Dairy foods and nutrients in relation to risk of ovarian cancer and major histological subtypes
Inconsistent results for the role of dairy food intake in relation to ovarian cancer risk may reflect the potential adverse effects of lactose, which has been hypothesized to increase gonadotropin levels, and the beneficial antiproliferative effects of calcium and vitamin D. Using data from the New England case–control study (1,909 cases and 1,989 controls), we examined dairy foods and nutrients in relation to risk of ovarian cancer overall, histological subtypes and rapidly fatal versus less aggressive disease. We used logistic regression and polytomous logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CIs). In models that were simultaneously adjusted for total (dietary plus supplements) calcium, total vitamin D and lactose, we observed a decreased overall risk of ovarian cancer with high intake of total calcium [Quartile 4 (Q4, >1,319 mg/day) vs. Quartile 1 (Q1, <655 mg/day), OR = 0.62, 95% CI = 0.49–0.79]; the inverse association was strongest for serous borderline and mucinous tumors. High intake of total vitamin D was not associated overall with ovarian cancer risk, but was inversely associated with risk of serous borderline (Q4, >559 IU/day vs. Q1, <164 IU/day, OR = 0.51, 95% CI = 0.34–0.76) and endometrioid tumors (Q4 vs. Q1, OR = 0.55, 95% CI = 0.39–0.80). We found no evidence that lactose intake influenced ovarian cancer risk or that risk varied by tumor aggressiveness in the analyses of intake of dairy foods and nutrients. The overall inverse association with high intake of calcium and the inverse associations of calcium and vitamin D with specific histological subtypes warrant further investigation.
Differences in ovarian cancer incidence rates worldwide1 suggest that lifestyle factors, including diet, may play an important role in risk for this disease. The observation that per capita milk consumption and lactase persistence (the ability to digest lactose after childhood) is significantly positively correlated with ovarian cancer incidence worldwide highlights the potential role of dairy food consumption in ovarian carcinogenesis.2 It has been hypothesized that the galactose component of dairy sugar or lactose might have a toxic effect on oocytes and prematurely raise gonadotropins in a similar manner to that observed in mouse/rat models in which high-lactose diets led to ovulatory dysfunction and hypogonadism.3–5
Subsequent epidemiological studies that evaluated consumption of dairy foods or lactose in relation to ovarian cancer risk produced conflicting results with case–control studies generally finding a null association,6 whereas cohort studies showed a more consistent positive association between high intake of lactose6–8 and/or skim/low-fat milk7 or milk (all types)6 and risk of ovarian cancer.
Inconsistencies could be due to the contrasts of intake that were examined, potential differences in the distribution of histological subtypes across studies or may reflect differences in long-term versus recent dairy food consumption or in the types of dairy products consumed and their different nutrient contents. We hypothesized that foods higher in lactose or fats could have a more harmful effect, whereas foods with higher levels of calcium or vitamin D might be beneficial by downregulation of circulating parathyroid hormone leading to decreased cell proliferation.9, 10 In this large population-based case–control study of ovarian cancer, we evaluated intake of dairy foods and their components (lactose, calcium and vitamin D) in relation to ovarian cancer risk overall, the main histological subtypes and rapidly fatal versus less aggressive disease.
Material and Methods
Details regarding case and control enrollment in the New England case–control (NECC) study were described previously.11, 12 Briefly, 3,957 women residing in eastern Massachusetts or New Hampshire with a diagnosis of incident ovarian cancer were identified through hospital tumor boards and statewide cancer registries. Of these, 3,083 (78%) cases met eligibility criteria and 2,203 (71%) were enrolled. This analysis was restricted to 2,076 cases with epithelial tumors. Controls were identified through random digit dialing, drivers' license lists and town resident lists. In the first study phase (1992–1997), 420 (72%) and 102 (51%) of the eligible controls identified through random digit dialing and town resident lists, respectively, agreed to participate. In the second (1998–2003) and third (2003–2008) phases, 4,366 potential controls were identified; of whom, 2,940 (67%) were eligible; 1,362 (46%) declined to participate by phone or by mail via an “opt-out” postcard and 1,578 (54%) were enrolled. Controls were frequency matched to cases on age and state of residence. Study participants were interviewed in-person at the time of enrollment about known and putative ovarian cancer risk factors that occurred at least 1 year before diagnosis (for cases) or enrollment (controls). Institutional review boards at the Brigham and Women's Hospital and the Dartmouth Medical School approved the study, and all participants provided written informed consent.
