Skip to main content
NCBI home page
Medicine logoLink to Medicine
. 2026 Jul 17;105(29):e49642. doi: 10.1097/MD.0000000000049642

Causal assessment of 187 dietary habits with ovarian cancer from multiple sources via Mendelian randomization and meta-analysis

Chengying Yu a, Zhang Ruoxi b, Jun Xu a, Zhaifang Ye a,*
PMCID: PMC13384660  PMID: 42469968

Abstract

Observational studies have suggested that dietary habits may influence ovarian cancer risk, but Mendelian randomization (MR) evidence remains limited. Genetic data for 187 dietary habit phenotypes and ovarian cancer were collected and preprocessed. MR analyses were performed using 2 independent ovarian cancer datasets. Inverse-variance weighted estimates were pooled by meta-analysis, and multiple-testing correction was applied. Reverse MR analyses were conducted for significant dietary phenotypes to assess causal direction. Among the 187 dietary habit phenotypes, only cherry preference (GCST90094731) showed a significant causal association with ovarian cancer risk. In the FinnGen R12 dataset, the inverse-variance weighted estimate was not significant (odds ratio [OR] = 0.902, 95% confidence interval [CI]: 0.703–1.158; P = .419). In the OpenGWAS dataset, cherry preference was associated with reduced ovarian cancer risk (OR = 0.809, 95% CI: 0.691–0.948; P = .0087). Meta-analysis yielded a combined OR of 0.835 (95% CI: 0.730–0.954; P = .0081), and the association remained significant after Bonferroni correction. Reverse MR analysis showed no evidence of reverse causality. Genetically predicted cherry preference was associated with a lower risk of ovarian cancer. These findings provide genetic evidence supporting a potential role of dietary habits in ovarian cancer prevention and may inform precision nutrition and public health strategies.

Keywords: dietary habit, Mendelian randomization analysis, meta-analysis, multiple corrections, ovarian cancer, reverse Mendelian randomization analysis

1. Introduction

Ovarian cancer is the seventh most common malignancy among women worldwide, with approximately 300,000 new cases and nearly 200,000 deaths each year. Its incidence rises markedly in women aged ≥50 years, peaking between 60 and 70 years of age. Due to the lack of specific symptoms in the early stages, more than 70% of patients are diagnosed at an advanced stage, and the overall 5-year survival rate is only about 45%.[1,2] Geographic variation exists, with higher incidence rates in North America and Europe and relatively lower rates in Asia and Africa.[3]

Risk factors include advanced age, family history, nulliparity or late childbirth, long-term hormone replacement therapy, polycystic ovary syndrome, irregular menstruation, obesity, and high-fat diets.[4] Clinical manifestations are often nonspecific, such as abdominal distension, abdominal pain, indigestion, loss of appetite, urinary frequency, weight loss, fatigue, and postmenopausal bleeding. Diagnosis mainly relies on imaging, tumor marker detection (e.g., cancer antigen 125 and human epididymis protein 4), and pathological examination. Surgery remains the cornerstone of treatment, supplemented by chemotherapy, radiotherapy, and targeted therapy. Preventive measures include routine gynecological examinations, family history assessment, early screening, and risk monitoring, with prophylactic surgery or close surveillance recommended for high-risk individuals.[5,6]

In recent years, the relationship between diet and ovarian cancer risk has attracted increasing attention, although epidemiological evidence remains inconsistent. Several prospective cohort studies and systematic reviews have suggested that overall dietary quality – such as adherence to the Mediterranean diet, Dietary Approaches to Stop Hypertension, or HEI indices – shows limited or inconclusive associations with ovarian cancer risk. However, potential links have been observed for specific dietary components.[7]

First, a high dietary glycemic load may be associated with increased ovarian cancer risk, whereas higher fiber intake appears to be protective. Second, in terms of dairy products, higher consumption of whole milk has been linked to increased risk, while low-fat dairy, calcium, and vitamin D intake may exert protective effects. For alcohol and coffee/tea, most studies have not reported significant associations, although some evidence suggests that green tea may provide modest protection, albeit with insufficient evidence. High intake of soy and phytoestrogens has been associated with reduced risk in some Asian populations, though findings vary across populations. Fish and ω-3 fatty acids generally show no significant association, whereas excessive trans fat intake may increase risk.[8,9]

Beyond cancer risk, dietary habits have also been examined in relation to prognosis among ovarian cancer patients. Evidence indicates that higher overall diet quality is associated with reduced all-cause mortality, while the “inflammatory dietary index” has not demonstrated a significant impact in multiple prospective analyses. These findings suggest that the role of diet in ovarian cancer may not solely depend on inflammatory pathways but also involve metabolic and hormonal regulation.[10,11] Overall, current evidence supports the potential importance of balanced dietary patterns in ovarian cancer prevention, but further research is required to elucidate the underlying mechanisms.

