Abstract
It has been proposed that high levels of galactose consumption increase the risk of ovarian cancer. Galactose levels are determined, in part, by the galactose-1-phosphate uridyl transferase gene (GALT). The N314D allele of the GALT gene has been associated with low GALT activity and with an increased risk of ovarian cancer. We screened for the presence of the N314D GALT allele in 891 incident cases of epithelial ovarian cancer and in 364 unaffected female controls. No significant difference in the prevalence of the N314D allele was observed between the cases (18.1%) and the controls (18.7%). The odds ratio associated with the presence of one N314D allele was 0.94 (95% confidence interval (CI), 0.68–1.3; P = 0.70), and the odds ratio associated with two N314D alleles was 1.62 (95% CI, 0.34–7.7; P = 0.54). Subanalyses of the cases by histological type, by age, by ethnic group, by family history, and by BRCA1/2 mutation status did not reveal any significant associations. We conclude that the GALT N314D allele does not predispose to epithelial ovarian cancer.
Introduction
The GALT2 gene codes for the GALT enzyme, which catalyzes the conversion of galactose-1-phosphate to uridyl diphosphate-galactose. An autosomal recessive deficiency of GALT activity leads to the accumulation of galactose-1-phosphate and is the basis for the metabolic defect in galactosemia (1). Women with galactosemia commonly have premature ovarian failure (2, 3, 4), and animal models have shown that high dietary galactose causes ovarian toxicity (5). On the basis of these observations, Cramer et al. (6, 7) hypothesized that women exposed to high galactose levels, either through consumption of dairy products or as a result of low GALT activity, might be at elevated risk for ovarian cancer. Two variants of the GALT gene, the Q188R mutation of classical galactosemia and the N314D allele (the Duarte variant), are associated with reduced GALT activity (8, 9). In classical galactosemia (e.g., with the Q188R mutation), the homozygote individual has no measurable GALT activity. Heterozygote carriers typically have 50% of normal activity. For homozygous carriers of the common Duarte N314D variant, GALT activities are ∼50% of normal, and for heterozygotes, the activity is ∼75% of normal (10). Association studies of these two GALT polymorphisms and ovarian cancer risk have reported both positive and negative results (11, 12, 13, 14, 15, 16). To investigate the relationship between the N314D polymorphism and ovarian cancer, we carried out a case-control study in Ontario, Canada.
Materials and Methods
Cases.
All patients in the province of Ontario, Canada, diagnosed with invasive and borderline epithelial ovarian tumors from January 1995 to December 1997 were identified through the Ontario Cancer Registry. The pathology report for each case was reviewed to confirm eligibility and to determine histological subtype. Each tumor was classified as borderline or invasive and was subclassified into one of the following five categories: endometrioid; clear cell; serous; mucinous; and undifferentiated/other. All patients were between 20 and 79 years of age and were resident in Ontario at the time of diagnosis. Cases were incident cases, but because of delay in ascertainment, up to 2 years may have elapsed between patient ascertainment and interview. Blood samples were obtained from 891 cases. The BRCA1 and BRCA2status of these cases has been determined previously (17).
Controls.
The control group consisted of 364 women without history of cancer. These women were invited to participate in a research study when they attended a comprehensive screening clinic for healthy women at Sunnybrook and Women’s College Hospital (Healthwatch) from 1997 to 1998. The Healthwatch Clinic offered screening for breast and cervical cancer, as well as other screening tests such as blood pressure, lipids and bone density, and provided information about diet and lifestyle. Attendees were invited to participate in a study of genetic factors and serum hormones but with no specific reference to cancer. Approximately 85% of invited women agreed to participate in the study. Women who attended this clinic were from throughout the Toronto region and from surrounding communities and were predominantly white. Controls were asked to complete a questionnaire detailing ethnic background and reproductive, medical, and family histories. The woman were asked to provide a urine sample for hormonal measurements and a blood sample for DNA analyses.
Detection of the N314DPolymorphism.
Lymphocyte DNA was extracted from whole blood specimens by standard procedures. PCR amplifications were performed in a total volume of 10 μl with primers specific for exon 6 of GALT(9). The PCR consists of one cycle of 3 min at 95°C, followed by 35 cycles of 95°C for 1 min, 57°C for 1 min, and 72°C for 1 min. The PCR products were digested with AvaII (New England Biolabs) overnight at 37°C and then analyzed on a 4% NuSieve 3:1 agarose gel (FMC).
Statistical Analysis.
The frequency of the N314Dallele was compared between cases and controls. Each subject was determined to be a carrier of zero, one, or two variant alleles. The observed N314Dgenotype distributions were compared with the expected numbers based on Hardy-Weinberg equilibrium. Odds ratios and corresponding 95% CIs were calculated for both heterozygous and homozygous states, with reference to women carrying two wild-type alleles. Subanalyses were performed by categorizing the cases according to age of diagnosis, family history of ovarian cancer, ethnic group, histological type of ovarian cancer, and presence of a BRCA1or BRCA2mutation. χ2 tests of statistical significance were performed, and all Ps are two-sided.
