Main

An increasing number of epidemiologic studies have examined associations between GSTM1, GSTT1, and GSTP1 and epithelial ovarian cancer. Associations with GSTM1, GSTT1, and GSTP1 genotype have not been observed in most studies, many of which have had small sample sizes or other design limitations. However, GSTM1 null genotype was found to be more frequent among cases than controls in a recent study of epithelial ovarian cancer that included more subjects (59% vs. 49%, OR = 1.5, 95% CI 1.1–2.1).54

Results from this and one other recent case-control study suggest that deletion of GSTM1 may increase the risk of specific histological subtypes of ovarian cancer (endometrioid and clear cell).

This Human Genome Epidemiology Network (HUGE Net) review critically evaluates published studies of these associations and offers suggestions for future research.

GENE

Glutathione S-transferases (GSTs) catalyze the conjugation of glutathione to numerous potentially genotoxic compounds, including aliphatic aromatic heterocyclic radicals, epoxides, and arene oxides.1–3 The enzymes detoxify carcinogenic polycyclic aromatic hydrocarbons and conjugate isothiocyanates. Six families of enzymes have been classified as alpha, mu, pi, theta, zeta, and omega. Glutathione S-transferase-mu (GSTM1), glutathione S-transferase-theta (GSTT1), and glutathione S-transferase-pi (GSTP1) have been studied most.3 The GSTM1 gene, which is located on chromosome 1p13.3, codes for the enzyme glutathione S-transferase-mu. The GSTT1 gene codes for the enzyme glutathione S-transferase-theta; the GSTT1 gene is located on chromosome 22q11.2. The GSTP1 gene codes for the enzyme glutathione S-transferase-pi and is located on chromosome 11q13. GSTA1 and GSTA2, the two major glutathione S-transferase-alpha genes, are mentioned less frequently in the literature and are located on chromosome 6p12. The GSTZ1 gene that encodes glutathione S-transferase-zeta was mapped to chromosome 14q24.3.4,5 The glutathione S-transferase-omega (GSTO1) gene was recently characterized.6 The GSTs have broad and overlapping substrate specificities that have been previously reviewed.7

GENE VARIANTS

GSTM1 is polymorphically expressed, and three alleles have been identified:GSTM1-0, GSTM1a, and GSTM1b.3 There is no evidence of functional differences between GSTM1a and GSTM1b, which differ by a substitution in one base pair.8 The GSTT1 locus has a single allele that is either present or absent (deleted). Two functionally different genotypes at the GSTT1 locus have been described. The first, GSTT1-0, is a homozygous deletion. The second, GSTT1-1, comprises genotypes with one or two functional alleles present. There are also three variant alleles of GSTP1. GSTP1a and GSTP1b differ by an A to G substitution at codon 105 that exchanges isoleucine for valine. GSTP1c varies from the “a” allele by the substitution of valine for alanine at codon 114. Individuals with homozygous deletions of GSTM or GSTT have reduced or no glutathione S-transferase activity and therefore may be unable to eliminate electrophilic carcinogens as efficiently. This may increase the risk of somatic mutations leading to tumor formation. Evidence is lacking on whether heterozygosity in either GSTM1 or GSTT1 affects gene function.8 An analysis of functional differences in the GSTP1 variant proteins was performed by evaluating the ability of recombinant GSTP1 proteins to catalyze the conjugation of glutathione with 1-chloro-2,4-dinitrobenzene, a universal GST substrate. Results showed that catalytic activity was 3- to 4-fold higher for the GSTP1a variant than for either GSTP1b or GSTP1c.9 This indicates that enzyme activity is significantly reduced in individuals who carry the valine allele.

Cotton et al.8 summarized information on the frequency of homozygosity and heterozygosity of GSTM1 and GSTT1 allelic variants in different populations and ethnic groups; updated information on the frequency of these variants was provided by Geisler and Olshan.10 Both of these articles are posted on the Centers for Disease Control website on the Internet.8,10 The frequency of the GSTM1 null genotype ranges from 23% to 48% in African populations, 33% to 63% in Asian populations, 39% to 62% in European populations, and 23% to 62% in U.S. populations.8 In selected U.S. populations, the frequency of the GSTM1 null genotype ranges from 23% to 41% in African-Americans, 35% to 62% in whites, 32% to 53% in Asians, and 40% to 53% in Mexican-Americans.

