN2A LOCUS IN EPITHELIAL OVARIAN TUMOURS Yang-Chia SHIH1, Judith KERR1, Jim LIU2, Terry HURST3, Soo-Keat KHOO3, Bruce WARD3, Brandon WAINWRIGHT4 and Georgia CHENEVIX-TRENCH1* 1The Queensland Institute of Medical Research, Brisbane, Australia 2Myriad Genetics Inc., Salt Lake City, UT, USA 3Department of Obstetrics and Gynaecology, The University of Queensland, Royal Brisbane Hospital, Herston, Brisbane, Australia 4Centre for Molecular and Cellular Biology, The University of Queensland, Brisbane, Australia The tumour-suppressor gene CDKN2A (p16, MTS1, CDK4I) encodes a cell cycle-regulatory protein and is located on chromosome 9p21, a region deleted in a wide variety of human cancers. To determine the role of the CDKN2A gene in the development of ovarian adenocarcinomas, we examined a large series of benign, low malignant potential (LMP) and invasive ovarian neoplasms for evidence of loss of heterozygosity (LOH), homozygous deletions, point mutations and hypermethylation of the CDKN2A locus. We have previously reported LOH on 9p in 45% of malignant ovarian neoplasms and a smaller percentage of benign and LMP tumours. In the current study, 6 malignant tumours were identified with partial deletions of 9p21. In 5 of these, the CDKN2A gene lays within the minimal deleted region. Homozygous deletions of CDKN2A were observed in only 2/88 invasive ovarian tumours and in 5/11 ovarian cancer cell lines. Of 15 primary ovarian tumours analyzed, one nonsense mutation was identified in a mucinous LMP tumour. No evidence of hypermethylation of the CDKN2A gene was found in 50 primary ovarian adenocarcinomas nor in 3 ovarian cancer cell lines. In conclusion, homozygous deletions, mutations and the de novo methylation of 58 CpG island are not frequent modes of inactivation of the CDKN2A gene in ovarian cancer. The target of 9p LOH in ovarian adenocarcinomas is therefore unknown. Int. J. Cancer, 508–511, 1997. r 1997 Wiley-Liss, Inc. Ovarian cancer is one of the major causes of death in women with gynaecological tumours. Early-stage ovarian cancer is hard to detect because of the absence of early warning symptoms, and the majority (60–70%) of patients present with advanced tumours, most of them dying of the disease. The molecular events associated with the development and subsequent progression and metastasis of epithelial ovarian tumours have not been well characterised. To gain insight into the molecular genetic changes of ovarian cancer, loss of heterozygosity (LOH) studies have been used to identify chromosomal regions containing putative tumour suppressor genes. Many studies of LOH in ovarian cancer have been reported at numerous loci (reviewed by Shelling et al., 1995). Frequent LOH on chromosome 9p has been observed in multiple tumour types, suggesting the presence of a tumour suppressor gene with a role in the tumorigenesis of a wide variety of tumours (Kishimoto et al., 1995). The involvement of chromosome 9 in ovarian adenocarcinoma was first reported by Cliby et al. (1993). The high frequency of LOH (54%) observed in this study was confirmed by others (Chenevix-Trench et al., 1994; Osborne and Leech, 1994a,b; Rodabaugh et al., 1995; Devlin et al., 1996). We have reported LOH on chromosome 9p in 45% of malignant ovarian adenocarcinomas, as well as in 1/20 benign and 4/25 low malignant potential (LMP) neoplasms (Chenevix-Trench et al., 1997). Further evidence that a tumour-suppressor gene in this region is inactivated in ovarian adenocarcinoma came from the finding of homozygous deletions in ovarian cancer cell lines between D9S171 and IFNA (Chenevix-Trench et al., 1994). The CDKN2A gene (also called CDKN2, MTS1 and CDK4I), which encodes an Mr 16,000 protein (p16), was identified as a candidate tumour-suppressor gene within this region (Kamb et al., 1994; Nobori et al., 1994). The p16 protein plays a central role in cell cycle control by binding to cyclin-dependent kinase 4 (CDK4) and preventing the association of CDK4 