Differences inp16Gene Methylation and Expression in Benign and Malignant Ovarian Tumors

Gynecologic Oncology 72, 87–92 (1999) Article ID gyno.1998.5235, available online at http://www.idealibrary.com on

Differences in p16 Gene Methylation and Expression in Benign and Malignant Ovarian Tumors Lisa L. McCluskey,* Chen Chen,† Erika Delgadillo,† Juan C. Felix,† Laila I. Muderspach,* and Louis Dubeau† *Division of Gynecologic Oncology, and †Department of Pathology, USC/Norris Comprehensive Cancer Center, University of Southern California School of Medicine, Los Angeles, California 90033 Received January 9, 1998

is usually associated with gene silencing, but may also be associated with increased gene expression. Patterns of DNA methylation are tissue-specific and heritable [3], suggesting a role in maintenance of differentiation. The extensive changes in global DNA methylation usually present in cancer cells [4, 5] also suggest a role in the development and perhaps maintenance of the malignant phenotype. There is indeed strong support for the idea that genomic DNA alterations may provide means of either activating cellular protooncogenes [6 –12] or silencing tumor suppressor genes [13– 15]. Evidence for methylation changes in benign neoplasms as well as in tumors with reduced invasive and metastatic abilities suggests that DNA methylation alterations may play a role early in the neoplastic transformation process [16, 17]. We recently examined and compared global levels of DNA methylation in ovarian cystadenomas, ovarian tumors of low malignant potential (LMP), and ovarian carcinomas in order to gather insights into the importance of DNA methylation changes in this tumor model [17]. We concluded that the state of methylation of genomic DNA was significantly lower in carcinomas and LMP tumors compared to cystadenomas [17]. Although these studies suggested that changes in DNA methylation may be important in the development of ovarian LMP tumors and carcinomas, the nature of the specific genes or chromosomal regions targeted by such changes during ovarian tumorigenesis remains unknown. We sought to examine and compare the role of DNA methylation changes in silencing the p16 tumor suppressor gene in ovarian cystadenomas, LMP tumors, and carcinomas in the present study. This gene is a cyclin-dependent kinase inhibitor which controls entry into the G1 phase of the cell cycle by preventing phosphorylation of the retinoblastoma gene product [18]. Defects in this gene are thought to be associated with the development of a large number of different tumor types [18] and have been observed in ovarian carcinoma cell lines. We examined the state of methylation of specific CpG dinucleotides near the transcription start site of this gene in each of these ovarian tumor subtypes and determined if a correlation existed between such changes and p16 gene expression. Our

Objective. The aim of this study was to test the hypothesis that DNA methylation is important for silencing the p16 tumor suppressor gene in ovarian epithelial tumors and to compare the prevalence of this mechanism among different ovarian epithelial tumor subtypes. Method. Methylation-specific PCR was used to analyze the p16 gene for DNA methylation in 20 ovarian cystadenomas, 15 low malignant potential (LMP) tumors, and 37 carcinomas. p16 expression was determined immunohistochemically in 58 of these tumors (16 cystadenomas, 13 LMP tumors, 29 carcinomas). Differences in methylation or expression rates between specific tumor subgroups were examined by Fisher’s exact test. Results. Fragments from the distal promoter and beginning of the first exon of the p16 gene were both methylated in 5 of 15 (33%) LMP tumors compared to 2 of 37 (5%) carcinomas (P 5 0.02). Those sites were also methylated in 5 of 20 (25%) cystadenomas. Lack of p16 expression was present in 7 of 16 cystadenomas, 4 of 13 LMP tumors, and 22 of 29 carcinomas (P [LMPs versus carcinomas] 5 0.01) and correlated with methylation changes in LMP tumors (P 5 0.05). p16 expression was correlated with mucinous differentiation in cystadenomas (P 5 0.001). Conclusion. p16 silencing may be important for the development of ovarian carcinomas and a subset of LMP tumors. Changes in DNA methylation may be more important for inactivation of this gene (and perhaps other tumor suppressor genes) in LMP tumors, which lack many of the alternative mechanisms present in carcinomas. p16 expression is primarily related to mucinous differentiation in cystadenomas. © 1999 Academic Press Key Words: DNA methylation; p16; ovarian cancer.