The frequency of the GSTT1 null genotype ranges from 15% to 26% in African populations, 16% to 64% in Asian populations, and 10% to 21% in European populations.8 In U.S. populations, the frequency of the GSTT1 null genotype ranges from 22% to 29% in African-Americans, 15% to 27% in whites, and 10% to 12% in Mexican-Americans.

To obtain information on the frequency of homozygosity and heterozygosity of GSTP1 allelic variants in different populations and ethnic groups, we searched MEDLINE by using the MeSH heading “glutathione transferase” and the text word “GSTP1.” Reference lists were also reviewed from published articles. Papers written in English and published between January 1993 and March 2001 were reviewed. The majority of the studies reviewed, including studies reporting genotype frequencies among individuals without cancer, had a case-control design.11–33 The frequencies for polymorphisms in exon 5 and exon 6 of the GSTP1 gene among control individuals are summarized in Table 1. Studies have been conducted on all continents with the exception of Africa and South America; only two studies have been conducted in the United States. The frequency of the homozygous wildtype genotype at exon 5 ranges from 39% to 59% in whites, 52% to 93% in Asians, and 35% in the single study involving African-Americans. The heterozygous genotype ranged from 30% to 50% among whites, 24% to 44% among Asians, and 46% among African-Americans. The homozygous mutant genotype ranged from 4% to 16% among whites, 4% to 5% among Asians, and 19% for African-Americans. For the exon 6 polymorphism, the frequency of the homozygous wildtype genotype ranged from 75% to 85% among whites, 98% to 100% among Asians, and 95% among African-Americans. Heterozygous genotypes ranged from 15% to 25% among whites, 0% to 2% among Asians, and 5% among African-Americans. The homozygous mutant genotype was only found among persons from India (2%).

Table 1 Population frequency of GSTP1 polymorphisms among controls
Full size table

DISEASE

Ovarian cancer is the sixth most common cancer among U.S. women.34 An estimated 23,300 new cases will be diagnosed in 2002 and an estimated 13,900 U.S. women will die from the disease this year. Based upon SEER data for the period 1988 to 1992, ovarian cancer incidence rates among U.S. women were highest among American-Indian women (age-adjusted rate 17.5 per 100,000), followed by non-Hispanic white (age-adjusted rate 15.8 per 100,000), Vietnamese, white Hispanic, and Hawaiian women.35 Rates were intermediate among African-American women (age-adjusted rate 10.2 per 100,000) and lowest among Korean and Chinese women in the United States over that time period. Worldwide, the highest ovarian cancer incidence rates are found in white women in Europe (especially Scandinavian countries), Israel, and North America. Rates among women in Central and South America and in Asia (including Japan) are relatively low.36

The majority of malignant ovarian tumors originate in epithelial cells. More is known about the descriptive epidemiology and etiology of epithelial ovarian cancer than germ cell or sex cord-stromal (follicular) tumors.36 The most common type of epithelial tumors is serous carcinoma (which comprises about 40% of epithelial ovarian cancers), followed by mucinous carcinomas (6–16%), endometrioid carcinomas (15–25%), and clear cell carcinomas (5–11%).37

Established risk factors for epithelial ovarian cancer include age, family history of ovarian cancer, personal history of endometrial or breast cancer, and nulliparity.36 Other risk factors that have been less consistently associated with ovarian cancer include history of infertility, use of infertility drugs, early age of menarche, late menopause, and use of talc.36 History of one or more full-term pregnancies, use of oral contraceptives, and possibly tubal ligation and lactation are protective.38–40 Several studies of ovarian cancer have examined possible associations with endogenous estrogens, hormone replacement therapy, various dietary factors, and other purported risk factors, but results to date have been inconsistent.36,41 Results from a recent cohort mortality study suggest that hormone replacement therapy of long duration may modestly increase risk.42

