Catechol-O-Methyltransferase Polymorphism Is Not Associated with Ovarian Cancer Risk¹

In 1979, Cassagrande et al. (1) postulated that proliferation or malignant transformation of the ovarian epithelium can be induced by exposure to estrogen-rich follicular fluid after ovulation. Evidence supports a carcinogenic role for estrogen in several tissues, including human breast, endometrium,and ovary (2, 3). For example, unopposed estrogen replacement therapy was found to be associated with an increased risk of endometrioid or clear-cell epithelial ovarian tumors (OR, 2.56; 95%CI, 1.32–4.94; Ref. 3).

The primary mechanism of estrogen carcinogenesis is thought to involve estrogen receptor-mediated cell proliferation associated with spontaneous DNA replication errors (4). In addition, there is accumulating evidence suggesting that the metabolism of estradiol to 16α-hydroxyestrone and the catechol estrogens, which are also estrogenic, may be a contributing factor in tumor formation via direct and indirect genotoxicity (5). In particular, the 2- and 4-hydroxylations of estradiol form catechol estrogens, which can cause oxidative DNA damage through participation in redox cycling processes and/or through subsequent metabolism to quinones, can directly adduct DNA (6). Sulfation, glucuronidation, and methylation inactivate catechol estrogens by inhibiting their estrogenicity and preventing additional oxidative metabolism (7). Furthermore, the methylation of 2-hydroxyestradiol to 2-methoxyestradiol has been shown to inhibit tumor cell growth,stimulate apoptosis, and inhibit angiogenesis (8).

The methylation of catechol estrogens is catalyzed by COMT³(9). Lachmann et al. (10) found that the valine-108-methionine transition in the COMTgene accounted for a 3- to 4-fold variation in COMT activity. Twenty five percent of Caucasians are homozygous for the low-activity allele COMT^LL, 50% are heterozygous(COMT^LH), and 25% are homozygous for the high-activity allele COMT^HH(11, 12, 13). It has been hypothesized that this COMT polymorphism may modulate the risk of hormonal cancers because of a decreased ability of the protein encoded by COMT^L to methylate catechol estrogens (9).

Several studies have examined the association between the COMT polymorphism and breast cancer risk. Three of four peer-reviewed case-control studies (14, 15, 16, 17) found that the low-activity COMT allele, COMT^L, was associated with an increase in breast cancer risk, although the populations at risk varied among the studies. We hypothesized that COMT^L would also be associated with ovarian cancer. To test this, we used DNA that was originally isolated for a case-control study conducted by Hengstler et al.(18), which examined the association between ovarian cancer risk and polymorphisms of GSTT1 and GSTM1.

GSTs are a superfamily of genes whose products catalyze the conjugation of reactive intermediates to glutathione (19). They also may play a role in the inactivation of lipid and DNA products of oxidative stress (20), including those caused by catechol estrogen-associated redox cycling (21, 22). The absence of both GSTM1 and GSTT1 activity is caused by the inheritance of two null alleles of these genes (23, 24). Lavigne et al. (14) found that a GSTM1 null genotype in combination with COMT^LLincreased a postmenopausal woman’s risk of breast cancer (OR, 4.0;95% CI, 1.17–14.27). Although Hengstler et al.(18) found no statistically significant association between either the GSTM1 or GSTT1 polymorphism and ovarian cancer risk in these subjects, we hypothesized that women who were both COMT^LL and GSTM1and/or GSTT1 null would be at higher risk for ovarian cancer. To the best of our knowledge, the present study is the first to examine the relationship between the COMT polymorphism alone and in conjunction with GSTM1 and GSTT1 genotypes and ovarian cancer risk.