with cyclin D. This in turn prevents the subsequent phosphorylation of critical substrates necessary for transit through the G1 phase of the cell cycle (Serrano et al., 1993). If the CDKN2A gene is inactivated, there is no regulatory mechanism to block the cell cycle, and this may lead to uncontrolled growth. Analysis of CDKN2A homozygous deletions and mutations in primary ovarian adenocarcinomas and ovarian adenocarcinoma cell lines has been previously reported (Chenevix-Trench et al., 1994; Kamb et al., 1994; Campbell et al., 1995; Hatta et al., 1995; Rodabaugh et al., 1995; Schultz et al., 1995). Although some reports showed that 14–50% of ovarian cancer cell lines had homozygous deletions of this locus, very few somatic mutations have been found in more than 140 primary ovarian neoplasms that have been examined. In primary tumours, no point mutations have been found, while homozygous deletions have been reported in just 1 malignant (Chenevix-Trench et al., 1994) and 2 LMP (Rodabaugh et al., 1995; Devlin et al., 1996) tumours. One explanation for the high rate of LOH at 9p21 but the very low level of mutations in CDKN2A may be that expression of the gene is reduced by ‘‘silencing’’. Such transcriptional blocking of p16 by methylation of its 58 CpG island has been described in non-small cell lung and head-and-neck squamous-cell cancer cell lines (Merlo et al., 1995; Otterson et al., 1995). The same hypermethylation was also found in about 20% of different primary neoplasms but not in the corresponding normal tissues. This suggests that de novo methylation of the 58 CpG island of CDKN2A is a common pathway of p16 inactivation in human cancers, but ovarian adenocarcinomas have not yet been evaluated. To determine the role of the CDKN2A tumour-suppressor gene in the development of ovarian adenocarcinomas, we examined homozygous deletions and mutations of CDKN2A as well as hypermethylation of the locus in a large series of benign, LMP and invasive ovarian neoplasms. MATERIAL AND METHODS Samples Up to 135 ovarian tumours of epithelial origin, obtained from patients undergoing surgery, were examined for LOH, homozygous deletions, point mutations and hypermethylation. Of the 135 tumours, there were 82 serous, 22 mucinous, 8 clear cell, 7 endometrioid and 16 of mixed histology. The series included 21 benign, 29 LMP and 85 malignant tumours. All patients were staged at laparotomy in accordance with the recommendations of the International Federation of Gynaecology and Obstetrics (FIGO). Germline DNA was obtained in all cases from peripheral blood, except for 3 archival specimens for which archival stromal material was used. *Correspondence to: The Queensland Institute of Medical Research, RBH Post Office, Herston, QLD 4029, Australia. Fax: 61 7 3362 0105. Received 23 August 1996; revised 22 October 1996. CDKN2A IN OVARIAN NEOPLASMS Cell lines Eleven ovarian tumour cell lines were examined: CI-80-13S 27/87, PEO14, SKOV3, PEO1, PEO4, OAW53128, OAW42, JAM, LANDRY and COLO 316 (as listed in Chenevix-Trench et al., 1994). DNA isolation, microsatellite and mutation analyses Tumour tissue was prepared as described previously in ChenevixTrench et al. (1992). DNA from normal and tumour specimens was analysed at 7 polymorphic markers on chromosome 9p. The short tandem repeats (STRs) were D9S162, IFNA, D9S942, D9S1604, D9S171, D9S126 and D9S169. For STR analysis 10–50 ng genomic DNA was amplified in 10 µl reactions containing 200 µM dCTP, dGTP and dTTP; 25 µM dATP; 0.5 µCi a33P-dATP; 5 pmol each primer; 1.5 mM MgCl2; 1 3 Taq polymerase buffer and 1 U Taq. Samples were amplified with an initial denaturation of 3 min at 94°C followed by 25–30 cycles of 94°C for 45 sec, 55°C for 1 min, 72°C for 1 min and a final extension of 5 min at 72°C. Products were then heat-denatured and run on a 5% polyacrylamide/7 M urea-sequencing gel followed by visualisation by autoradiography analysis. Gels were analysed by 2 independent observers, and