Although the focus of this review is on glutathione S-transferase polymorphisms, we briefly summarize here major genes that have been implicated in some cases of ovarian cancer. Hereditary ovarian cancer due to BRCA1 or BRCA2 gene mutations accounts for roughly 5% to 11.7% of all cases of ovarian cancer.43–45 Women carrying germline mutations in BRCA1 or BRCA2 have a high lifetime risk of developing breast and/or ovarian cancer.43 Other genetic syndromes that may include ovarian cancer as an associated feature include Peutz-Jeghers syndrome, multiple endocrine neoplasia type 1 (MEN1), and hereditary nonpolyposis colon cancer (HNPCC).46,47 Germline mutations in the LKB1 gene are associated with Peutz-Jeghers syndrome.46 Mutations of MLH1 and MSH2 have been implicated in HNPCC.47

DISEASE ASSOCIATIONS

The studies reviewed in this section were identified by searching MEDLINE by using the MeSH headings “ovarian neoplasms” and “glutathione transferase,” and the text words “GST” and “glutathione S-transferase.” Reference lists were also reviewed from published articles. Papers written in English and published between January 1993 and March 2001 were reviewed. Studies that looked at prognosis or chemoresistance were excluded.48–50

Lallas et al.51 examined associations with GSTM1 genotype in 138 genotyped patients with epithelial ovarian cancer in Iowa and 77 unmatched controls who were paid volunteers (Table 2). Women with a family history of ovarian, breast, or colon cancer were excluded from the control group. Women who had the null GSTM1 genotype were not at increased risk of ovarian cancer (OR = 0.7, no confidence interval given). No associations were observed between GSTM1 genotype and age at diagnosis, histological type, or stage of ovarian cancer. This study is limited by the use of volunteer controls and by the small sample size. Associations with GSTM1a and GSTM1b genotype were examined by Sarhanis et al.52 in a study involving 84 patients with epithelial ovarian cancer and 312 control patients from the same hospital in North Staffordshire. No associations were observed with the null GSTM1 genotype in comparisons with homozygosity for either the GSTM1a or GSTM1b genotypes (Table 2). The study is limited by the use of patient controls and by the small number of cases. Of 23 patients with tumors that produced or expressed p53 protein, 20 (87%) were GSTM1 null. The frequency distributions of GSTM1 genotypes in p53-positive and p53-negative tumor samples were significantly different (P = 0.002). Hengstler et al.53 examined associations with GSTM1 genotype in 103 patients with epithelial ovarian cancer and 115 control patients. The controls included patients from the same hospital in Mainz, Germany, and also nursing home residents. No associations were observed with the null GSTM1 genotype (OR = 0.8, 95% CI 0.4–1.3). This study is limited by the use of patient controls and by the small sample size. GSTM1 genotype was unrelated to histological type, grade, or stage of ovarian cancer.

Table 2 Summary of studies of epithelial ovarian cancer and GSTM1
Full size table

Different results were obtained by Baxter et al.54 in a study of 219 women with epithelial ovarian cancer who were treated at hospitals in southeast England and 219 healthy female volunteer controls who resided in the same region. The GSTM1 null genotype was more frequent among the cases than controls (59% vs. 49%, OR = 1.5, 95% CI 1.1–2.1). This study is limited by the use of patient controls and by the small sample size. After stratification by histological subtype, a significantly increased frequency of the GSTM1 null genotype was only observed for endometrioid (65%, P = 0.013) and combined endometrioid/clear cell ovarian cancers (67%, P = 0.004). Further evidence that deletion of GSTM1 may specifically increase the risk of these histological subtypes was observed by Spurdle et al.24 in a case-control study of epithelial ovarian cancer from Brisbane, Australia (285 hospitalized cases and 295 convenience controls from a twin study who were matched to the cases on age). These investigators observed no association with GSTM1 null genotype overall (age-adjusted OR = 1.0, 95% CI 0.7–1.5). However, GSTM1 null genotype was associated with increased risk of endometrioid/clear cell invasive cancer (age-adjusted OR = 2.0, 95% CI 1.0–4.1). Because only 32 cases with endometrioid cancer and 15 cases with clear cell carcinoma were included, the study had limited ability to look at associations within histologic subtypes of ovarian cancer. Evidence of (positive or inverse) associations with other histological types was not observed.