Tumor histological subtype and behavior (invasive vs. borderline) were abstracted from pathology reports that had been reviewed by a gynecologic pathologist. We examined five major histological subtypes of ovarian cancer: serous borderline or invasive, mucinous, endometrioid and clear cell tumors. We also evaluated invasive ovarian cancer cases by tumor aggressiveness (rapidly fatal and less aggressive) based on the time interval between diagnosis and death reported in the National Death Index. Rapidly fatal cases were those who died due to any cause within 3 years (the cause of death was unavailable), and less aggressive cases were those who died >3 years postdiagnosis or those who were alive at follow-up.
Dietary intake that occurred at least 1 year before diagnosis or study enrollment was assessed at the time of enrollment using a semiquantitative food frequency questionnaire (FFQ),13, 14 which has been previously shown to provide valid estimates of skim/low-fat milk, whole milk and yogurt intake with correlation coefficients between the dietary questionnaire and the 1-week diet records of 0.81, 0.62 and 0.94, respectively.15 For calcium, correlation coefficients of 0.56 with supplements and 0.51 without supplements were reported between the dietary questionnaire and 1-week diet records.16 For vitamin D, reported intake via the dietary questionnaire and plasma concentrations of 25-OH vitamin D were moderately correlated (correlation coefficients of 0.35 and 0.25 with and without supplements, respectively).17
Specific dairy foods that were assessed for our study included skim/low-fat milk, whole milk, cream, sour cream, nondairy coffee whitener, sherbet/ice milk, ice cream, yogurt, cottage/ricotta cheese, other cheeses and butter. Additional food items that are important sources of calcium, including broccoli and leafy green vegetables, were also included in the FFQ. The questionnaire included questions on supplemental vitamin usage; for multivitamins, respondents were asked about the number of pills per week and the specific brand, and for single supplements, respondents were asked about the dosage per day and the duration of use. Dietary nutrient intake (lactose, calcium and vitamin D) was calculated by multiplying the frequency of intake of each food containing the nutrient by the nutrient content of specified portions determined from the food composition values available from the US Department of Agriculture food composition data.18 To provide information about the primary dietary sources of each nutrient, we confirmed that there were similar frequencies of intake of dairy foods in the NECC and the Nurses' Health Study (NHS), which is also composed of US residents, and diet was assessed over a similar time period.7 In the NHS in 2002, the primary dietary sources of lactose were skim/low-fat milk (63%) and yogurt (15%); the main dietary sources of calcium were skim/low-fat milk (29%), fortified orange juice (8%) and hard cheeses (7%) and the major dietary sources of vitamin D were fortified skim/low-fat milk (41%) and fish (32%) (S. Tworoger, personal communication). Total (dietary and supplemental) nutrient intake was calculated by summing the contributions of the nutrient from dietary, multivitamin and single supplement sources. Intakes of total and saturated fat were calculated as the sum of the contributions from all foods based on the U.S. Department of Agriculture data and also included margarine and fats used in cooking or baking.
Individuals were excluded from the analysis if they did not complete a FFQ (n = 134) or if they had an implausibly low or high caloric intake (<500 or >3,500 kcal/day; n = 143). Energy-adjusted nutrient intake was calculated using the residuals from the regression of nutrient intake based on a total caloric intake of 1,600 kcal/day.19 Nutrient quartile cutpoints were calculated based on the distribution of intake among controls. Unconditional logistic regression was used to estimate the odds ratio (OR) and 95% confidence intervals (CIs) for the main effects of dairy foods and nutrients. All multivariable analyses were adjusted for age (continuous), number of pregnancies (continuous), oral contraceptive (OC) use (0, >3 months to 2 years, >2 to 5 years, >5 to 10 years, >10 years), tubal ligation (yes vs. no), family history of ovarian cancer in a first-degree relative (yes vs. no), study center (MA, NH), study phase (3, 4, 5) and energy intake (continuous). Analyses of lactose intolerance were additionally adjusted for race. To simultaneously evaluate the effects of lactose, total calcium and total vitamin D, we included all three terms in the same regression model. Additional potential confounders were evaluated but not included in the final models because they did not substantially alter the risk estimates. We calculated Pearson's correlation coefficients (r) to assess colinearity between continuous nutrient variables. Tests for linear trend were performed using the Wald test with a trend variable based on the median number of servings per day for each category of dairy food intake or using a trend variable based on the median value for each quartile (for dairy nutrients). We assessed effect modification by age (< 50, ≥ 50 years), menopausal status (unknown or premenopause, postmenopause), body mass index (BMI; < 25, ≥ 25), OC use (ever/never), parity and median total fat intake (high, low). The p for interaction was calculated using a likelihood ratio test to compare models with and without interaction terms.
