We recruited 215 ovarian cancer patients diagnosed between Jan. 2004 and Jan. 2007 for this study. 218 Controls were randomly selected from female people who requested general health examinations in the same hospital during the same period. Controls were required to be without any history of any type of cancer and frequency matched by five-year age groups. The average ages of cases and controls were 47.2 ± 7.5 years and 47.6 ± 8.1 years, respectively (Table 1). There were no significant differences between cases and controls in age, drinking, tobacco use, menopausal status, or hormone replacement use. Ovarian cancer patients tended to have borne fewer children than had controls. Oral contraceptive use was associated with a lower risk of ovarian cancer, while ovarian cancer in first-degree relatives heightened cancer risk. Most of the cancer patients received chemotherapy; only 6.0% of patients received radiotherapy.

There was a significant difference in C677T polymorphisms between patients and controls (Table 2). Individuals with the TT genotype had an increased risk of developing ovarian cancer relative to those with the CC genotype (Adjusted OR = 2.43, 95% CI = 1.32–6.32). Individuals with high folate intake were less likely to develop ovarian cancer (OR = 0.54, 95% CI = 0.32–0.94).

Patients were followed from diagnosis until the end of Dec. 2010. Seven of the 215 patients were lost to follow-up due to migration, while the remaining 208 patients completed the study. The median follow-up time was 48.2 months (range: 2–72 months). A total of 106 patients (50.9%) died of ovarian cancer during the follow-up period (Table 3). The 677TT genotype was associated with poor survival (HR = 2.17, 95% CI = 1.20–4.79) (Figure 1). We found an inverse relationship between folate intake and risk of death, with an average survival time of 40.5 months for individuals with low folate intake, and 56.2 months for those with the highest folate consumption (Figure 2). The HRs (95% CI) for moderate and high folate intake were 0.67 (0.40–1.03) and 0.43 (0.33–0.88), respectively (p for trend = 0.011).

We further explored the association between MTHFR C677T polymorphisms and ovarian cancer prognosis according to folate intake level (Table 4). Among individuals with a folate intake of 230–300 μg/day and >300 μg/day, MTHFR 677CC carriers showed a significantly lower risk of death risk from ovarian cancer, with adjusted HRs (95% CI) of 0.56 (0.39–0.96) and 0.32 (0.27–0.82), respectively. A significant interaction was found between folate consumption and MTHFR genotype (p = 0.024, data not shown).

Our results indicate that MTHFR C677T polymorphisms are associated with susceptibility to ovarian cancer, and high folate consumption with a decreased risk of ovarian cancer. Moreover, ovarian cancer patients with high folate consumption may have better survival than individuals with low diet folate intake. This protective effect was more evident in patients carrying MTHFR 677CC genotypes. This may be due to the putative effect of folate metabolism on the development and prognosis of ovarian cancer.

Folate mediates the transfer of one-carbon moieties both in the synthesis of nucleotides necessary for DNA synthesis, replication, and repair, and in DNA methylation reactions [17]. These functions may play a critical role in carcinogenesis. An abundant intake of folate-rich foodstuffs appears to convey protection against developing some cancers [18]. Our study provides further evidence of the protective role of folate intake. Although previous epidemiological studies have found an inverse association between folate intake and cancer risk [19], the effect of folate supplements on carcinogenesis remains controversial for some ethnic groups [6,20]. The prevalence of variant genotypes of the MTHFR C677T polymorphism varies greatly among different human populations. In Africans, the frequency of MTHFR 677T/T is below 1%, whereas in Mexicans, it is above 30%. Its frequency in our control group was 8.8%, which is comparable to previous reports on Chinese populations [21,22]. The different results found for the effect of folate intake on ovarian cancer risk in different ethnicities may be due to differences in population background, study design, sample size, environmental factors, and chance variations. Further confirmation of such differential effects is therefore needed.

We found a significant association between MTHFR C677T polymorphism and ovarian cancer development and prognosis. However, previous studies have found inconsistent results for whether this association exists [13,23]. A study conducted in the United States and one Polish study both reported an association between these MTHFR SNPs and ovarian cancer development and prognosis [23,24], while another large sample study from the United States did not find evidence for an association between MTHFR C677T and ovarian cancer risk [13]. Some biological data support a role for MTHFR C677T polymorphism in cancer survival. Namely, the balance of the 5,10-MTHF and 5-MTHF controlled by MTHFR may influence the efficacy of chemotherapy with 5-fluorouracil (5-FU). The inactive 677TT form could increase the amount of 5,10-MTHF. Cell line experiments suggest that fluoropyrimidines like 5-FU inhibit thymidylate synthetase by forming a ternary complex involving 5-FU, thymidylate synthetase, and 5,10-MTHF [25]. Therefore, these MTHFR polymorphisms may enhance the effect of 5-FU by increasing the amount of 5,10-MTHF. In one animal study, administration of 5,10-MTHF enhanced the efficacy of 5-FU [26]. Another study showed that colon cancer cases with inactive MTHFR 677T/T are three times more likely to respond to 5-FU chemotherapy than are MTHFR 677C/C carriers [27]. In our study, most of the cases were treated with chemotherapy, and our results showed 677TT is associated with a higher risk of death from ovarian cancer, which indicates an interaction between MTHFR C677T polymorphism and chemotherapy.

Our study showed high folate intake was associated with reduced risk of ovarian cancer among 677C/C carriers, which indicates that folate and the active 677C/C genotype have a synergistic effect on ovarian cancer prognosis. This association between folate intake and reduced risk of death from ovarian cancer was only found for the 677C/C genotype. Similarly, a study conducted in China on esophageal cancer also found a significant synergy between folate intake and the 677C/C genotype [28].