Given the inconsistencies in current evidence, this study employed Mendelian randomization (MR) to systematically assess the causal relationships between 187 dietary habit phenotypes and ovarian cancer risk. The study design included the collection and preprocessing of dietary exposure and ovarian cancer outcome data; MR analyses conducted separately in 2 independent datasets; and meta-analysis of inverse-variance weighted (IVW) results from both analyses, with multiple testing correction to enhance robustness and precision. The advantage of MR combined with meta-analysis lies in integrating data across studies, reducing bias, investigating potential heterogeneity and strengthening the generalizability of the conclusions. Finally, reverse MR analyses were performed for selected positive dietary phenotypes and ovarian cancer to clarify the causal direction, thereby providing more reliable evidence to uncover the role of dietary habits in the development and progression of ovarian cancer. The analytical framework and interpretation of MR findings followed established methodological guidance for two-sample MR studies.

2. Materials and methods

2.1. Study design

In this study, we first collected and organized data on exposures and outcomes, followed by preliminary cleaning and standardization. Subsequently, 187 dietary habit phenotypes were analyzed with ovarian cancer data from 2 independent databases using MR. The IVW results from both MR analyses were then subjected to meta-analysis, and multiple testing correction was applied to improve the reliability and precision of the findings. The advantage of MR combined with meta-analysis is that it allows the integration of evidence across multiple studies, reduces the risk of bias, addresses potential heterogeneity, and enhances the generalizability of the conclusions. By integrating data from multiple sources, this approach enables a more comprehensive assessment of the causal relationship between exposures and outcomes, thereby producing more robust and credible results and deepening the understanding of the research question. Finally, reverse MR analyses were performed for selected positive dietary phenotypes and ovarian cancer to further verify the causal direction and strengthen the scientific validity of the conclusions.[12,13] In addition, this study adhered to and refined the Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization reporting guidelines to ensure transparency and reproducibility.

2.2. Genome-wide association study (GWAS) data sources for the 187 dietary habit phenotypes

The dietary exposure data were derived from GWASs of large-scale cohorts investigating food liking and food preference phenotypes, aiming to systematically explore the genetic determinants of dietary habits. Importantly, these phenotypes reflect genetically influenced food preference tendencies rather than direct quantitative measures of habitual intake. Therefore, they are informative for studying behavioral predisposition toward specific foods, but they should not be interpreted as exact proxies for the amount or frequency of food consumption. The study included multiple categories of food preference phenotypes and, through genome-wide analyses, identified genetic variants associated with olfactory and gustatory function, metabolism, and reward systems. Further genetic correlation analyses revealed shared genetic underpinnings between food preferences and brain structure, neurotransmitter function, and psychological and behavioral traits, underscoring the important role of neurophysiological mechanisms in dietary choices. These findings not only elucidate the genetic basis of food preferences but also provide theoretical support for personalized nutritional interventions and the prevention of metabolic diseases. The original dietary habit dataset involved 187 dietary habit phenotypes, with GWAS accession codes ranging from GCST90094687 to GCST90094873.[14] According to recent studies, we applied the following criteria for single nucleotide polymorphism (SNP) selection: P < 5 × 10−6, F > 10, minor allele frequency (MAF) > 0.01, clump_kb = 10,000, and clump_r2 = 0.001.

2.3. Sources of GWAS data on ovarian cancer

To avoid overlap between antibody immune response data and ovarian cancer data, outcome data on ovarian cancer were obtained from 2 independent sources. The first dataset was derived from the FinnGen R12 database, comprising 2339 cases and 222,078 controls.[15] The second dataset was obtained from the OpenGWAS database, contributed by the Ovarian Cancer Association Consortium, with GWAS ID ieu-a-1120.[16] This dataset included a total of 66,450 individuals, consisting of 25,509 cases and 40,941 controls. All participants in both datasets were of European ancestry. The criteria for selecting outcome data included prioritizing studies with larger sample sizes, widely representative data sources, and study populations consistent with the exposures of interest.

