Methylation-mediated suppression of detoxification, DNA repair, and tumor suppressor genes has been implicated in cancer development and progression. Studies also have indicated that concordant methylation of multiple genes (methylator phenotypes), rather than a single gene, may predict cancer prognosis. The current study was designed to determine whether a methylator phenotype exists in ovarian cancer, whether methylation frequencies differ between malignant ovarian tumors and ovarian tumors with low malignant potential (LMP or borderline), and whether methylation of multiple genes affects patient survival.
The current study included 234 consecutively diagnosed patients with either LMP (n = 19 patients) or malignant (n = 215 patients) ovarian tumors. DNA samples were extracted from fresh frozen tissues and were analyzed for methylation in the promoter region of 6 genes (p16, breast cancer 1 [BRCA1], insulin-like growth factor-binding protein 3 [IGFBP-3], glutathione S-transferase π 1 [GSTP1], estrogen receptor-α [ER-α], and human MutL homologue 1 [hMLH1]) by using methylation-specific polymerase chain reaction analysis.
The frequencies of methylation in malignant tumors and LMP tumors were 0% and 0% for GSTP1, respectively; 9% and 0% for hMLH1, respectively; 21% and 5% for BRCA1, respectively; 42% and 21% for p16, respectively; 44% and 26% for IGFBP-3, respectively; and 57% and 42% for ER-α, respectively. A methylator phenotype was not detected, but a calculated methylation index (MI) that was based on the total number of genes methylated in each tumor was associated with ovarian cancer risk and progression. A higher MI was associated with malignant tumors (odds ratio, 10.11; 95% confidence interval [95% CI], 1.19–85.75) and disease progression (hazards ratio, 6.53; 95% CI, 1.39–30.65).
Ovarian cancer is one of the most lethal malignancies and is the leading cause of death among women with gynecologic cancers.1 In 2005, it was estimated that 22,220 women would develop ovarian cancer, and 16,210 women would die from the disease in the U.S.2 Despite advances in cancer research and treatment, survival for patients with ovarian cancer remains low; >50% of patients die within 5 years of their ovarian cancer diagnosis.3 This poor survival rate is due in part to the lack of sensitive and specific methods of early detection. Because symptoms in early-stage ovarian cancer generally are nonspecific, patients with ovarian cancer usually are diagnosed with either Stage III or Stage IV disease that already has spread beyond the ovary.4 A better understanding of the molecular mechanisms that are responsible for ovarian cancer development and progression will help improve the diagnosis and treatment of the disease.
Aberrant DNA methylation is now recognized as one of the most common molecular abnormalities in cancer.5 This epigenetic modification occurs at the cytosines of CpG dinucleotides, which often exist in clusters called CpG islands. When methylation of these sites occurs in the promoter region of a gene, it can result in chromatin condensation and gene silencing. In cancer cells, aberrant methylation frequently has been reported in tumor suppressor genes, DNA repair genes, and genes related to cancer metastasis and invasion.6 The silencing of these functionally important genes leads to a shift of cells from a normal cellular cycle to a state of high proliferation that favors tumor development and progression.7, 8 It has been observed that promoter methylation of specific genes in cancer occurs in both a tissue-specific and a cell-specific manner, making the identification of methylation patterns a potentially useful tool for cancer management.9 This may be especially important for patients with ovarian cancer, because early detection and accurate disease characterization can improve survival; there is a large difference in 5-year survival between patients with localized, Stage I tumors (94%) and patients with Stage III or IV disease (29%).2
Recent research has focused on the identification of groups of genes with consistent, concurrent methylation (methylator profiles or phenotypes). This phenomenon initially was observed in colorectal cancer and was described as a CpG island methylator phenotype.10 By using a panel of 7 genes, a distinct group of tumors was identified that had a 3-fold to 5-fold elevation in methylation levels.11 Similar findings subsequently were reported for numerous other cancers, including gynecologic cancers like cervical,12, 13 uterine,14 and ovarian cancers.15–17 However, with the relatively small number of genes analyzed and with the differences in selection of genes and analytic methods among studies, there remains a great deal of uncertainty regarding the presence and classification of these methylator phenotypes. What remains particularly unclear is which genes should be used to identify a methylator phenotype that best describes the process of either tumorigenesis or progression in a tissue-specific manner.