Results
The average age of the 891 cases was 56.5 years (range, 20–81 years), compared with an average of 57.7 years (range, 26–94 years) for the 364 controls. The average age of diagnosis of ovarian cancer in the cases was 55.5 years (range, 20–79 years). Among the cases, 161 (18.1%) carried at least one variant N314D allele, compared with 68 (18.7%) of the controls (Table 1). The frequency of the N314D allele was similar in the cases (9.5%) as in the controls (9.6%). The distribution of homozygotes and heterozygotes closely approximated the expected values under Hardy-Weinberg equilibrium. The OR associated with heterozygosity for the N314D allele was 0.94 (95% CI, 0.68–1.3) and for homozygosity was 1.62 (95% CI, 0.34–7.7).
Subanalyses of the ovarian cancer cases were conducted according age at diagnosis, ethnic group, histological type, family history of ovarian cancer, and the presence of a BRCA1 or BRCA2 mutation (Tables 2 and 3). Ethnicity data were obtained from 853 cases and 348 controls. The frequency of the N314D allele varied between ethnic groups, and there was some variation in the OR by ethnic group (Table 2). However, most of the ethnically defined subgroups were small, and no significant associations were seen (Table 2). The Mantel-Haenszel OR for heterozygosity, adjusted for ethnicity, was 0.91 (95% CI, 0.66–1.3).
If there were an association between the GALT N314D allele and a particular subgroup of ovarian cancer cases, we would expect to see the frequency of the GALT allele vary by subgroup. There were no appreciable differences in the prevalence of the GALT allele among subgroups defined by age of diagnosis, by ovarian cancer histology, by family history, or by BRCA mutation status (Table 3).
Discussion
We did not find a significant difference in the frequency of the GALT N314D allele between women with ovarian cancer and healthy controls. The prevalence of the N314D allele was, in fact, slightly higher in the controls (18.7%) than in the cases (18.1%). Previous association studies between GALT variant alleles and ovarian cancer have at most provided modest support for this hypothesis or have been negative (11, 12, 13, 14, 15, 16). The most recent study by Cramer et al. (11) did not confirm an association of adult lactose consumption or of GALT enzyme activity with ovarian cancer. Cramer et al. (11) did report a significant difference in the genotype distribution of GALT N314D between patients with endometrioid or clear cell cancer and controls. In this subgroup, the OR associated with heterozygosity was 1.40 (95% CI, 0.84–2.3) and with homozygosity was 14.2 (95% CI, 2.6–77). In contrast, Morland et al. (13) suggested that the association was present only for the serous type of ovarian cancer. Other studies have found no associations among the various histological subgroups (15, 16). We also found no significant relationship between the GALT N314D allele and ovarian cancer of any histological subtype.
We did not see a significant association between GALT N314D homozygosity and ovarian cancer risk (OR = 1.7; 95% CI, 0.34–7.7), but homozygotes were rare, and the power to detect a significant association (P < 0.05) at OR of 2.0 or greater was only 16%. Webb et al. (12) reported an association between ovarian cancer and GALT activity but only among women with abnormally low GALT activity. In this study, neither GALT activity or galactose consumption was measured, and it was not possible to evaluate the association in these subgroups.
The N314D allele is present in ∼20% of the North American population and is potentially of importance in ovarian cancer etiology. Our observed frequency of the N314D allele in the cases from Ontario (0.095) was slightly lower than that reported by Morland et al. (0.143; Ref. 13) or by Cramer et al. (0.114; Ref. 11). Samples from Morland et al. (13) were taken from patients in the United Kingdom, whereas those from Cramer et al. (11), Goodman et al. (15), and Cozen et al. (16) were taken from patients in the United States.
In this study, the controls were different from the cases in that controls lived in the greater Toronto region, whereas the cases were drawn from throughout the province. To control for possible differences in the ethnic composition for cases and controls, ethnicity-specific comparisons were made. Ethnic differences might have resulted in differences in allele frequencies between the studies. In our study, analyzing the data controlling for ethnic group did not modify our conclusions. Factors other than ethnic group (e.g., age) should not influence the likelihood that a variant GALT allele is present (Table 2) and could not explain our negative results. In summary, in a large population-based study of unselected incident cases of ovarian cancer, no association was found between the N314D polymorphism of GALT and ovarian cancer.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: GALT, galactose-1-phosphate uridyl transferase; CI, confidence interval; OR, odds ratio.