Included in this study were 108 Caucasian incident cases of ovarian cancer treated between 1994 and 1996 at the Department of Gynecology,University of Mainz. Histological typing was performed according to World Health Organization criteria, and tumor staging was done according to the guidelines of the Federation of Gynecology and Obstetrics. Non-epithelial and borderline tumors were excluded from the study. Epithelial tumors were divided into serous and nonserous carcinomas. The control group consisted of 124 Caucasian women who were either residents of a nursing home in Mainz or patients at the University of Mainz. As there were no specific criteria for individuals to join the nursing home, the nursing home residents were likely to be representative of the aging population. Controls were excluded if they were ever clinically diagnosed with a neoplastic disease or if they suffered acute diseases such as pneumonia or diabetes mellitus type I. Subjects with age-related chronic diseases,such as arthrosis or diabetes type II, were not excluded from this study. One hundred percent of incident cases and 67% of controls who were asked agreed to participate in the study.

All subjects signed an informed consent form and were asked about their age, smoking habits, and if any of their first- or second-degree relatives had ever had cancer. Blood samples were obtained from all study subjects. Whereas samples for COMT genotype analyses were available for all of the 108 cases, samples were available for only 106 controls. DNA samples from the remaining controls were depleted in a previous study. The analysis of the DNA samples from these subjects for their COMT genotype was approved by The Johns Hopkins School of Hygiene and Public Health Committee on Human Research.

Heparinized venous blood samples were taken from subjects. After the addition of 10% DMSO, the blood samples were stored at −80°C until shortly before the DNA was isolated. DNA was isolated from venous blood using either the IsoQuick Nucleic Acid Extraction Kit (Microprobe Corporation) or the QIAmp Blood Kit (Qiagen, Valencia, CA).

Sample DNA concentrations were low; consequently, semi-nested PCR and subsequent RFLP was performed using modified methods of Lavigne et al. (14) and Hoda et al.(25). A 293-bp fragment of the COMT gene was amplified using the forward primer c: 5′-TGGACGCCGTGATTCAGG-3′ and the reverse primer b: 5′-GTGAACGTGGTGGAACACC-3′. A 25-μl PCR reaction containing picogram-to-nanogram quantities of genomic DNA, 100 μm dNTPs, 2.5 μl 10×reaction buffer, 2.5 mmMgCl₂, 0.5 U Taq polymerase (all from Qiagen, Valencia, CA), and 300 nm primers was amplified in a Hybaid OmniGene thermocycler. The DNA was first denatured for 5 min at 94°C, and then was amplified during 30 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s, and subsequently by a 10-min extension at 72°C.

After the first round of PCR, each sample was diluted 100-fold, and 1μl of this was submitted to a second round of PCR and subsequent RFLP as described by Lavigne et al. (14). A reaction with DNA from a known COMT heterozygote and a reaction containing all reagents except DNA were included in each batch as a positive and negative control, respectively. A random subset (10%) of the samples was reanalyzed blindly for quality control. In the reanalysis, the genotype of one sample was unable to be determined,whereas the remaining samples confirmed the initial genotype.

All statistical analyses were performed using STATA Version 6 (Stata Corporation, College Station, TX). The association between the COMT polymorphism and ovarian cancer risk was examined using Mantel-Haenszel statistics for calculating ORs and 95% CIs, as well asχ ² tests for trend. Individuals homozygous for COMT^HH were designated as the referent category.

The association between COMT genotype and ovarian cancer was also examined within categories of other available potential risk factors. These variables included age group (≤50 years), having a first- or second-degree relative with cancer, smoking status, and GSTM1 and GSTT1 polymorphisms. The associations between COMT genotype and serous ovarian cancer risk and nonserous ovarian cancer risk were determined by comparing cases with each histology to the total control population. Results from age-adjusted analyses are presented because they differed from those of unadjusted analyses.

The available data for ovarian cancer cases (n =108) and controls (n = 124) from this case-control study are shown in Table 1. The mean age of cases was 58.9 ± 13.1 and that of controls was 51.0 ±14.8 (P < 0.001). Cases were more likely than controls to be 50 years of age and older. In other analyses presented in Table 1, it can be seen that cases were more likely than controls to have had a first- or second-degree relative with cancer. In addition, no statistically significant association was found between smoking and ovarian cancer risk.