INTRODUCTION DNA methylation is an epigenetic alteration thought to be important for the control of gene expression [1, 2]. These changes usually occur in regions rich in cytosine– guanine dinucleotides, called CpG islands, which are often associated with promoter regions. Increased methylation of such regions Presented at the 29th Annual Meeting of the Society of Gynecologic Oncologists, Orlando, FL, February 7–11, 1998. 87

0090-8258/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

88

MCCLUSKEY ET AL.

aim was to test the hypothesis that methylation of those promoter sequences, which were previously reported [19] to correlate with decreased p16 expression in other tumor models, could represent an important mechanism for silencing this tumor suppressor in ovarian epithelial neoplasms. p16 mutations are rare in ovarian tumors [20, 21] and loss of heterozygosity, which is another important mechanism of tumor suppressor gene inactivation in a variety of tumor types, affects the p16 locus only in about 30% of ovarian carcinomas and rare in cystadenomas and LMP tumors [22]. We reasoned that DNA methylation changes could provide an alternative mechanism for p16 inactivation in tumors with no loss of heterozygosity at this locus. Evidence that such changes are indeed instrumental in turning off p16 expression was previously reported in bladder, breast, prostate, colon, and hematologic cancers [13, 14, 19, 23]. MATERIALS AND METHODS Source and Initial Processing of Tissue Specimens Tumor specimens were obtained fresh from the operating rooms of either the USC/Norris Comprehensive Cancer Center or Women’s Hospital of the Los Angeles County Medical Center and stored frozen at 280°C. All tissues were obtained in compliance with the rules of the Institutional Tissue Committee at the University of Southern California and after approval was obtained from the committee. Diagnostic verification and evaluation of the amount of stromal contamination for each case was done by one of us (L.D.), a surgical pathologist familiar with ovarian tumor histopathology. The criteria used for histological grading of carcinomas were described [22]. A total of 72 tumors (20 ovarian cystadenomas, 15 LMP tumors, and 37 carcinomas) were examined for changes in DNA methylation. Fifty-eight of these tumors (16 cystadenomas, 13 LMP tumors, and 29 carcinomas) were also examined for p16 gene expression because archival tissue blocks were no longer available from the remaining 15 cases. Sample Processing for DNA Methylation Studies and Methylation-Specific PCR We used the methylation-specific PCR technique first developed by Herman et al. [24] for evaluation of DNA methylation in specific portions of exon 1 of the p16 gene. Deamination of unmethylated cytosine residues in genomic DNA was achieved first by heating the DNA samples (2 mg of genomic DNA dissolved in 20 ml of water) at 95°C for 20 min. After addition of 5 ml of fresh 3 M NaOH, the samples were further incubated for 20 min at 45°C. Twelve microliters of 0.1 M hydroquinone and 208 ml 3.6 M sodium bisulfite were added and each sample was overlaid with oil and incubated overnight at 55°C for 16 h without being exposed to light. The deaminated DNA was purified using a Wizard clean-up kit (Promega, Madison, WI, cat. No. A-7500) and the resulting 50-ml samples were desulfonated with 5 ml of 3 M NaOH at 40°C for 15 min, precipi-

tated in ethanol, and resuspended in 40 ml of water. This procedure yielded approximately 50 ng/ml of bisulfite-treated DNA. After an initial denaturation step at 95°C for 5 min, PCR was carried out for 35 cycles using the following conditions: 45 s at 95°C, 45 s at annealing temperature (1°C below the lowest primer melting temperature), and 45 s at 72°C. Primer sequences were described by Herman et al. [24]. Immunohistochemical Staining for p16 Formalin-fixed, paraffin-embedded archival tissue sections (4 mm in thickness) were mounted on glass slides, rehydrated, treated with 0.3% hydrogen peroxide, and heated in TrisCl (pH 1.0) for 6 min in a microwave oven purchased from Sharp (Mahwah, NJ, model No. R-4A4B6). Sections were stained using a primary mouse monoclonal antibody directed against the p16 protein purchased from Pharmingen (San Diego, CA, cat. No. 13251A). The Elit avidin– biotin complex kit (Vector Laboratories, Burlingame, CA) was used for antibody detection. Sections processed in parallel but for which the primary antibody was replaced by normal mouse serum were used as controls. Cases were scored as positive for p16 expression if a group of at least 10 adjacent tumor cells or if 5% or more of the total tumor cell population showed discrete, homogeneous nuclear staining with the anti-p16 antibody. Loss of Heterozygosity Determinations All carcinomas and most LMP tumors used in this study had been previously examined for loss of heterozygosity in chromosome 9p in a previous study [22]. About 30% of the carcinomas showed losses of heterozygosity in this chromosome whereas none of the LMP tumors examined showed such losses. Statistical Analyses Differences in either expression or methylation of the p16 gene in various tumor subgroups were compared using Fisher’s exact test [25]. All P values quoted are two-sided. RESULTS Methylation of the p16 Gene in Ovarian Cystadenomas, LMP Tumors, and Carcinomas We used an approach developed by Herman et al. [24] to examine the state of methylation of the p16 gene in ovarian tumors. This approach, called methylation-specific PCR, is based on the fact that treatment of genomic DNA with bisulfite ions deaminates unmethylated cytosine residues to uracil, leaving methylated residues unaltered. A methylation-specific PCR can therefore be achieved by designing primers complementary to potentially methylated genomic regions near their 39 ends and engineering their sequences to complement either the