Polytomous logistic regression (PLR) was used to simultaneously estimate separate risk factor associations across (i) histological subtypes and (ii) rapidly fatal versus less aggressive invasive ovarian cancers. For PLR, the likelihood ratio test was used to calculate a p-value for heterogeneity comparing a model in which all of the associations were held constant between the case subgroups to a model that allowed only the association of interest to differ between the case subgroups.20 In all statistical analyses, a p-value < 0.05 was considered as significant. Analyses were performed using SAS v9.2 (SAS Institute, Cary, NC) and Stata v9 (StataCorp, College Station, TX).
The final study population included 1,909 women with epithelial ovarian cancer (invasive and borderline) and 1,989 controls. Cases reported a lower proportion and shorter duration of OC use, were less likely to be parous or to have had a tubal ligation and were more likely to report a family history of ovarian cancer (Table 1). Consumption of selected dairy products (yogurt, cottage/ricotta cheese and hard cheeses) was significant and inversely related to risk of ovarian cancer, and all showed a trend (p ≤ 0.02) of decreasing risk with increasing intake (Table 2, Model 1). Increased consumption of skim/low-fat milk was also inversely associated with ovarian cancer risk, although only an intermediate category of intake was statistically significant (2–7 times per week vs. never/less than monthly, OR = 0.83, 95% CI = 0.70–0.97, ptrend = 0.28). In contrast, we observed an increased risk of ovarian cancer with high intake of whole milk (≥2 times per week vs. never/less than monthly, OR = 1.43, 95% CI = 1.15–1.78, ptrend = 0.002) or cream cheese (≥2 times per week vs. never/< monthly, OR = 1.39, 95% CI = 1.09–1.77, ptrend = 0.008). We observed a nonsignificant positive association between consumption of ice cream and ovarian cancer risk. In additional analyses of dairy foods with a higher fat content (whole milk, hard cheeses, cottage/ricotta cheese, ice cream and cream cheese), additionally adjusting for quartiles of total fat intake (Table 2, Model 2) only minimally altered the risk estimates. Quartiles of total fat and saturated fat intake were not associated with risk of ovarian cancer overall (data not shown).
Table 1. Descriptive characteristics of ovarian cancer cases and controls in the New England case–control (NECC) study
Table 2. Adjusted odds ratios (ORs) for epithelial ovarian cancer associated with frequency of consumption of dairy foods
The energy-adjusted intake of both dietary and total calcium and vitamin D was higher in controls; however, lactose intake was similar in cases and controls (Table 1). We observed a statistically significant inverse association with ovarian cancer risk with higher intake of both dietary and total (dietary plus supplements) calcium and vitamin D (Table 3, Model 1). The protective effects for intake of total calcium [Quartile 4 (Q4) vs. Q1, OR = 0.61, 95% CI = 0.50–0.74, ptrend < 0.001] and total vitamin D (Q4 vs. Q1, OR = 0.76, 95% CI = 0.63–0.92, ptrend = 0.01) were slightly stronger than for dietary intake of either nutrient. We observed a nonsignificant inverse association for high intake of lactose with ovarian cancer risk (Q4 vs. Q1, OR = 0.88, 95% CI = 0.73–1.05, ptrend = 0.07). Dietary intake of calcium and lactose were strongly correlated (r = 0.89) as was dietary intake of vitamin D with lactose or calcium (both correlations r = 0.54). When all three nutrients were included in the same multivariable model (Table 3, Model 2), we observed a similar inverse association for total calcium intake and risk of ovarian cancer (Q4 vs. Q1, OR = 0.62, 95% CI = 0.49–0.79, ptrend < 0.001); however, the OR for the highest compared to the lowest quartile of total vitamin D intake was attenuated (Q4 vs. Q1, OR = 0.93, 95% CI = 0.74–1.16, ptrend = 0.75).