In the current study, we analyzed a clinical cohort of women with malignant or low malignant potential (LMP) epithelial ovarian tumors for the concurrent methylation of 6 genes: estrogen receptor-α (ER-α), insulin-like growth factor-binding protein 3 (IGFBP-3), p16, breast cancer 1 (BRCA1), human MutL homologue 1 (hMLH1), and glutathione S-transferase π 1 (GSTP1). These genes are known as important in the development and progression of cancer. Suppression of p16, BRCA1, and hMLH1 expression by methylation has been reported in various types of malignancies, including ovarian cancer.6, 16IGFBP-3, ER-α, and GSTP1 methylation has also been found in a number of malignancies, whereas their role or presence in ovarian cancer is less clear.18–20 Our previous research showed that IGFBP-3 methylation was associated with the survival of patients with early-stage ovarian cancer.21 Although methylation in these functionally important genes has been observed in cancer, they have not been examined previously in the context of methylation patterns specifically in ovarian cancer. In the current study, we assessed whether concurrent methylation exists within this panel of genes, whether methylation frequencies are greater in malignant than in LMP ovarian tumors, and whether a higher methylation index (MI) corresponds to unfavorable survival outcomes.
MATERIALS AND METHODS
Study Population and Data Collection
Two hundred sixty-four women underwent surgery for suspected ovarian cancer between October 1991 and February 2000 at the University Hospital, University of Turin (Turin, Italy). Staging was performed according to the International Federation of Gynecologists and Obstetricians criteria.22 Histologic grade and type were determined according to World Health Organization standards.23 Patients were excluded if they had metastatic tumors of nonovarian origin (n = 23 patients), primary ovarian tumors of nonepithelial origin (n = 6 patients), or endometriosis (n = 1 patient). Of the remaining patients, 19 women were diagnosed with LMP (or borderline) tumors (Grade 0), and 215 women had histologically confirmed primary epithelial ovarian cancer. Women with LMP tumors and malignant tumors had similar age distributions (Table 1). Histologic types of malignant tumors included serous (40.2%; n = 86 tumors), endometrioid (19.6%; n = 42 tumors), undifferentiated (17.3%; n = 37 tumors), mucinous (8.4%; n = 18 tumors), clear cell (7.5%; n = 16 tumors), and other epithelial (7.0%; n = 15 tumors). For LMP tumors, the main histologic type was serous (68.4%; n = 13 tumors), and the rest were endometrioid (10.5%; n = 2 tumors) and mucinous (10.5%; n = 2 tumors). LMP tumors tended to be early stage (Stages I and II; 81%), whereas malignant tumors predominately were late stage (Stages III and IV; 70%) (Table 1).
Table 1. Clinical and Pathologic Characteristics of the Patient Population
P values were based on chi-square tests, Fisher exact tests, or Student t tests, as appropriate.
Mean ± SD age, y
53.3 ± 19.8
57.7 ± 11.4
The average age of the patients with ovarian cancer who underwent surgery was 57.2 years (standard deviation, 12.3 years), and the patients ranged in age from 25 years to 89 years. The prevailing postoperative treatment for these patients was chemotherapy, which was administered to the majority of patients (n = 182 patients; 84.7%). Chemotherapeutic regimens consisted of combinations of doxorubicin, carboplatin, cisplatin, cyclophosphamide, epirubicin, and taxol. All but 1 patient (n = 181 patients; 99.5%) on chemotherapy received a platinum-based compound (either cisplatin or carboplatin). In addition to platinum-based compounds, taxol was administered to 76 patients (35.4%). Follow-up information was collected after surgery through June 2001 for 205 patients with primary epithelial ovarian cancer (95.3%). Overall survival (OS), defined as the time from the date of surgery to the date of death or last contact if the patient was still alive, ranged from 0.6 months to 114.1 months (median, 31.1 months). Disease progression-free survival (PFS) was defined as the duration from the date of surgery to the date of first evidence of local recurrence, distant metastasis, or last contact. PFS ranged from 0.6 months to 108.8 months (median, 20.6 months). Of 205 patients who had complete follow-up information available, 42.4% (n = 87 patients) were alive without disease progression, 12.7% (n = 26 patients) remained alive after disease progression, 33.2% (n = 68 patients) died after recurrence, and 11.7% (n = 24 patients) died of the disease without remission.