Associations with GSTT1 genotype have also been examined (Table 3). Sarhanis et al.52 found no association with GSTT1 genotype in their study of 81 patients and 325 controls. Hengstler et al.53 examined associations with GSTT1 genotype in their study of 103 patients with ovarian cancer and 115 controls. No associations were observed with the null GSTT1 genotype (OR = 0.9, 95% CI 0.4–1.9). GSTT1 genotype was unrelated to histological type, grade, or stage of ovarian cancer.

Table 3 Summary of studies of epithelial ovarian cancer and GSTT1
Full size table

Overall, Spurdle et al.24 found no association with GSTT1 genotype (age-adjusted OR = 1.1, 95% CI 0.7–1.6) or with GSTP1 genotype (results not shown) in their case-control study of epithelial ovarian cancer from Brisbane, Australia. However, there was a suggestion that ovarian cancers of the endometrioid or clear cell histological subtype had a higher frequency of the GSTT1 null genotype than other histological subgroups [P = 0.04 for comparison of endometrioid and clear cell subtypes (29% and 33%, respectively) versus serous subtype (17%)], but the differences were not statistically significant overall.

With respect to limitations of existing studies that should be overcome in future research, and gaps in the current epidemiologic knowledge base, only one study has looked for an association with GSTP1 genotype. In addition, reported studies of GSTM1, GSTT1, and GSTP1 and epithelial ovarian cancer have only been carried out among Caucasian populations. Studies that involve other racial and ethnic populations are needed. Furthermore, existing studies have often been based upon patient controls and hospital-based study designs. Studies with a population-based case-control study design are needed, because the latter design is more likely to be scientifically valid. Future studies of GSTM1, GSTT1, and GSTP1 and epithelial ovarian cancer should have a larger sample size with adequate numbers of incident cases and population controls, and they should be carried out in diverse populations. Gene-environment interactions, such as possible interactive effects between exogenous hormones and genotype, have not been examined in reported studies. Such interactions should be examined in future studies. A further gap in the existing knowledge base is that few studies have investigated the combined effect of polymorphic alleles in different genes on ovarian cancer risk, as discussed below.

Although results of studies carried out to date do not confirm associations between GSTM1, GSTT1, and GSTP1 and epithelial ovarian cancer, the associations are biologically plausible because glutathione S-transferases metabolize exogenous and endogenous substances that may have a role in ovarian cancer carcinogenesis.50,52 Studies that examine associations within specific histologic types of ovarian cancer are a promising area for further research.

INTERACTIONS

Spurdle et al.24 analyzed the combined effect of the GSTM1 and GSTT1 genotypes. The combination of the GSTM1 null and GSTT1 null genotypes resulted in an age-adjusted odds ratio of 1.1 (95% CI 0.6–2.0), compared with the GSTM1 positive and GSTT1 positive genotypes. Similarly, no association was observed with having all three genotypes hypothesized to increase risk of ovarian cancer (GSTM1 null, GSTT1 null, and GSTP1 GG/AG; age-adjusted OR = 0.8, 95% CI 0.4–1.9). In comparisons of endometrioid/clear cell cases and controls, however, the age-adjusted odds ratio was 3.9 (95% CI 1.4–11.1) for GSTM1 null combined with GSTT1 null (P = 0.01), compared with the GSTM1 positive and GSTT1 positive genotypes.

We did not identify any published studies that evaluated gene-environment interactions and risk of ovarian cancer.

LABORATORY TESTS

Molecular methods for determining GSTM1 and GSTT1 null genotype were summarized by Cotton et al.8 All of the studies reviewed here extracted genomic DNA from blood samples and genotyping methods used in these studies were consistent with standard techniques employed for PCR and PCR-restriction fragment length polymorphisms.

POPULATION TESTING

There is insufficient evidence associating GSTM1, GSTT1, or GSTP1 with ovarian cancer for population testing to be considered. More evidence is needed from molecular epidemiology studies to assess the causality of the associations with ovarian cancer.