89

p16 METHYLATION AND EXPRESSION IN OVARIAN TUMORS

FIG. 1. Diagram of the p16 gene fragments examined for changes in DNA methylation.

deaminated or undeaminated forms. We used a set of primers for enzymatic amplification of two potentially methylated regions of the p16 gene, the methylation status of which was previously shown to correlate with p16 gene expression [24]. A diagram of these two regions, called respectively M 1 and M 2, is shown in Fig. 1. M 1 corresponds to a 140-base-pair region in the promoter near the transcription start site whereas M 2 is a 224-base-pair sequence that overlaps with the entire M 1 region but also contains an additional 84-base-pair fragment from the beginning of exon 1. We used the above primers to examine the methylation status of the M 1 and M 2 fragments of the p16 gene in bisulfitetreated genomic DNA from 72 different ovarian neoplasms which included 20 cystadenomas, 15 tumors of low malignant potential, and 37 carcinomas. Representative results are shown in Fig. 2 for the M 2 fragment. The top portion shows PCR products obtained with methylation-specific primers (M 2) whereas the bottom portion shows results obtained from the same tumors using primers complementary to the deaminated (unmethylated) sequence (U 2). The arrows indicate the position of the expected 224-base-pair product for each primer set. The remaining bands seen in the figure correspond to nonspecific PCR products and are unrelated to methylation status. A detectable signal corresponding to the expected 224-base-pair fragment was obtained with cases 1, 2, and 3 when the M 2 primers were used (Fig. 2). Cases 1 and 3 showed no detectable signal at the corresponding position when the U 2 primers were used whereas a weak band was seen with case 2 (Fig. 2). We conclude that both alleles of the M 2/U 2 sequence were methylated in cases 1 and 3 while at least one allele was methylated in case 2. Cases 3–7 showed signals only with the U 2 primers and therefore contained no detectable DNA methylation in this fragment (Fig. 2). The last two lanes shown in Fig. 2 are for positive and negative controls, respectively. The rates of methylation of the M 1/U 1 and M 2/U 2 sequences in our 20 ovarian cystadenomas, 15 tumors of low malignant potential, and 37 carcinomas are summarized in Table 1. Methylation of both sequences was generally lower in carcinomas compared to either cystadenomas or LMP tumors. This difference was most marked for the M 2/U 2 sequence, which is distinguished from the M 1/U 1 sequence by an additional 84 base pairs in the translated region of the gene (Fig. 1). The difference in frequencies of DNA methylation of this sequence between tumors of low malignant potential (5 of 15 cases) and carcinomas (2 of 37 cases) was statistically significant (two-

FIG. 2. The M 2/U 2 region of the p16 gene was amplified enzymatically from bisulfite-treated genomic DNA obtained from seven different ovarian epithelial tumors as well as from the EJ bladder carcinoma cell line (positive control). A sample without genomic DNA was also included as negative control. Primers specific for either methylated (M 2) or unmethylated (U 2) DNA were used. The PCR products were electrophoresed on 3.5% agarose and visualized under ultraviolet light.

sided P 5 0.02). Both of the carcinomas that showed methylation in this sequence were of low histological grade (not shown). Neither of those two carcinomas showed loss of heterozygosity at the p16 locus (not shown). All ovarian tumors examined that showed evidence of methylation using the M 2 TABLE 1 P16 Methylation in Ovarian Epithelial Tumors Tumors with methylated M 1 fragment

Cystadenomas LMPs Carcinomas

Tumors with methylated M 2 fragment

N

%

N

%

14/20 11/15 21/37

70 73 57

5/20 5/15 2/37

25 37 5

Note. P [M2 methylation in LMP tumors versus carcinomas] 5 0.05.