Sodium Bisulfite Modification and Methylation-Specific Polymerase Chain Reaction Analysis
Fresh tumor tissue was collected from each patient during surgery. Specimens were snap-frozen in liquid nitrogen immediately after surgical resection and stored at −80 °C until analysis. Tissue sections were examined by 2 pathologists independently; the majority of tissue sections had from 80% to 90% tumor content. Genomic DNA was extracted from approximately 100 mg of ovarian tumor tissue by using standard phenol:chloroform methods and then submitted for chemical modification using sodium bisulfite. Briefly, DNA was denatured with 2 N NaOH, followed by treatment with 10 mM hydroquinone and 3 M sodium bisulfite. Modified DNA was then purified by using the QIAEX II Gel Extraction Kit (QIAGEN, Valencia, CA) and concentrated with Pellet Paint NF Co-Precipitant (Novagen, San Diego, CA). Final suspension volume was in 100 μL of Tris-HCl ethylenediamine tetraacetic acid buffer.
Promoter methylation status was analyzed by using methylation-specific polymerase chain reaction (PCR) analysis (MSP) for 6 genes: ER-α, IGFBP-3, hMLH1, BRCA1, p16, and GSTP1 (Fig. 1). Two sets of PCR primers were used to amplify each promoter region of interest: One pair recognized a sequence in which the CpG sites were unmethylated (bisulfite-modified to TpG), whereas the other pair recognized methylated CpG sites (unmodified by bisulfite treatment). Table 1 contains information on primer sequences, annealing temperatures, product sizes, and references for the MSP assays. In Table 2, primer bases in bold indicate a C-to-T transition because of bisulfite treatment of the DNA, and underlined bases indicate the cytosine of a CpG site. All MSP reactions began with initial denaturing for 5 minutes at 94°C and concluded with a final extension for 7 minutes at 72°C. Annealing temperatures and times varied from assay to assay, ranging from 57°C to 65°C and from 30 seconds to 1 minute. The number of amplification cycles ranged from 35 cycles to 40 cycles. MSP products were visualized on 2.0% to 2.5% agarose gels stained with ethidium bromide. Determination of methylation status for each sample was made without knowledge of the patient's clinical or tumor characteristics or survival outcome. Methylation status was confirmed by direct sequencing for a subset of samples from each assay; results demonstrated that MSP results were reliable.
Table 2. The Primers Used for Detecting Methylation in p16, Breast Cancer 1, Insulin-Like Growth Factor-Binding Protein 3, Estrogen Receptor α, Human MutL Homologue 1, and Glutathione S-Transferase π
Temp indicates temperature; bp, base pairs; BRCA1, breast cancer 1; IGFBP-3, insulin-like growth factor-binding protein 3; ER-α; estrogen receptor-α; hMLH1, human MutL homologue 1; GSTP, glutathione S-transferase π.
Bases in bold represent a C-to-T transition as a result of sodium bisulfite treatment; underlined bases represent CpG sites.
5′-TGC GTT CGG CGG TTG CGG-3′
5′-GAC CCC GAA CCG CGA CCG-3′
5′-G TGT GTT TGG TGG TTG TGG AGA-3′
5′-CCC AAC CCC AAA CCA CAA CCA TAA-3′
5′-TCG TGG TAA CGG AAA AGC GC-3′
5′-AAA TCT CAA CGA ACT CAC GCC G-3′
5′-T TGG TTT TTG TGG TAA TGG AAA AGT GT-3′
5′-C AAA AAA TCT CAA CAA ACT CAC ACC A-3′
5′-CGA AGT ACG GGT TTC GTA GTC G-3′
5′-CGA CCC GAA CGC GCC GAC C-3′
5′-TT GGT TGT TTA GGG TGA AGT ATG GGT-3′
5′-CAC CCA ACC ACA ATA CTC ACA TC-3′
5′-CGA GTT GGA GTT TTT GAA TCG TTC-3′
5′-CTA CGC GTT AAC GAC GAC CG-3′
5′-A TGA GTT GGA GTT TTT GAA TTG TTT-3′
5′-ATA AAC CTA CAC ATT AAC AAC AAC CA-3′
5′-ACG TAG ACG TTTTAT TAG GGTCGC-3′
5′-CCT CAT CGT AAC TAC CCG CG-3′
5′-TTT TGA TGT AGA TGT TTT ATT AGG GTT GT-3′
5′-ACC ACC TCA TCA TAA CTA CCC ACA-3′
5′-TTC GGG GTG TAG CGG TCG TC-3′
5′-GCC CCA ATA CTA AAT CAC GAC G-3′
5′-GAT GTTTGG GGT GTA GTG GTT GTT-3′
5′-CCA CCC CAA TAC TAA ATC ACAACA-3′
Associations between clinical and pathologic characteristics and individual promoter methylation status were examined by using chi-square or Fisher exact tests, as appropriate. Logistic regression was used to evaluate differences in methylation status between LMP and malignant tumor samples. To examine the methylation of multiple gene promoters, an MI was created (MI = total number of genes methylated/total number of genes analyzed). Two MIs were calculated: 1 that included ER-α (6 genes) and 1 that did not (5 genes). MIs were analyzed numerically and categorically. In categorical analysis, the indices were dichotomized into high-MI and low-MI based on median values. Differences in MI between clinical and pathologic variables were compared by using the Mann–Whitney U test. The effects of methylation on PFS and OS were evaluated by using Kaplan–Meier survival curves, and differences in survival were examined by using the log-rank test. Cox proportional hazards regression was then used to assess the role of methylation in ovarian cancer prognosis; associations were evaluated by using hazards ratios (HRs) and their 95% confidence intervals (95% CIs). Both unadjusted and adjusted HRs were calculated; covariates included patient age at surgery, disease stage, tumor grade, and residual tumor size. Analyses were performed by using SAS software (version 8.02; SAS Institute, Cary, NC).