Table 3. Adjusted odds ratios (ORs) for epithelial ovarian cancer associated with intake of dairy nutrients
We evaluated the association of intake of total calcium, total vitamin D and lactose with ovarian cancer risk among participants who were <50 years or ≥50 years. For participants who were <50 years, any category of calcium intake that was higher than the referent group had a significantly protective effect (e.g., Q4 vs. Q1, OR = 0.56, 95% CI = 0.39–0.80, ptrend = 0.002). Among participants who were ≥50 years, only the highest quartile of calcium intake was significantly protective (Q4 vs. Q1, OR = 0.69, 95% CI = 0.54–0.87, ptrend < 0.001) (pint = 0.01; Table 4). Associations were similar for total vitamin D intake with ovarian cancer risk across both age groups (pint = 0.41). For lactose intake among participants who were <50 years, we observed a statistically significant inverse association with ovarian cancer risk for any category of intake that was higher than the referent group (e.g., Q4 vs. Q1, OR = 0.70, 95% CI = 0.52–0.95, ptrend = 0.04); however, there was no association between lactose intake and ovarian cancer risk among participants who were ≥50 years (pint = 0.01). We observed similar results to those reported using the age of 50 years cutoff in the analyses stratified by menopausal status (pre/dubious menopause vs. postmenopausal; data not shown). In the analyses of dairy food and nutrient intake, similar estimates to the overall findings were observed when participants were stratified by parity, OC use, BMI and median total fat or saturated fat intake (data not shown).
Table 4. Associations between calcium and lactose intake with ovarian cancer risk are influenced by age
In an earlier report, we observed an excess of controls who reported lactose intolerance beginning at an early age (<20 years). Consistent with this finding, we evaluated the association with lactose intolerance beginning before the age of 20 years and found a suggestive inverse association with ovarian cancer risk (OR = 0.72, 95% CI = 0.42–1.22). We did not observe any association with ovarian cancer risk for individuals who reported lactose intolerance beginning at or after the age of 20 years (OR = 0.96, 95% CI = 0.75–1.22; data not shown). As a crude measure of lactose intolerance, we also asked respondents about their use of lactase enzyme tablets/lactose-free dairy products and observed a nonsignificant inverse association with reported use of these products (OR = 0.87, 95% CI = 0.67–1.12).
We identified significant differences in the risk associations for intake of total calcium and total vitamin D across the histological subtypes [using PLR combined with the likelihood ratio test to calculate a p-value for heterogeneity, phet (model) ≤ 0.04; Table 5]. The strongest inverse associations for total calcium intake were observed in serous borderline (Q4 vs. Q1, OR = 0.41, 95% CI = 0.27–0.62, ptrend < 0.001) and mucinous tumors (Q4 vs. Q1, OR = 0.46, 95% CI = 0.29–0.72, ptrend < 0.001). The ORs associated with total calcium intake remained essentially unchanged after additional adjustment for total vitamin D and lactose (data not shown).
Table 5. Adjusted odds ratios (ORs) for ovarian cancer associated with consumption of dairy nutrients by histologic subtype
The strongest inverse associations for total vitamin D intake were observed in the serous borderline (Q4 vs. Q1, OR = 0.51, 95% CI = 0.34–0.76, ptrend = 0.001) and endometrioid tumors (Q4 vs. Q1, OR = 0.55, 95% CI = 0.39–0.80, ptrend = 0.002) (Table 5). In the model that was mutually adjusted for all three nutrients, risk estimates for total vitamin D intake were slightly attenuated but remained significantly protective for serous borderline (Q4 vs. Q1, OR = 0.61, 95% CI = 0.40–0.93, ptrend = 0.02) and endometrioid tumors (Q4 vs. Q1, OR = 0.66, 95% CI = 0.45–0.97, ptrend = 0.046) (data not shown). We observed no difference in the associations between lactose intake and risk of the different histological subtypes of ovarian cancer [phet (model) = 0.66]. In the analyses assessing intake of individual dairy foods, ORs for the histological subtypes resembled those for ovarian cancer risk overall (data not shown).
Based on previous suggestions that high intake of dairy-related factors was modestly associated with poorer survival, we evaluated whether the risk associations with dairy food and/or nutrient consumption varied between rapidly fatal and less aggressive invasive cases compared to controls using the time interval between the date of diagnosis to the date of death to define rapidly fatal (died within 3 years of diagnosis) and less aggressive (died >3 years postdiagnosis or still alive) cases. We observed no statistically significant differences in risk by tumor aggressiveness (rapidly fatal vs. less aggressive invasive tumors) in the analyses of intake of dairy foods and nutrients (data not shown). Further adjustment for tumor stage made no difference to the risk associations. In the subset of cases for which detailed chemotherapy data were available (n = 946), the majority (90% and 84%) of the rapidly fatal and less aggressive cases, respectively, were treated with a cisplatin and/or carboplatin regimen (data not shown).