2.4. Criteria for selection of instrumental variables (IVs)

In MR studies, the selection of valid IVs is crucial. In the primary analysis, SNPs associated with each dietary habit phenotype at P < 5 × 10−6 were selected as IVs. This threshold was chosen to balance instrument strength and instrument number, thereby retaining sufficient variants for stable MR estimation and adequate statistical power. To further assess robustness, an additional sensitivity analysis was conducted using the conventional genome-wide significance threshold of P < 5 × 10−8. For the key exposure phenotype cherry preference, 58 SNPs were retained under P < 5 × 10−6, whereas only 5 SNPs were available under P < 5 × 10−8 (Supplementary Table 1, Supplemental Digital Content 1), suggesting that the stricter threshold substantially reduced the number of eligible instruments and may have decreased statistical power.[17]

Next, to further screen for strong instruments, the F-statistic was calculated for each SNP using the formula F = (β/SE)2. Only SNPs with an F-value > 10 were retained, which helped eliminate weak instruments and enhance the reliability of the results (Supplementary Table 2, Supplemental Digital Content 2). Thus, although a relatively relaxed association threshold was used in the primary analysis, the retained variants still satisfied the conventional criterion for instrument strength. In addition, the MAF was calculated based on the effect allele frequency (EAF): if EAF was <0.5, MAF was set equal to EAF; otherwise, it was set to 1 − EAF. Only SNPs with MAF > 0.01 were retained to exclude rare variants that might bias the results.[18,19]

Finally, the filtered data were reformatted for MR analysis, and linkage disequilibrium clumping was applied to avoid bias from correlated variants. The specific criteria were set as a physical distance threshold of 10,000 kilobases and a linkage disequilibrium threshold of r2 = 0.001.[20] These steps ensured the independence of the IVs and the accuracy of the results.

3. Statistical analysis

3.1. Causal validation of the 187 dietary habit phenotypes to ovarian cancer

All data analyses in this study were conducted using R version 4.5.1 (R Foundation for Statistical Computing, https://www.r-project.org/). First, SNPs from ovarian cancer outcome data that matched the 187 dietary habit phenotypes were selected. Palindromic SNPs were then processed according to the specific criterion action = 2. In addition, SNPs with mr_keep = false were excluded.[20]

Before applying Mendelian Randomization Pleiotropy RESidual Sum and Outlier (MR-PRESSO), horizontal pleiotropy tests were performed on the processed data. If a SNP had a P-value < .05, it was considered pleiotropy-positive and treated as an outlier, which was subsequently removed using the MR-PRESSO method to ensure data accuracy. The parameters for MR-PRESSO were set as NbDistribution = 3000 and SignifThreshold = 0.05. For SNPs with P-values > .05, no outliers were considered to exist.[21]

After this refinement, heterogeneity tests were performed prior to MR analyses. Although heterogeneity has a limited impact on results, SNPs showing significant heterogeneity (Q_pval < 0.05) were analyzed using the IVW random-effects model, whereas those without significant heterogeneity were analyzed using the IVW fixed-effects model. In addition, regardless of heterogeneity status, MR-Egger regression and the weighted median approach were applied, and odds ratios (ORs) were calculated.[22]

To further enhance the robustness of the results, MR findings for dietary habit phenotypes and ovarian cancer from the 2 databases were combined using meta-analysis. The statistical significance of the meta-analysis results was adjusted for multiple testing using the Bonferroni method to minimize the probability of type I error.[23]

3.2. The causal link between ovarian cancer and dietary habit phenotypes

In this process, the positive dietary habit phenotypes were treated as outcome data, while ovarian cancer was considered the exposure. The same procedures for IV selection and data analysis as in the forward MR analysis were applied. The primary aim of this step was to validate the direction of causality between the 2 traits. Accordingly, the same thresholds and criteria were used to select valid IVs, and similar data processing and analytical methods were employed to determine the causal direction.

3.3. Sensitivity analysis

Horizontal pleiotropy refers to situations in which different treatments or exposures may have varying effects across individuals or contexts, and these effects may be mistakenly attributed to differences between the experimental and control groups rather than reflecting the true causal effect. To minimize the impact of horizontal pleiotropy on the results, GWAS data were tested for pleiotropy, and SNPs with P < .05 were considered pleiotropy-positive outliers. These outliers were removed using the MR-PRESSO method, with parameters set as NbDistribution = 3000 and SignifThreshold = 0.05 (Supplementary Table 3, Supplemental Digital Content 3). In addition, a supplementary sensitivity analysis using the more stringent criterion of P < 5 × 10−8 was performed for the main positive phenotype.

Heterogeneity refers to variability or differences among study subjects, observations, or experimental conditions. In statistical and methodological contexts, heterogeneity often reflects differences between individuals that may arise from physiological or psychological factors, socioeconomic status, or environmental influences. Such diversity enhances the generalizability and representativeness of study findings but also increases analytical complexity and interpretive challenges. In this study, heterogeneity tests were performed, and SNPs with significant heterogeneity (Q_pval < 0.05) were analyzed using the IVW random-effects model, whereas those without significant heterogeneity were analyzed using the IVW fixed-effects model. This approach ensured the accuracy and reliability of the MR results (Supplementary Table 4, Supplemental Digital Content 4).