Promoter methylation in 6 genes (ER-α, IGFBP-3, p16, BRCA1, hMLH1, and GSTP1) was evaluated in 215 patients with epithelial ovarian cancer and 19 patients with LMP tumors. Table 3 shows the frequency of methylation for each gene and the corresponding estimate of relative risk for ovarian cancer. Promoter methylation was observed frequently for ER-α, IGFBP-3, and p16, in which methylation occurred in 56.5%, 43.7%, and 41.6% of patients with malignant tumors, respectively. Lower rates of methylation were observed for BRCA1 (21%) and hMLH1 (8.9%), whereas methylation of GSTP1 was completely absent in this patient population. Methylation of the IGFBP-3 promoter was concordant with methylation of the ER-α promoter (P = .012); no significant associations between other genes were observed.
Table 3. Differences in Methylation Status between Patients with Malignant and Borderline Ovarian Tumors
Malignant (n = 215)
Borderline (n = 19)
OR (95% CI)
OR (95% CI)
OR indicates odds ratio; 95% CI, 95% confidence interval; ER-α; estrogen receptor–α; IGFBP-3, insulin-like growth factor-binding protein 3; BRCA1, breast cancer 1.; hMLH1, human MutL homologue 1; GSTP1, glutathione S-transferase π1; ND, not determined (because of 0 cell counts or limited degrees of freedom).
With the exception of GSTP1, the frequency of promoter methylation was higher in patients with malignant compared to LMP tumors. After adjusting for age, patients with malignant tumors were 3 times more likely to have methylation in the promoter region of the IGFBP-3 gene (odds ratio [OR], 3.83; 95% CI; 1.06–13.81). Although there were no statistically significant differences, methylation of the other 4 genes, ER-α, p16, BRCA1, and hMLH1, did occur more frequently in malignant tumors than in LMP tumors (Table 2).
By using an MI, we also evaluated promoter methylation in these genes together. MIs were significantly higher in malignant tumors (median MI, 0.33) than in LMP tumors (median MI, 0.17; P = .006). When comparing the total number of methylated genes, a trend was observed in which increasing numbers of methylated genes were more likely to occur in patients who had malignant tumors than in patients who had LMP tumors. Patients who had methylation in ≥3 genes, for example, were 10 times more likely to have malignant tumors than LMP tumors (OR, 10.11; 95% CI, 1.19–85.75). This trend retained borderline significance after adjustment for age (P = .050) (Table 2).
No associations were found between promoter methylation and age at surgery, disease stage, or residual tumor size for any of the 6 genes when they were analyzed individually (data not shown). Histologic grade, however, was associated with promoter methylation of the BRCA1 gene (P = .025); methylation occurred more frequently in patients with higher grade tumors. Patients with methylation of the p16 gene were more likely to experience disease progression (P = .048). The median MI among the patients was 0.33, which corresponded to methylation in 2 of 6 genes studied. Based on this median value, patients were classified into 2 groups: high-MI (MI values > 0.33) or low-MI (MI values ≤ 0.33). No significant associations were observed between high MI and age at surgery, disease stage, histologic grade, or residual tumor size (data not shown). Because promoter methylation of the ER-α gene was high in both LMP and malignant tumors, and because there is a possible correlation between ER methylation and aging,24, 25 a second MI was calculated that excluded ER-α. This MI had a median value of 0.2, which corresponded to having 1 of the 5 genes methylated. This MI also was used to categorize patients into high-MI (MI values > 0.2) or low-MI (MI values ≤ 0.2) groups. Again, no associations were observed between MI and any clinicopathologic variable (data not shown).