90

MCCLUSKEY ET AL.

FIG. 3. Immunohistochemical demonstration of p16 gene expression. Histological sections stained with anti-p16 antibody are shown for mucinous (a) and serous (b) LMP tumors. The discrete nuclear staining attests to the presence of p16 gene expression in each case.

primers also showed methylation with the M 1 primers (not shown). p16 Gene Expression in Ovarian Cystadenomas, LMP Tumors, and Carcinomas We next examined the state of expression of the p16 gene in ovarian epithelial tumors in order to determine if methylation changes in the promoter region of this gene were associated with changes in gene expression. This was evaluated immunohistochemically using an anti-p16 monoclonal antibody. An example of p16 expression demonstrated by this approach is shown in Fig. 3. The discrete nuclear staining present in the two tumor samples shown in the figure is interpreted as indicative of p16 expression. A few of the ovarian tumors examined for DNA methylation at the p16 locus could not be examined for p16 expression because histological sections of these tumors were no longer available at our institution. Results of p16 immunoreactivity in the remaining 29 carcinomas, 13 LMP tumors, and 16 cysta-

denomas are shown in Table 2, together with the corresponding M 2/U 2 methylation data and tumor histological subtypes. Only 7 of the 29 (24%) carcinomas examined by immunohistochemistry showed detectable expression of this gene compared to 9 of 13 (69%) LMP tumors (P 5 0.01). These differences suggest that silencing of the p16 gene is comparatively more important in the development of carcinomas than LMP tumors. However, p16 was not expressed in all LMP tumors, as 4 such tumors showed no detectable expression of this tumor suppressor. DNA methylation may have played an important role in silencing of this gene in these 4 LMP tumors because a statistically significant correlation between M 2/U 2 methylation and gene expression was seen with these tumors in contrast to carcinomas (Table 2). Expression of the p16 gene in cystadenomas was not related to the state of p16 gene expression (Table 2), but was strongly correlated with the mucinous phenotype (P 5 0.001). All mucinous cystadenomas showed detectable p16 gene expression whereas only a single serous cystadenoma showed such

TABLE 2 Correlation between Expression and Methylation Status of the p16 Gene in Ovarian Epithelial Tumors Expression1 methylation1

Expression1 methylation2

Expression2 methylation1

Expression2 methylation2

Expression versus methylation

Serous carcinomas Mucinous carcinomas Endometrioid carcinomas Clear cell carcinoma Total carcinomas

0 0 0 0 0

2 0 5 0 7

1 1 0 0 2

11 0 8 1 20

P 5 1.00

Serous LMPs Mucinous LMPs Total LMPs

0 1 1

6 2 8

3 0 3

1 0 1

P 5 0.05

Serous cystadenomas Mucinous cystadenomas Total cystadenomas

0 3 3

1 5 6

0 0 0

7 0 7

P 5 0.21

p16 METHYLATION AND EXPRESSION IN OVARIAN TUMORS

expression. A similar correlation between mucinous differentiation and methylation was not observed in either LMP tumors or carcinomas. DISCUSSION Inactivation of the p16 gene is thought to contribute to the development of several tumor types [18] and has been associated with biologically more aggressive cancers [26]. The results of our experiments clearly show that this tumor suppressor is silenced in the majority of ovarian carcinomas. Although it was expressed in the majority of our ovarian LMP tumors, some tumors belonging to this category did not express this gene. This raises the possibility that p16 inactivation may be important for the development of a subset of LMP tumors. All mucinous cystadenomas examined expressed the p16 gene, whereas all serous cystadenomas except one showed lack of p16 expression. The status of p16 expression in cystadenomas may therefore be primarily determined by their state of differentiation and, in contrast to LMP tumors and carcinomas, may have little to do with neoplastic development. Our results also show that sequences near the translation start site of the p16 gene are methylated in a significant proportion of ovarian epithelial tumors. Substantial differences in rates of methylation were seen between the M 1/U 1 and M 2/U 2 sequences. Although methylation of the former sequence was seen in 64% (46 of 72) of all tumors examined, only 17% (12 of 72) of the same tumors showed methylation of the latter. None of the cases examined showed methylation of the M 2 sequence exclusively. This raises the possibility that methylation signals seen with the M 2/U 2 primers reflected changes at the site of the sense primer because this primer was the same for both the M 1/U 1 and the M 2/U 2 fragments. The proportion of carcinomas showing methylation changes in both the M 1/U 1 and the M 2/U 2 sequences in our tumor population was small, in support of recent studies from other laboratories [20, 27]. This suggests that mechanisms other than p16 methylation may be responsible for p16 gene silencing in the majority of carcinomas. However, neither of the two carcinomas with methylation of the M 2/U 2 sequence showed p16 expression, suggesting that methylation changes may have contributed to p16 inactivation in these two tumors. In support of this conclusion, different authors have previously shown that methylation of these sequences correlate with decreased p16 gene expression in other tumor types [13, 14, 19, 23]. A better correlation was observed between the M 2/U 2 methylation status and p16 expression in LMP tumors, suggesting that methylation changes may be a more important mechanism of p16 silencing in this group of tumors. This may point to important differences between the mechanisms of development of LMP tumors compared to carcinomas. Many different abnormalities, including loss of heterozygosity, homozygous deletions, and others can contribute to p16 silencing in carcinomas in addition to DNA methylation alterations. The finding of