The distribution of COMT^HH, COMT^LH, and COMT^LL genotypes was 25.0, 50.0, and 25.0%, respectively, in the cases and 23.6, 49.1, and 27.4%,respectively, in the controls. No statistically significant associations were found between COMT genotype and ovarian cancer risk (P_trend > 0.70; Table 2). This association did not vary by age,family history of cancer, GSTT1 or GSTM1genotype, or ovarian cancer histology(P_trend > 0.30 for all strata). The distribution of COMT genotype differed among women <50 years of age and women ≥50 years [the distribution of COMT^HH, COMT^LH, and COMT^LL was 29.7, 46.9, and 23.4%,respectively, in controls 50 years of age and older, and 14.7, 44.1,and 41.2%, respectively, in those younger than 50 years of age(P = 0.11)].

A valine-to-methionine polymorphism in exon 4 of COMTcauses a 3- to 4-fold reduction in enzyme activity and has been associated with an increased risk of breast cancer in several studies (26). Because ovarian cancer is also associated with estrogen exposure, we hypothesized that this polymorphism might also play a role in this disease. The results of this study do not support this hypothesis.

In our study population, the distribution of COMT^HH, COMT^LH, and COMT^LL genotypes was 25.0, 50.0, and 25.0%, respectively, in the cases and 23.6, 49.1, and 27.4%,respectively, in the controls. These distributions are similar to those of Caucasian controls in other studies. In the study by Lavigne et al. (14), the distribution was 24, 49, and 27%, respectively, in that by Thompson et al.(16), it was 27, 48, and 25%, respectively, and in that by Millikan et al. (17), it was 22, 50, and 28%, respectively. Therefore, the lack of association between COMT genotype and ovarian cancer risk in our study cannot be attributed to a skewed distribution of the COMT genotype in the control subjects, as both the cases and controls had the expected distribution for healthy Caucasian individuals. The distribution of COMT genotype differed among women younger than 50 years of age versus those 50 years of age and older. The distribution in older women was not statistically different from that of the entire population; however, in the younger age group, there were more women with a COMT^LL genotype and less with a COMT^HH genotype than in the entire population. This could possibly be attributed to small numbers in this age group.

The lack of association between COMT genotype and ovarian cancer risk in this study could be attributable to a number of factors. Because the number of study subjects is small, only moderate and large effects can be ruled out and, in stratified analyses, even moderate effects may have been undetected. Furthermore, information on specific factors that might have influenced the association between the COMT genotype and ovarian cancer risk, such as history of breast or ovarian cancer and oral contraceptive or postmenopausal hormone use, was unknown.

Besides analytical limitations, there are several possible biological reasons why an association between the COMT polymorphism and ovarian cancer risk was not detected. Although evidence indicates that catechol estrogens are produced in the human ovary, there is currently no method for measuring directly the estrogen levels in this tissue. It is feasible that catechol levels saturate the COMT enzyme in the ovary,leading to an excess of catechols, regardless of COMTgenotype. Furthermore, there are several other enzymes, such as sulfotransferases and glucuronosyltransferases, that can conjugate catechol estrogens. It is not known how much each individual enzyme contributes to this metabolic pathway. In addition, several enzymes that generate catechol estrogens, such as cytochromes P4501A1 and 1B1,have also been shown to be polymorphic in the human population (26, 27). It is possible that combinations of all of these polymorphic enzymes, including COMT, may be better predictors of ovarian cancer risk than polymorphisms in one or two genes alone. As a first approach to addressing this issue, we determined the association between the polymorphisms of either GSTM1 or GSTT1 in combination with the COMT polymorphism and ovarian cancer risk. Whereas we didn’t find a statistically significant association, it is possible this was attributable to a small number of study subjects. Future studies should focus on combinations of polymorphisms of all of the enzymes that are involved with estrogen metabolism and carcinogenesis and how they affect a woman’s risk of ovarian cancer.