LOH was scored conservatively as a substantial reduction of intensity of 1 allele. In a comparison of 2 scoring methods, visual scoring was found to be compatible with densitometric analysis (Leggett et al., 1995). Homozygotes were declared ‘‘not informative’’. Sequencing of CDKN2A exons 1 and 2 was carried out as previously described (Liu et al., 1995). Southern analysis of the CDKN2A locus Southern analysis was carried out as previously described (Chenevix-Trench et al., 1994). EcoRI digests were hybridised with a radiolabelled CDKN2A cDNA probe to detect homozygous deletions, using a probe (p24a3VI) from the COL6A3 locus as a loading control. For the analysis of hypermethylation, DNAs were digested with EcoRI alone and in combination with the methylationsensitive enzymes SmaI, SacII and EagI (15 U/µg) for 16 hr. These 509 digests were hybridised with the PEI probe, which we amplified by PCR (Merlo et al., 1995). RESULTS Deletion mapping of chromosome 9p in malignant tumours Tumours from 135 patients were examined for LOH using 7 9p STR markers. Overall LOH frequencies have been reported before (Chenevix-Trench et al., 1996), and we present here only the deletion mapping data of 81 malignant tumours. Six malignant tumours were identified with partial deletions of 9p21, in which the smallest region of overlap was between D9S942 and D9S171 (Figs. 1, 2). These 6 tumours can be classified into 3 groups: group 1 (tumour 104) had LOH at and proximal to D9S171, group 2 (tumours 85/85, 25/90 and 76/90) had LOH distal to D9S171 but not including this locus and group 3 (tumours 13/89 and 64/92) had LOH between and including the D9S942 and D9S171 loci. Homozygous deletions of CDKN2A Eighty-eight primary ovarian neoplasms (including 4 benign, 15 LMP and 69 malignant tumours) and 11 cell lines were examined for homozygous deletion by Southern blot analysis using a probe from the CDKN2A locus. Homozygous deletions of CDKN2A were found in 5/11 cell lines (CI-80-13S, 27/87, SKOV3, PEO1 and PEO4), and in 2/88 primary tumours (27/87 and 77/93) (Fig. 3). Southern analysis with IFNB1 and D9S3 probes previously revealed homozygous deletions only in cell lines 27/87, PEO1 and PEO4 and in the primary tumour from which the 27/87 cell line was derived (Chenevix-Trench et al., 1994). Tumour 77/93 was a stage 1 mucinous LMP tumour, while 27/87 was a stage 3 endometrioid malignant tumour. Mutation analysis The CDKN2A gene was sequenced in 15 ovarian neoplasms (5 of which showed LOH of chromosome 9p) to detect exonic mutations. These tumours included 3 benign; 4 LMP; and 3 stage I, 4 stage III and 1 stage IV malignant tumours. Only 1 mutation was FIGURE 1 – Schematic representation of 6 patients with partial deletions on chromosome 9p. Patient numbers, stage and histology are indicated at the top of each column. Black box, LOH; white box, no LOH; patched box, not informative; UD, undetermined; CCC, clear cell carcinomas; SER, serous tumours. 510 SHIH ET AL. DISCUSSION FIGURE 2 – Representative results at the 3 9p STR markers examined. Patient numbers are indicated at the top of each column. N and T indicate normal germline and tumour DNA, respectively. Arrows indicate LOH. FIGURE 3 – Southern analysis showing homozygous deletion of CDKN2A locus in tumour sample 27/87. Tumour sample 18/92 is included to demonstrate a tumour without a CDKN2A deletion. Hybridisation of the same filters with a probe (p24a3VI) from the COL6A3 locus on chromosome 2 is shown as a loading control. identified, a non-sense mutation, G = T at nucleotide 175 (data not shown). This mutation occurred in a mucinous LMP tumour (11/92) with LOH of 9p. Hypermethylation of the CDKN2A locus Fifty ovarian adenocarcinomas (2 benign, 3 LMP and 45 malignant tumours) and 3 ovarian cell lines without homozygous deletions were examined for evidence of hypermethylation of the CDKN2A gene. No evidence of hypermethylation was detected with double digests of EcoRI/SmaI, EcoRI/SacII