91

a poor correlation between absence of p16 expression and p16 methylation in these tumors is not surprising given the availability of so many alternative mechanisms. In contrast, most of these mechanisms are not present in LMP tumors [28], leaving methylation changes as perhaps one of the few means of silencing the p16 gene (and other tumor suppressor genes) in these tumors. The fact that the only carcinomas found to be methylated with our M 2 primers were of low histological grade, and that such tumors are usually associated with fewer losses of heterozygosity based on previous studies [17, 29], is also consistent with the hypothesis that changes in DNA methylation are a comparatively more important mechanism of tumor suppressor gene inactivation in tumors with otherwise no or few losses of heterozygosity. This also suggests that agents with mechanisms of action which interfere with DNA methylation may be more effective in the treatment of a subset of LMP tumors or low grade carcinomas than the more aggressive ovarian epithelial tumors. ACKNOWLEDGMENTS This work was supported by NIH Grant RO1 CA51167 to L.D. and by pilot funds from Grant 5P30 CA14089-22 available to L.I.M. Lisa McCluskey was supported by the gynecologic oncology fellowship program at the University of Southern California. We thank Dr. Richard Cote and Mrs. Lilian Young for their help with immunohistochemical staining procedures. We thank Dr. Peter Laird for his helpful suggestions and comments.

REFERENCES 1. Ehrlich M, Wang RY: 5-Methylcytosine in eukaryotic DNA. Science 212:1350 –1357, 1981 2. Keshet I, Yisraeli J, Cedar H: Effect of regional DNA methylation on gene expression. Proc Natl Acad Sci USA 82:2560 –2564, 1985 3. Bird AP: CpG-rich islands and the function of DNA methylation. Nature 321:209 –213, 1986 4. Laird PW, Jaenisch R: DNA methylation and cancer. Hum Mol Genet 3:1487–1495, 1994 5. Spruck CH, Rideout WM, Jones PA: DNA Methylation: Molecular Biology and Biological Significance. Basel, Birkhauser Verlag, 1993 6. Munzel PA, Pfohl LA, Rohrdanz E, Keith G, Dirheimer G, Bock KW: Site-specific hypomethylation of c-myc protooncogene in liver nodules and inhibition of DNA methylation by N-nitrosomorpholine. Biochem Pharmacol 42:365–371, 1991 7. Sharrard RM, Royds JA, Rogers S, Shorthouse AJ: Patterns of methylation of the c-myc gene in human colorectal cancer progression. Br J Cancer 65:667– 672, 1992 8. Bhave MR, Wilson MJ, Poirier LA: c-H-ras and c-K-ras gene hypomethylation in the livers and hepatomas of rats fed methyl-deficient, amino acid-defined diets. Carcinogenesis 9:343–348, 1988 9. Ray JS, Harbison ML, McClain RM, Goodman JI: Alterations in the methylation status and expression of the raf oncogene in phenobarbitalinduced and spontaneous B6C3F1 mouse live tumors. Mol Carcinog 9:155–166, 1994 10. Hanada M, Delia D, Aiello A, Stadtmauer E, Reed JC: Bcl-2 gene hypomethylation and high-level expression in B-cell chronic lymphocytic leukemia. Blood 82:1820 –1828, 1993 11. Lipsanen V, Leinonen P, Alhonen L, Janne J: Hypomethylation of orni-

92

12.

13.

14.

15.

16.

17.

18.

19.

20.