Patients who had methylation of the p16 promoter region were more likely to experience disease progression compared with patients who did not have p16 methylation (data not shown; P = .013). None of the other genes individually demonstrated an association between methylation and either PFS or OS. However, when the effect of MI on survival was examined, a significant difference was observed for PFS. Patients in the high-MI group had significantly shorter PFS compared with patients in the low-MI group when analysis both included ER-α (data not shown; P = .0016) and excluded ER-α (Fig. 2) (P = .0015). A similar but nonsignificant, difference also was observed for OS either including ER-α (P = .2068) or excluding ER-α (Fig. 2) (P = .0859).
Multivariate Cox regression analysis showed that, when adjusting for age at surgery, disease stage, histologic grade, and residual tumor size, promoter methylation of p16 or hMLH1 was associated significantly with an increased risk of disease progression (Table 4). However, methylation of any individual gene was not associated significantly with OS in either univariate or multivariate analysis. For methylation of multiple genes, a significant trend was observed between the increase in risk of disease-progression and the increasing numbers of methylated genes (P = .002). Patients with ≥3 methylated genes had an approximately 3-fold to 7-fold increase in their risk of disease progression. A similar trend was observed for OS, although statistical significance was not observed until ER-α was excluded from the analysis. Patients with ≥3 methylated genes had an approximately 2-fold to 4-fold increase in their risk of death (test for trend; P = .040; data not shown).
Table 4. Survival Analysis of Promoter Methylation in Patients with Ovarian Cancer*
It has been demonstrated previously that promoter hypermethylation is among the most common and critical epigenetic events in cancer.5, 26, 27 However, methylation of single genes may have limited value in clinical applications. The objective of the current study was to determine whether the evaluation of methylation in a panel of genes could help to distinguish between LMP and malignant ovarian tumors and help to predict the prognosis of patients with ovarian cancer. Although gene expression was not evaluated in this study, previous studies have shown that DNA promoter methylation in these genes is correlated highly with decreased levels of gene expression.6, 28 In this study, methylation was found in 5 of 6 genes analyzed, with methylation frequencies ranging from 9% in hMLH1 to 56.5% in ER-α. With the exception of GSTP1 (which was unmethylated in all samples), methylation frequencies were higher in malignant ovarian tumors (9–57%) than in LMP ovarian tumors (0-42%). Furthermore, patients with malignant tumors were more likely to have a higher MI than patients with borderline tumors (P = .006), and patients with ovarian cancer who had a high MI were more likely to have a poor prognosis compared with patients who had a low MI. Therefore, increasing numbers of methylated genes appear to be important in the development and progression of ovarian cancer.
The methylation frequencies among malignant ovarian tumors in this study were similar to those reported previously for all of the genes except p16.16, 29 A frequency of 41.6% in p16 methylation was relatively high compared with other studies. However, this likely was because of the difference in MSP assay design. McCluskey et al. found a p16 methylation frequency of only 5% when primers were designed to amplify a 224-base pair (bp) product, which included 140 bp of the promoter and 84 bp of exon 1.30 However, when MSP was performed using primers to amplify only the 140-bp region of the promoter, the methylation frequency was much higher (57%) and was close to that observed in the current study. Our p16 primers amplified a 125-bp, methylated product and a 129-bp, unmethylated product, which were similar in location and size to those used successfully by Herman et al. to examine methylation in a number of common cancers.31 To verify our MSP results, representative MSP samples were analyzed further by direct sequencing, and methylation status was confirmed. In contrast to the high rate of methylation found in the p16 gene, no methylation was observed in the GSTP1 promoter region. Because GSTP1 methylation has been found frequently in prostate cancer, this lack of methylation in ovarian cancer confirms the tissue specificity of GSTP1 methylation.32, 33 The finding of no GSTP1 methylation in the current study also indicates that our MSP analysis was reliable and that tissue processing and analysis were contamination free. In this regard, the selection of GSTP1 for this panel well may serve as a negative internal control for the analysis.