or EcoRI/EagI, and the expected banding pattern for unmethylated DNA was observed in all cases (data not shown). Previous LOH studies have implicated the short arm of chromosome 9 as a frequent target for allelic loss in a variety of tumour types (reviewed in Eiriksdottir et al., 1995). LOH on 9p has also been reported in ovarian adenocarcinoma (Chenevix-Trench et al., 1994; Campbell et al., 1995; Rodabaugh et al., 1995; Schultz et al., 1995; Devlin et al., 1996). The frequency of LOH on chromosome 9p in these reports varied from 14% to 70%, and the smallest regions of deletion have been described around marker D9S171 at 9p21-22, close to the CDKN2A gene. Our previous findings have shown 32% LOH in ovarian tumours on chromosome 9p and indicated that the smallest region of overlap was between IFNA and D9S171 (Chenevix-Trench et al., 1994). We have now identified 6 ovarian tumours with partial deletions on chromosome 9p which delineate the interval further (Figs. 1, 2). In 5 of these tumours, CDKN2A could be the target of the deletions. However, just 1 tumour (N104) had a deletion that excluded CDKN2A but instead could be targeting a more proximal gene, such as CDKN2B (p15). These findings support results showing that CDKN2A may not be the only target involved in ovarian tumorigenesis (Campbell et al., 1995; Rodabaugh et al., 1995; Schultz et al., 1995). Support for the hypothesis that CDKN2A is the target of 9p deletions in ovarian cancer comes from data showing that both homologues of the gene are inactivated by deletion or mutation in primary tumours. The CDKN2A gene has been reported to be homozygously deleted in various tumour cell lines, including those derived from ovarian cancers (Kamb et al., 1994). However, there may be a low frequency (3–14%) of homozygous deletions at the CDKN2A locus in primary ovarian tumours (Campbell et al., 1995; Schultz et al., 1995; Devlin et al., 1996), with a much higher frequency (20–50%) of homozygous deletions in cell lines (Chenevix-Trench et al., 1994; Schultz et al., 1995; Rodabaugh et al., 1995). We report here a similarly low rate of homozygous deletions in only 2/88 (2%) ovarian tumours, while 5/11 (45%) ovarian cell lines examined had homozygous deletions of CDKN2A. This confirms that homozygous deletion is not a major mechanism of CDKN2A gene inactivation in ovarian tumorigenesis. Point mutations of the CDKN2A gene have been reported in many different tumour cell lines (Serrano et al., 1993; Kamb et al., 1994), but for most tumour types, mutations are less frequent in primary tumours (Okajima et al., 1996). This is also the case in ovarian neoplasia, for which there are no previous reports of somatic point mutations of CDKN2A in primary tumours (Campbell et al., 1995; Hatta et al., 1995; Rodabaugh et al., 1995; Schultz et al., 1995). In our mutation analysis, one nonsense mutation was identified out of 15 ovarian tumours analysed. Interestingly, this mutation occurred in a mucinous LMP tumour. Combining all available data, there is a significantly higher rate of point mutations in LMP tumours (1/22) compared with malignant tumours (0/138) (2-sided p 5 0.037, Fisher’s exact test). Furthermore, homozygous deletions have been reported more often in LMP tumours (3/35) than in malignant ovarian adenocarcinomas (2/177) (2-sided p 5 0.018, Fisher’s exact test) (Campbell et al., 1995; Devlin et al., 1994; Rodabaugh et al., 1995). This may indicate that inactivation of CDKN2A is slightly more common in LMP tumours, as has been reported for K-ras mutations (Teneriello et al., 1993), but deletions and point mutations are clearly rare mechanisms of CDKN2A inactivation in epithelial ovarian neoplasms. Another possible explanation for the high rate of LOH at 9p21 around the CDKN2A locus (in the absence of second mutations) is that expression of p16 is reduced by hypermethylation, or ‘‘gene silencing’’. This pathway of inactivation, which involves reduced transcription associated