In this large, pooled analysis of 7 prospective cohort studies, circulating 25(OH)D concentrations were not associated with ovarian cancer risk overall or in analyses stratified by tumor subtype, age at blood draw, or oral contraceptive use. However, stratified analyses by body mass index suggested a possible inverse association between circulating vitamin D and ovarian cancer risk among overweight and obese women.
Despite the biologic plausibility, results from this pooled analysis did not support an association between circulating 25(OH)D and ovarian cancer risk overall. These results are consistent with those from other studies to date (29
). In a pooled analysis of the Nurses’ Health Study, the Nurses’ Health Study II, and the Women's Health Study by Tworoger et al. (29
), fourths of 25(OH)D or 1,25(OH)2
D were not significantly associated with ovarian cancer risk overall. Both our study and the study by Tworoger et al. (29
) included the same Nurses’ Health Study participants (about 25% of the cases in the current study). However, excluding Nurses’ Health Study cases and controls from our study did not materially alter the results obtained from all studies combined. Similarly, overall null associations were observed between serum 25(OH)D and ovarian cancer risk in a nested case-control study in the Finnish Maternity Cohort (30
). Serum or plasma 25(OH)D was not associated with ovarian cancer risk in a case-control study nested within the New York University Women's Health Study and the Northern Sweden Health and Disease Study, nor was there evidence of an interaction with vitamin D receptor polymorphisms or haplotypes (31
). Excluding subjects (18% of cases in the current pooled analysis) did not alter the results reported in this paper.
Among women with a body mass index of ≥25 kg/m2
, Tworoger et al. (29
) observed a significant inverse association between 25(OH)D quartiles and ovarian cancer risk (Ptrend
= 0.04). A similar inverse association was found in this study (P
= 0.01). In a sensitivity analysis excluding the Nurses’ Health Study participants, who were also included in the study by Tworoger et al. (29
), our results became statistically nonsignificant (Ptrend
= 0.16) but the point estimates remained similar. Thus, data from these 2 studies suggest that vitamin D concentrations may be inversely associated with ovarian cancer risk among overweight and obese individuals.
Although body mass index is not a clear risk factor for ovarian cancer (40
), some data suggest adiposity is inversely associated with circulating 25(OH)D concentrations, likely because vitamin D is fat soluble (41
). In rats fed vitamin D supplements, adipose tissue concentrations of vitamin D increased significantly; interestingly, the study reported that vitamin D was released from adipose tissue, particularly during fasting. This suggests that adipose tissue may be an important factor in determining long-term vitamin D status (45
). Thus, it is possible that women with a higher body mass index may have more bioavailable vitamin D at the tissue level or that circulating concentrations may better reflect long-term tissue exposure in this subpopulation. More research is needed to elucidate this potential relation.
The median level of 25(OH)D was the lowest among participants from the Shanghai Women's Health Study. Excluding women from this study, however, did not materially affect the study results. Results stratified by age at blood draw and oral contraceptive use suggested no statistically significant interaction of these factors on the association of circulating vitamin D with ovarian cancer risk. Further, no associations were observed for specific histologic subtypes, although power was limited for some subtype analyses.
Strengths of the current study included a large sample size, the use of prediagnostic blood samples, and a wide exposure range due to inclusion of studies in different geographic locations. Additionally, information was available on numerous potential demographic and lifestyle factors, allowing evaluation of potential confounding effects by established ovarian cancer risk factors as well as potential effect modification.
One limitation of a pooled analysis is that data were collected using different methods across studies. Because of the different study methods, harmonization of data was a challenge, and information on some potential confounders, such as tubal ligation, was not collected in all studies and therefore not included in analyses. However, information was available on most established risk factors for ovarian cancer, and it is unlikely that factors such as tubal ligation are correlated with circulating vitamin D levels.
Even with a relatively large number of cases, sample sizes for some subanalyses were small, such as those stratified by tumor subtype and other ovarian cancer risk factors.
Only one blood measurement per person was taken, and within-person variation in circulating 25(OH)D could have obscured a true association. However, the correlation over 3 years is around 0.70 for 25(OH)D (46
), and unpublished results from 2 studies in our analysis (New York University Women's Health Study and Nurses’ Health Study) showed high intraclass correlations (35
), suggesting that concentrations are relatively stable over time. Additionally, seasonal variation was addressed in multiple ways, including matching cases and controls on season and constructing cutpoints based on season-specific quartiles among controls. Finally, as with any observational study, it is not possible to completely eliminate potential confounding effects, although results were similar in crude and fully adjusted models.
In conclusion, this pooled analysis did not find evidence of a strong overall association between circulating 25(OH)D and ovarian cancer risk. However, there was some suggestion that low circulating vitamin D might be associated with an increased risk of ovarian cancer among overweight women.