MCCLUSKEY ET AL. thine decarboxylase gene and erb-A1 oncogene in human chronic lymphatic leukemia. Blood 72:2042–2044, 1988 Felgner J, Kreipe H, Heidorn K, Jaquet K, Zschunke F, Radzun HJ, Parwaresch MR: Increased methylation of the c-fms protooncogene in acute myelomonocytic leukemias. Pathobiology 59:293–298, 1991 Merlo A, Herman JG, Mao L, Lee DJ, Gabrielson E, Burger P, Baylin SB, Sidransky D: 59 CpG island methylation is associated with transcriptional silencing of the tumour suppressor p16/CDKN2/MTS1 in human cancers. Nature Med 1:686 – 692, 1995 Gonzalez-Zulueta M, Bender CM, Yang AS, Nguyen TT, Beart RW, Van Tornout JM, Jones PA: Methylation of the 59 CpG island of the p16/ CDKN2 tumor suppressor gene in normal and transformed human tissues correlates with gene silencing. Cancer Res 55:4531– 4535, 1995 Herman JG, Latif F, Weng Y, Lerman MI, Zbar B, Liu S, Samid D, Duan DS, Gnarra JR, Linehan WM: Silencing of the VHL tumor-suppressor gene by DNA methylation in renal carcinoma. Proc Natl Acad Sci USA 91:9700 –9704, 1994 Goelz SE, Vogelstein B, Hamilton SR, Feinberg AP: Hypomethylation of DNA from benign and malignant human colon neoplasms. Science 228: 187–190, 1985 Cheng P, Schmutte C, Cofer KF, Felix JC, Yu MC, Dubeau L: Alterations in DNA methylation are early, but not initial, events in ovarian tumorigenesis. Br J Cancer 75:396 – 402, 1997 Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman K, Tavtigian SV, Stockert E, Day RS, Johnson BE, Skolnick MH: A cell cycle regulator potentially involved in the genesis of many tumor types. Science 264: 436 – 440, 1994 Herman JG, Merlo A, Mao L, Lapidus RG, Issa JPL, Davidson NE, Sidransky D, Baylin SB: 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 Shih YC, Kerr J, Liu J, Hurst T, Khoo SK, Ward B, Wainwright B,

21.

22.

23. 24.

25. 26.

27.

28. 29.

Chenevix-Trench G: Rare mutations and no hypermethylation at the CDKN2A locus in epithelial ovarian tumours. Int J Cancer 70:508 –511, 1997 Shigemasa K, Hu C, West CM, Clarke J, Parham GP, Parmley TH, Korourian S, Baker VV, O’Brien TJ: p16 overexpression: a potential early indicator of transformation in ovarian carcinoma. J Soc Gynecol Invest 4:95–102, 1997 Cheng PC, Gosewehr JA, Kim TM, Velicescu MMW, Zheng J, Felix JC, Cofer KF, Luo P, Biela BH, Godorov G, Dubeau L: Potential role of the inactivated X chromosome in ovarian epithelial tumor development. J Natl Cancer Inst 88:510 –518, 1996 Jarrard DF, Bova GS, Isaacs WB: DNA methylation, molecular genetic and linkage studies in prostate cancer. Prostate Suppl 6:36 – 44, 1996 Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB: Methylationspecific PCR: a novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci USA 93:9821–9826, 1996 Dixon WJ, Massey FJ: Introduction to Statistical Analysis. New York, McGraw-Hill, 1969 Reed JA, Loganzo F, Shea C, Walker GJ, Flores JF, Glendening JM, Bogdany JK, Shiel MJ, Haluska MG, Fountain JW, Albino AP: Loss of expression of the p16/cyclin-dependent kinase inhibitor 2 tumor suppressor gene in melanocytic lesions correlates with invasive stage of tumor progression. Cancer Res 55:2713–2718, 1995 Ryan A, Al-Jehani RM, Mulligan KT, Jacobs IJ: No evidence exists for methylation inactivation of the p16 tumor suppressor gene in ovarian carcinogenesis. Gynecol Oncol 68:14 –17, 1998 McCluskey LL, Dubeau L: Biology of ovarian cancer. Curr Opin Oncol 9:465– 470, 1997 Dodson MK, Hartmann LC, Cliby WA, DeLacey KA, Keeney GL, Ritland SR, Su JQ, Podratz KC, Jenkins RB: Comparison of loss of heterozygosity patterns in invasive low-grade and high-grade epithelial ovarian carcinomas. Cancer Res 53:4456 – 4460, 1993