with de novo methylation of a 58 CpG island of CDKN2A, has been observed frequently (39–92%) in several cancer cell lines and in breast (31%), colon (42%), non-small cell lung cancer (53%) and glioma (28%) primary tumours (Herman et al., 1995, 1996). However, we have found no evidence of hypermethylation of the CDKN2A gene in 50 primary CDKN2A IN OVARIAN NEOPLASMS varian tumours (including 2 benign, 3 LMP and 45 malignant tumours) and 3 ovarian cancer cell lines without homozygous deletion. This indicates that the de novo methylation of the 58 CpG island is not a frequent mode of inactivation of the CDKN2A gene in ovarian cancer. In conclusion, our current results and those reported previously by others show a high frequency of LOH on 9p in ovarian carcinomas, which usually includes the CDKN2A locus. However, there is little evidence of homozygous inactivation of the CDKN2A gene by deletion, mutation or silencing. It would appear from monochromosomal transfer experiments that a single copy of chromosome 9 is sufficient to induce growth arrest; therefore, haplo-insufficiency (resulting from LOH and without a ‘‘second 511 hit’’) would not be sufficient to promote a growth advantage for the cell (England et al., 1996). It therefore remains a conundrum as to whether CDKN2A is indeed the target of 9p LOH or whether other closely linked tumour suppressor genes (perhaps including CDKN2B) may be involved. ACKNOWLEDGEMENTS We thank Dr M. Friedlander and Ms J. Leary for contributing some tumour samples; Drs A. Wilson, J.F. Smyth, T. Bradley and I. Bertoncello for the cell lines; and Dr D. Beach for the CDKN2A cDNA probe. This work was supported by the Queensland Cancer Fund and the National Health and Medical Research Council. REFERENCES CAMPBELL, I.G., BEYNON, G., DAVIS, M. and ENGLEFIELD, P., LOH and mutation analysis of CDKN2A in primary human ovarian cancers. Int. J. Cancer, 63, 222–225 (1995). CHENEVIX-TRENCH, G., KERR, J., FRIEDLANDER, M., HURST, T., SANDERSON, B., COGLAN, M., WARD, B., LEARY, J. and KHOO, S.K., Homozygous deletions on the short arm of chromosome 9 in ovarian adenocarcinoma cell lines and loss of heterozygosity in sporadic tumours. Amer. J. hum. Genet., 55, 143–149 (1994). CHENEVIX-TRENCH, G. and 19 OTHERS, Analysis of loss of heterozygosity and K-ras mutations in ovarian neoplasms: clinicopathological correlations. Genes Chromosomes Cancer (1997) (In press). CHENEVIX-TRENCH, G., LEARY, J., KERR, J., MICHEL, J., KEFFORD, R., HURST, T., PARSONS, P.G., FRIEDLANDER, M. and KHOO, S.K., Frequent loss of heterozygosity on chromosome 18 in ovarian adenocarcinoma which does not always include the DCC locus. Oncogene, 7, 1059–1065 (1992). CLIBY, W., RITLAND, S., HARTMANN, L., DOBSON, M., HALLING, K.C., KEENEY, G., PODRATZ, K.C. and JENKINS, R.B., Human epithelial ovarian cancer allelotype. Cancer Res., 53, 2393–2398 (1993). DEVLIN, J., ELDER, P.A., GABRA, H., STEEL, C.M. and KNOWLES, M.A., High frequency of chromosome 9 deletion in ovarian cancer: evidence for three tumour-suppressor loci. Brit. J. Cancer, 73, 420–423 (1996). DEVLIN, J., KEEN, A.J. and KNOWLES, M.A., Homozygous deletion mapping at 9p21 in bladder carcinoma defines a critical region within 2cM of IFNA. Oncogene, 9, 2757–2760 (1994). EIRIKSDOTTIR, G., SIGURDSSON, A., JONASSON, J.G., AGNARSSON, B.A., SIGURDSSON, H., GUDMMUNDSSON, J.T., BERGTHJORSSON, J.T., BARKARDOTTIR, R.B., EGILSSON, V. and INGVARSSON, S., Loss of heterozygosity on chromosome 9 in human breast cancer: association with clinical variables and genetic changes at other chromosome regions. Int. J. Cancer, 64, 378–382 (1995). ENGLAND, N.L., CUTHBERT, A.P., TROTT, D.A., JEZZARD, S., NOBORI, T., CARSON, D.A. and NEWBOLD, R.F., Identification of human tumour suppressor genes by monochromosome transfer: rapid growth-arrest response mapped to 9p21 is mediated solely by the cyclin-D-dependent kinase inhibitor gene, CDKN2A (p16INK4A). Carcinogenesis (1996) (In