In this updated analysis of the NECC21 (including 1,909 cases and 1,989 controls), we evaluated the association of intake of dairy foods and nutrients (including calcium and vitamin D) in relation to ovarian cancer risk overall and assessed whether these associations differed among the histological subtypes of ovarian cancer. Consistent with previous case–control studies,9, 22–24 we observed that high intake of skim/low-fat milk was significantly inversely associated with ovarian cancer risk. Among three additional case–control studies that evaluated consumption of any type of milk, two studies reported an inverse association between high milk intake and ovarian cancer risk,25, 26 and the third study found no association.27 In contrast, three cohort studies reported a suggestive but nonsignificant increased risk of invasive ovarian cancer with high consumption of skim/low-fat milk7, 28 or any type of milk,29 whereas another cohort study30 and a pooled analysis of 12 cohort studies8 found no association with intake of any type of milk. In separate analyses of whole milk, we observed that high intake was associated with significantly increased risk of ovarian cancer. This finding is consistent with previous case–control studies23, 24 and some,7 but not all,8, 31 cohort studies.
We found that high intake of yogurt and ricotta/cottage cheese was significantly inversely associated with ovarian cancer risk. In contrast, previous case–control and cohort studies reported no association between consumption of yogurt and/or ricotta/cottage cheese and ovarian cancer risk,7–9, 22, 29, 30 and one study23 reported an elevated risk with increasing consumption of full-fat (but not low fat) yogurt. We were unable to distinguish low-fat yogurt from full-fat yogurt in our study. We observed that high consumption of hard cheeses was significantly inversely associated with risk of ovarian cancer. This result is consistent with the NHS,7 but not with other studies that reported no association8, 9, 28–30 or an increased risk23 with high cheese intake. To our knowledge, this was the first report to find that high intake of cream cheese was associated with increased ovarian cancer risk. The increased risk that we observed for selected high-fat foods was unlikely to be due to their fat content, because total fat and saturated fat intake alone were not associated with ovarian cancer risk in our study. There have been conflicting findings regarding the association between total fat intake and ovarian cancer risk; a meta-analysis of seven case–control and one cohort study32 and a recent report from the NIH-AARP Diet and Health Study33 found significantly increased risk with high total fat intake, whereas a pooled analysis of 12 cohort studies found no association with total fat intake.34
A component of dairy foods that has been hypothesized to increase ovarian cancer risk is lactose (and its metabolite galactose) through toxic effects on the ovarian germ cells leading to subsequent gonadotropin stimulation of the ovaries.35 In support of this, most,7, 8, 28, 29 but not all,30, 31 cohort studies reported a positive association between lactose intake and ovarian cancer risk, and two of these studies reported a stronger association for serous invasive tumors.7, 29 In contrast, case–control studies, including the current report, have found little evidence for an association between lactose intake and ovarian cancer risk,6, 21, 23, 36–39 and one study reported a significant inverse association.9 We did not observe differences in the association with lactose intake across the histological subtypes.
It has been suggested that differences in the association between lactose intake and ovarian cancer risk in case–control and cohort studies may relate to the time periods measured by the dietary questionnaire (i.e., recent diet in case–control studies versus long-term or baseline diet in cohort studies).6 Although we were unable to evaluate diet in an earlier life period in our study, we observed a nonsignificant but suggestive inverse association with ovarian cancer risk among individuals who reported symptoms of lactose intolerance that began before the age of 20 years and did not observe an association if symptoms began at or after the age of 20 years. Considering that lactose-intolerant women are likely to consume less lactose, our finding suggests that decreased lactose intake early in life may reduce ovarian cancer risk; however, further studies are needed to confirm this finding.