Our study has several limitations. First, we selected controls from hospital visitors, which may be a threat to the validity of the results. This method brings a certain risk of selection bias, since the controls were not a random sample of the general population and may not fully represent the underlying base population. However, all control subjects in our study were those who came to the hospital for routine health examinations, not hospitalized patients with specific diseases, so they were likely representative of the general population, and thus provide a reasonable estimate of the allele frequencies of MTHFR C677T variants. Second, the small number of patients carrying the homozygous mutation TT in this study limited the statistical power of our analysis of differences among carriers of different genotypes. Increasing the number of controls to a control-case ratio of 2 or more can, at least to some extent, increase the study power. Therefore, further large-sample studies on this question are warranted. Third, collection of dietary habits information brings a risk of recall bias, which may have exaggerated or reduced the effects we found. We suggest that, for the survival analysis, future studies could collect information on dietary habits during follow-up, which would greatly reduce recall bias.

This was a case-control and case-only cohort study conducted in the Tianjin Medical University Cancer Institute and Hospital. All Chinese female cases newly diagnosed with primary ovarian cancer in the hospital between Jan. 2004 and Jan. 2007 were invited to face-to-face interviews within two months after diagnosis. All cases recruited in this study were histologically confirmed. A total of 215 cases were selected and interviewed. Controls were randomly selected from people who requested general health examinations in the same hospital during the same period. Controls had no history of any type of cancer, and were frequency-matched by 5-year age groups. Among a total of 246 eligible controls, 218 were successfully interviewed and donated blood samples, with a participation rate of 88.6%. All the patients were followed up until Dec. 2010, and we recorded all deaths of patients.

A face-to-face interview was conducted with a structured questionnaire to collect information on demographic factors and clinical characteristics, menstrual and reproductive history, hormone use, dietary habits, tobacco and alcohol habits, and family history of ovarian cancer. The tobacco smoking status were divided into smokers (smokers were defined as individuals who smoked more than 20 packets of cigarettes per year, or smoked more than one cigarette per day and continued for 6 months) and nonsmokers, drinking status were divided into drinkers (defined as individuals who drank more than 100 g alcohol (400 mL beers, 250 wine mL and 100 mL white spirit) per month and continued for 6 months) and non-drinkers. Information on usual dietary intake was collected using a comprehensive quantitative food-frequency questionnaire (FFQ) with 95 items. The FFQ used in this study was based on a questionnaire designed by the US National Institutes of Health and modified according to common Chinese food items and cooking habits. For each food item, we collected the frequency and quantity of consumption, then calculated daily intake by multiplying the reported consumption frequency for each food item by the portion size specified by the respondent. Soy food intake was estimated with 16 soy food items, including tofu, soy milk, soy yogurt, soy frozen yogurt, soy ice cream, soy cheese, soy hot dogs and cold cuts, other meat substitutes made from soy, tempeh, miso, soybeans, roasted soy nuts, soy sauce, soybean sprouts, alfalfa sprouts, and protein powder supplements made from soy.

The study protocol was approved by the Ethics Committee of the Affiliated Hospital of Sichuan University. Informed consent was obtained before each interview.

Genomic DNA was extracted from whole-blood samples using the Qiagen Blood Kit (Qiagen, Chatsworth, CA). Genotyping was conducted by TaqMan assays using the ABI Prism 7911HT Sequence Detection System (Applied Biosystems, Foster City, CA). Primer, probes, and reaction conditions were based on those of our previous study [29]. Genotyping was done by laboratory personnel blinded to case-control status. For quality control purposes, we also genotyped internal positive control samples, used no-template controls, and replicated the genotyping of 10% of the samples.

All statistical analyses were performed using STATA 10.0 (College Station, TX, USA). Differences in demographic and clinic characteristics, as well as potential confounding factors, were tested using a chi-squared test. Previous studies showed that family history of ovarian cancer, reproductive factors, menopausal status and hormone replacement therapy were the risk factors for ovarian cancer [30], and these factors may confound the association between MTHFR C677T polymorphism and ovarian cancer prognosis. Moreover, previous studies showed that there was interaction between MTHFR gene polymorphism and tobacco smoking and alcohol consumption [31–33]. Therefore, we adjusted the age, tobacco smoking, alcohol consumption, number of deliveries, menopausal status, hormone replacement, oral contraceptive use, and ovarian cancer history when analyzing the association between MTHFR C677T polymorphism and ovarian cancer prognosis. Unconditional logistic regression was undertaken to estimate odds ratios (ORs) and their 95% confidence intervals (95% CIs) after controlling for potential confounding factors. In the logistic regression equation, the age was continuous variable, and tobacco smoking, alcohol consumption, number of deliveries, menopausal status, hormone replacement, oral contraceptive use, and ovarian cancer history were categorical variables.

The outcome measured was overall survival, which was estimated using the Kaplan-Meir method. A univariate Cox regression analysis was used to assess the association between folate intake, MTHFR C677T-Arg399Gln gene polymorphisms, and survival. A primary death from ovarian cancer was defined as a failure event, and the survival time was defined as the time between diagnosis and death. The cause of death was defined by specialists based on clinical documents and reports by patients’ family members. If a patient died from a cause other than ovarian cancer, her data was censored at the date of death. All surviving patients were censored at the date of last follow-up. Folate intake was computed by multiplying food intake (in grams) by folate content (per gram) of the food, then summing all folate intake from various foods to obtain the total folate intake. The continuous variable of folate intake was categorized as low, moderate, or high using tertiles as cut-off points. The relative risk [hazard ratio (HR)] and 95% CI were calculated with the Cox regression model for all significant predictors from cancer diagnosis to the endpoint of the study (event). All statistical tests were two sided, and differences were taken as significant when the P value was less than 0.05.