In the current study, methylation frequencies for patients with borderline tumors were similar to those reported previously.16, 34 Makarla et al. reported a high MI for invasive ovarian carcinomas, a low MI for benign cystadenomas, and an intermediate MI for LMP tumors. This included a p16 methylation rate of 22%, which is similar to our rate of 21%. Our results showed a significantly lower MI of 0.15 for LMP tumors compared with 0.28 for malignant tumors (P = .006). Relatively few patients with LMP ovarian tumors (26%) had methylation in >1 of the analyzed genes, and even fewer patients had methylation in >2 genes (5%). This is in contrast to patients with malignant tumors, 56% of whom had methylation of >1 gene and 24% of whom had methylation of >2 genes. This offers further evidence to support the idea that there is an intermediate level of methylation in LMP tumors that is elevated slightly above the level found in benign tumors or normal tissue (based on findings in the literature) but significantly lower than the level of methylation found in malignant tumors.
Methylation of multiple genes was associated with both PFS and OS in the current study. Patients were more likely to experience disease progression or death as the MI increased. In fact, patients with the highest MI were 4 to 5 times more likely to have disease progression or to die. This is one of the few studies to examine the prognostic value of multiple gene methylation in ovarian cancer using a candidate gene approach. In their study, Wei et al. used differential methylation hybridization technology to classify 19 patients into groups with either high or low levels of methylation, and the results indicated that high levels of methylation were associated with poor PFS.35 This association or trend also was observed in other types of cancer. In bladder cancer, it was observed that patients with a high MI (≥0.2) had reduced OS.36 In esophageal adenocarcinoma, Brock et al. observed that patients with a high MI (>0.5) had a significantly decreased OS rate and a significantly increased recurrence rate.37 The results of our current study confirm that methylation is an important event in ovarian cancer and that an MI may be used for ovarian cancer prognosis if a panel of relevant genes can be developed.
Although the majority of epithelial ovarian cancers express ERs, uncertainty remains about the prognostic value of ER status, expression, and promoter methylation.38 Some investigators have reported that malignant ovarian cells exhibit decreased ER-α expression compared with human ovarian surface epithelial cells.7, 39 Other investigators have reported no change in ER-α expression when comparing normal and malignant ovarian cells, citing instead that the ratio of ER-α/ER-β (and specifically changes in ER-β expression) is what determines the change to neoplastic growth.19, 40 In the current study, we observed that the ER-α gene was highly methylated in both malignant tumors (56.5%) and LMP tumors (42.1%). Because of the unclear role of ER-α in ovarian cancer and the possible correlation between aging and ER-α methylation,24 our statistical analysis was performed both including and excluding this gene. After excluding the ER-α gene, the results of our analysis were more pronounced, suggesting that this gene is not a good candidate for future methylation analyses.
In this study, we were unable to confirm the presence of a methylator phenotype in ovarian cancer. Although tumors did exhibit methylation of multiple genes, no significant patterns of concurrent methylation were observed. Our failure to detect a methylator phenotype may have been because of the relatively small number or varied nature of the genes that were analyzed in the study. By using differential methylation hybridization, for example, Ahluwalia et al. were able to analyze 742 loci and found concurrent methylation at 4 loci in 10 ovarian tumor samples.15 With an estimated 45,000 CpG islands present in the human genome and with the methylation of a predicted 600 to 4900 CpG islands in ovarian cancer, the selection of only 6 loci may limit our ability to find concurrent methylation patterns in this population of patients with ovarian cancer.41 Alternatively, several researchers have suggested that quantitative methylation assays are needed to identify methylator phenoypes accurately.11 However, Strathdee et al. used MSP and found 2 distinct methylator phenotypes: One set of tumors had methylation of BRCA1, and a second group had methylation of HIC1, MINT25, MINT31, and p73.17 Our results are similar to those reported by other groups that used MSP, including Rathi et al., who examined 9 genes in ovarian tumors and also were unable to identify a methylator phenotype.16 Thus, it remains unclear whether methylator phenotypes are a relatively rare occurrence in ovarian cancer or whether analytic methods dictate the indentification.11
The current results illustrate the importance of methylation in the development and progression of ovarian cancer. Clearly, methylation is a common event: Greater than 80% of patients had methylation in ≥1 of 6 genes analyzed, and approximately 25% had methylation in ≥3 genes. In contrast, only 1 of 19 patients with LMP tumors had methylation in ≥3 of the genes analyzed; this suggests a potential usefulness of methylation as a marker to distinguish between malignant and LMP tumor types. In addition, although a methylator phenotype was not identified in this study, a high MI was associated with both PFS and OS. This suggests that an MI composed of important tumor suppressor and DNA repair genes may have clinical implications in the prognosis for patients with ovarian cancer. Methylation in these important genes also may serve as a potential therapeutic target for future studies that examine demethylating agents.