press). HATTA, Y., HIRAMA, T., TAKEUCHI, S., LEE, E., PHAM, E., MILLER, C.W., STROHMEYER, T., WILCZYNSKI, S.P., MELMED, S. and KOEFFLER, H.P.,Alterations of the p16 (MTS1) gene in testicular, ovarian and endometrial malignancies. J. Urology, 154, 1954–1957 (1995). HERMAN, J.G., JEN, J., MERLO, A. and BAYLIN, S.B., Hypermethylation associated inactivation indicates a tumor suppressor role for p15INK4B1. Cancer Res., 56, 722–727 (1996). HERMAN, J.G., MERLO, A., MAO, L., LAPIDUS, R.G., OSSA, J.P.J., DAVIDSON, N.E., SIDRANSKY, D. and BAYLIN, S.B., Inactivation of the CDKN2/p16/ MTS1 gene is frequently associated with aberrant DNA methylation in all common human cancers. Cancer Res., 55, 4525–4530 (1995). KAMB, A., CRUIS, N.A., WEAVER-FELDHAUS, J., LUI, Q., HARSHMAN, K., TAVTIGIAN, S.V., STOCKERT, E., DAY III, R.S., JOHNSON, B.E. and SKOLNICK, M.H., A cell cycle regulator potentially involved in genesis of many tumour types. Science, 264, 436–440 (1994). KISHIMOTO, Y., SUGIO, K., HUNG, J.Y., VIRMANI, A.K., MCLNTIRE, D.D., MINNA, J.D. and GAZDAR, A.F., Allele-specific loss in chromosome 9p loci in preneoplastic lesions accompanying non-small-cell lung cancers. J. nat. Cancer Inst., 87, 1224–1229 (1995). LEGGETT, B.A., YOUNG, J.P., BUTTENSHAW, R., THOMAS, L.R., YOUNG, B., CHENEVIX-TRENCH, G., SEARLE, J. and WARD, M., Colorectal carcinomas show frequent allelic loss on the long arm of chromosome 17 with evidence for a specific target region. Brit. J. Cancer, 71, 1070–1073 (1995). LIU, Q., NEUHAUSEN, S., MCCLURE, M., FRYE, C., WEAVER-FELDHAUS, J., GRUIS, N.A., EDDINGTON, K., ALLALUNIS, M.J., SKOLNICK, H.M., FUJIMURA, F.K. and KAMB, A., CDKN2A (MTS1) tumour suppressor gene mutations in human tumor cell lines. Oncogene, 10, 1061–1067 (1995). MERLO, A., HERMAN, J.G., MAO, L., LEE, D.J., GABRIELSON, E., BURGER, P.C., BAYLIN, S.B. and SIDRANSKY, D., 58CpG island methylation is associated with transcriptional silencing of the tumour suppressor p16/ CDKN2/MTS1 in human cancers. Nature (Med.), 1, 686–692 (1995). NOBORI, T., MIURA, K., WU, D.J., LOIS, A., TAKABAYASHI, K. and CARSON, D.A., Deletions of the cyclin-dependent kinase-4 inhibitor gene in multiple human cancers. Nature (Lond.), 368, 753–756 (1994). OKAJIMA, E., FUKUDA, T., OKITA, S., TSUTSUMI, M., HIRAO, Y., OKAJIMA, E. and KONISHI, Y., Infrequent somatic alteration of p16/MTS1 in human primary superficial bladder cancers. Cancer Lett., 203, 227–231 (1996). OSBORNE, R.J. and LEECH, V., Polymerase chain reaction allelotyping of human ovarian cancer. Brit. J. Cancer, 69, 429–438 (1994a). OSBORNE, R.J. and LEECH, V., Chromosome 9q deletion mapping in epithelial ovarian cancer. Proc. Amer. Assoc. Cancer Res., 35, 601 (1994b). OTTERSON, G.A., KHLEIF, S.N., CHEN, W., COXON, A.B. and KAYE, F.J., CDKN2 gene silencing in lung cancer by DNA hypermethylation and kinetics of p16INK4 protein induction by 5-aza 28deoxycytidine. Oncogene, 11, 1211–1216 (1995). RODABAUGH, K.J., BIGGS, R.B., QURESHI, J.A., BARRETT, A.J., WILLIAM, R.W., BELL, D.A., BERKOWITZ, R.S. and MOK, S., Detailed deletion mapping of chromosome 9p and p16 gene alterations in human borderline and invasive epithelial ovarian tumours. Oncogene, 11, 1249–1254 (1995). SCHULTZ, D.C., VANDERVEER, L., BUETOW, K.H., BOENTE, M.P., OZOLS, R.F., HAMILTON, T.C. and GODWIN, A.K., Characterization of chromosome 9 in human ovarian neoplasia identifies frequent genetic imbalance on 9q and rare alterations involving 9p, including CDKN2. Cancer Res., 55, 2150–2157 (1995). SERRANO, M., HANNON, G.J. and BEACH, D., A new regulatory motif in cell-cycle control causing specific inhibition of cyclin D/CDK4. Nature (Lond.), 366, 704–707 (1993). SHELLING, A.N., COOKE, I.E. and GANESAN, T.S., The genetic analysis of ovarian cancer. Brit. J. Cancer, 72, 521–527 (1995). TENERIELLO, M.G., EBINA, M., LINNOILA, R.I., HENRY, M., NASH, J.D., PARK, R.C. and BIRRER, M.J., p53 and Ki-ras gene mutations in epithelial ovarian neoplasms. Cancer Res., 53, 3103–3108 (1993).