We also evaluated intake of calcium and vitamin D in relation to ovarian cancer risk and identified a significant inverse association and dose response with higher dietary or total (diet plus supplements) intake of either nutrient. We observed a stronger protective effect for total calcium and total vitamin D compared to dietary intake of either nutrient alone and attributed this to the higher intake among individuals who use supplements. Because of the high correlations observed for these dairy nutrients, it was difficult to identify independent associations with ovarian cancer risk. Nevertheless, in models that were mutually adjusted for total calcium, total vitamin D and lactose intake, the statistically significant inverse association with total calcium intake remained unchanged, whereas the association with total vitamin D intake was attenuated and no longer significant. Findings of an inverse association between dietary and/or total calcium intake and ovarian cancer risk are consistent with some,9, 31 but not all,28 previous studies, and a pooled analysis of cohort studies found no association.8 We also observed that the association with calcium intake was influenced by the participant's age. For women aged ≥50 years, only the highest quartile of calcium intake was significantly inversely associated with ovarian cancer risk, whereas any intake higher than the referent category was protective for ovarian cancer in younger (<50 years) women, possibly due to the role of estrogens in enhancing intestinal calcium absorption.40
In our data, total calcium showed a strong inverse association with serous borderline and mucinous tumors. Higher levels of calcium intake have been postulated to reduce ovarian cancer risk by inhibiting the release of parathyroid hormone, which may have protumorigenic effects by stimulating IGF-1.9, 10 Our observation that mucinous tumors showed a strong inverse association with total calcium intake was of interest because intestinal-type mucinous ovarian tumors have been recognized,41 and recent studies have suggested that advanced stage of ovarian mucinous carcinomas could be metastases from the colorectum42 and the protective effects for colorectal cancer with increased calcium intake are well established.43 Altogether, these observations suggest that calcium intake may influence the risk of developing a mucinous neoplasm regardless of the primary tumor site; however, these results require further confirmation.
We found that high intake of total vitamin D was inversely associated with risk of serous borderline and endometrioid tumors. Complementary to our results, a pooled analysis of four studies (including the current study) found that the VDR Fok1 polymorphism, leading to reduced functionality of the VDR, was strongly associated with increased risk of serous borderline and endometrioid tumors.44 However, a recent pooled analysis of five additional studies observed an increased risk of ovarian cancer (all types) among carriers of the variant VDR Fok1 allele, but they did not observe a difference in the association by tumor histology.45 In contrast to our findings, the pooled analysis of cohort studies reported a nonsignificant positive association between total vitamin D intake and risk of endometrioid ovarian cancer.8 However, we speculated that the inverse association between vitamin D intake and endometrioid ovarian cancer risk is biologically plausible based on the known increased risk for endometrioid ovarian tumors with a higher BMI.46, 47 A recent pooled analysis found a stronger inverse association between circulating 25-hydroxyvitamin D levels and ovarian cancer risk among women with a BMI ≥ 25.48 Further studies are needed to evaluate whether a high vitamin D intake may be important for the prevention of endometrioid ovarian tumors among women with a higher BMI.
Potential limitations of our study include potential differences in the reported diet due to disease status and the highly correlated nature of dairy-related nutrients. However, we observed an increased ovarian cancer risk with high intake of certain dairy foods (whole milk and cream cheese) and a decreased risk with others (including consumption of skim/low-fat milk and yogurt), indicating that cases did not consistently under-report or over-report their dairy food intake. The intake levels of dairy nutrients were highly correlated, which can be problematic in attempts to isolate their mechanistic effects, particularly when lactose is postulated to increase ovarian cancer risk, whereas calcium and vitamin D were inversely associated with risk. To address this issue, we evaluated the dairy nutrients simultaneously in the same model; however, even using this approach, it is possible that residual confounding could explain some of the observed associations. Lastly, we cannot exclude the possibility that components of dairy foods other than lactose, vitamin D or calcium might explain the observed associations with ovarian cancer risk; however, these dairy nutrients were selected a priori based on their potential role in ovarian carcinogenesis.
In conclusion, we found that high intake of certain dairy foods was inversely associated with ovarian cancer risk (skim/low-fat milk, yogurt and cheeses), whereas increased consumption of other foods (whole milk and cream cheese) was associated with increased risk. The analyses of dairy nutrients identified a significant inverse association with high intake of total calcium (dietary plus supplements) for ovarian cancer risk overall. Furthermore, increased intake of total calcium was significantly inversely associated with risk of serous borderline and mucinous tumors, whereas high intake of total vitamin D was significantly protective for serous borderline and endometrioid tumors. Future studies are needed to confirm the role of calcium and vitamin D in the prevention of these specific histological subtypes of ovarian cancer. If these findings are confirmed in future studies, increasing intakes of calcium and/or vitamin D may be important factors to consider for the prevention of epithelial ovarian cancer.
The authors thank Mary De Pari and Cameron Fraer for their assistance in all aspects of this study. They are also grateful to the participants of the NECC study for their contribution to this research.