Genotype . | Cases n (%) . | Controls n (%) . | OR (95% CI) . | P . |
---|---|---|---|---|
−/− | 730 (81.9%) | 296 (81.3%) | 1.00 | |
+/− | 153 (17.2%) | 66 (18.1%) | 0.94 (0.68–1.3) | 0.70 |
+/+ | 8 (0.9%) | 2 (0.6%) | 1.62 (0.34–7.7) | 0.54 |
Genotype . | Cases n (%) . | Controls n (%) . | OR (95% CI) . | P . |
---|---|---|---|---|
−/− | 730 (81.9%) | 296 (81.3%) | 1.00 | |
+/− | 153 (17.2%) | 66 (18.1%) | 0.94 (0.68–1.3) | 0.70 |
+/+ | 8 (0.9%) | 2 (0.6%) | 1.62 (0.34–7.7) | 0.54 |
Ethnic group . | Cases . | . | Controls . | . | OR (95% CI) . | P . | ||
---|---|---|---|---|---|---|---|---|
. | Total . | N314D . | Total . | N314D . | . | . | ||
French-Canadian | 80 | 15 (18.8%) | 11 | 4 (36.4%) | 0.40 (0.10–1.59) | 0.18 | ||
Jewish | 20 | 2 (10.0%) | 2 | 1 (50.0%) | 0.11 (0.00–3.34) | 0.12 | ||
Other white | 692 | 124 (17.9%) | 315 | 61 (19.4%) | 0.91 (0.65–1.28) | 0.58 | ||
Asian | 42 | 9 (21.4%) | 4 | 0 (0%) | 0.31 | |||
Othera | 19 | 3 (15.8%) | 16 | 0 (0%) | 0.10 |
Ethnic group . | Cases . | . | Controls . | . | OR (95% CI) . | P . | ||
---|---|---|---|---|---|---|---|---|
. | Total . | N314D . | Total . | N314D . | . | . | ||
French-Canadian | 80 | 15 (18.8%) | 11 | 4 (36.4%) | 0.40 (0.10–1.59) | 0.18 | ||
Jewish | 20 | 2 (10.0%) | 2 | 1 (50.0%) | 0.11 (0.00–3.34) | 0.12 | ||
Other white | 692 | 124 (17.9%) | 315 | 61 (19.4%) | 0.91 (0.65–1.28) | 0.58 | ||
Asian | 42 | 9 (21.4%) | 4 | 0 (0%) | 0.31 | |||
Othera | 19 | 3 (15.8%) | 16 | 0 (0%) | 0.10 |
Other includes black, Hispanic, native Canadian, and mixed ethnicity.
. | Total . | No with N314D variant . | Prevalence . |
---|---|---|---|
By age of diagnosis (yr) | |||
<40 | 115 | 23 | 20.0% |
40–49 | 185 | 37 | 20.0% |
50–59 | 235 | 39 | 16.6% |
60–69 | 208 | 37 | 17.8% |
70+ | 148 | 25 | 16.9% |
All | 891 | 161 | 18.1% |
By histology | |||
Invasive Serous | 399 | 64 | 16.0% |
Endometrioid | 131 | 21 | 16.0% |
Clear cell | 40 | 8 | 20.0% |
Mucinous | 70 | 16 | 22.9% |
Other | 59 | 14 | 23.7% |
Total invasive | 699 | 123 | 17.6% |
Borderline | 192 | 38 | 19.8% |
By family history | |||
Family history | |||
No affected relative | 826 | 148 | 17.9% |
One or more affected relative | 47 | 9 | 19.1% |
BRCA mutation status | |||
BRCA1 mutation | 42 | 7 | 16.7% |
BRCA2 mutation | 33 | 4 | 12.1% |
No mutation | 816 | 150 | 18.4% |
. | Total . | No with N314D variant . | Prevalence . |
---|---|---|---|
By age of diagnosis (yr) | |||
<40 | 115 | 23 | 20.0% |
40–49 | 185 | 37 | 20.0% |
50–59 | 235 | 39 | 16.6% |
60–69 | 208 | 37 | 17.8% |
70+ | 148 | 25 | 16.9% |
All | 891 | 161 | 18.1% |
By histology | |||
Invasive Serous | 399 | 64 | 16.0% |
Endometrioid | 131 | 21 | 16.0% |
Clear cell | 40 | 8 | 20.0% |
Mucinous | 70 | 16 | 22.9% |
Other | 59 | 14 | 23.7% |
Total invasive | 699 | 123 | 17.6% |
Borderline | 192 | 38 | 19.8% |
By family history | |||
Family history | |||
No affected relative | 826 | 148 | 17.9% |
One or more affected relative | 47 | 9 | 19.1% |
BRCA mutation status | |||
BRCA1 mutation | 42 | 7 | 16.7% |
BRCA2 mutation | 33 | 4 | 12.1% |
No mutation | 816 | 150 | 18.4% |