4. Results

4.1. Causal validation of the 187 dietary habit phenotypes for ovarian cancer

A total of 187 dietary habit phenotypes were analyzed with ovarian cancer data from 2 independent databases using MR. The IVW results were subsequently pooled through meta-analysis, followed by multiple testing correction. Ultimately, only the dietary habit phenotype cherry preference (GCST90094731) showed a significant association with ovarian cancer. Specifically, in the FinnGen R12 database, the IVW analysis yielded an OR = 0.902 (95% confidence interval [CI]: 0.703–1.158, P = .419), which was not statistically significant (Fig. 1), whereas in the OpenGWAS database, the result was OR = 0.809 (95% CI: 0.691–0.948, P = .0087), suggesting that cherry preference may act as a protective factor for ovarian cancer (Supplementary Table 5, Supplemental Digital Content 5; Fig. 2). Moreover, the directions of the β estimates were consistent across IVW, MR-Egger, and weighted median approaches. Based on these 2 MR analyses, a meta-analysis of the primary IVW results was performed (Supplementary Table 6, Supplemental Digital Content 6), and the significance P-value was corrected using the Bonferroni method. The combined result was OR = 0.835 (95% CI: 0.730–0.954, P = .0081; Supplementary Table 7, Supplemental Digital Content 7). Collectively, the evidence demonstrated that the dietary habit phenotype cherry preference had a significant causal association with ovarian cancer after MR analysis, meta-analysis, and multiple testing correction. A forest plot was also generated to visualize the findings (Fig. 3). Under the stricter threshold of P < 5 × 10−8, only 5 SNPs were available for cherry preference, compared with 58 SNPs under P < 5 × 10−6. Therefore, the P < 5 × 10−8 analysis was treated as a sensitivity analysis rather than the primary analysis because the markedly reduced number of instruments was likely to weaken estimation precision and statistical power.

Figure 1.

Figure 1.

Open in a new tab

Combined Mendelian randomization plots of cherry preference on ovarian cancer in the FinnGen dataset. (A) Leave‐one‐out sensitivity analysis evaluating the influence of each SNP on the overall causal estimate. (B) Forest plot showing the causal effect estimates for individual SNPs and the pooled estimates obtained using the IVW and MR‐Egger methods. (C) Funnel plot assessing the symmetry of SNP‐specific causal estimates and potential directional pleiotropy. (D) Scatter plot showing the associations of genetic variants with cherry preference and ovarian cancer, with regression lines corresponding to the IVW, MR‐Egger, weighted median, simple mode, and weighted mode methods. IVW = inverse‐variance weighted, MR = Mendelian randomization, SNP = single‐nucleotide polymorphism.

Figure 2.

Figure 2.

Open in a new tab

Combined Mendelian randomization plots evaluating the causal effect of cherry preference on ovarian cancer in the OpenGWAS dataset. (A) Leave‐one‐out sensitivity analysis assessing whether the overall causal estimate was driven by any individual SNP. (B) Forest plot showing the causal effect estimates of individual SNPs and the pooled estimates obtained using the IVW and MR‐Egger methods. (C) Funnel plot assessing the symmetry of SNP‐specific causal estimates and potential directional pleiotropy. (D) Scatter plot showing the associations of SNPs with cherry preference and ovarian cancer, with regression lines representing the IVW, MR‐Egger, weighted median, simple mode, and weighted mode methods. IVW = inverse‐variance weighted, MR = Mendelian randomization, SNP = single‐nucleotide polymorphism.

Figure 3.

Figure 3.

Open in a new tab

Forest plot of positive results after meta-analysis. CI = confidence interval, OR = odds ratio, SNP = single nucleotide polymorphism.

4.2. The causal link between ovarian cancer and positive dietary habit phenotypes

In the reverse MR validation, no significant association was observed between cherry preference (GCST90094731) and ovarian cancer. Specifically, when ovarian cancer data from the FinnGen R12 dataset were treated as the exposure, the IVW analysis yielded an OR = 1.016 (95% CI: 0.999–1.034, P = .063). When ovarian cancer data from the OpenGWAS dataset were used as the exposure, the IVW result was OR = 0.991 (95% CI: 0.959–1.024, P = .589). Taken together, these results indicate that the dietary habit phenotype cherry preference does not exhibit evidence of reverse causality with ovarian cancer across the 2 independent datasets (Supplementary Table 8, Supplemental Digital Content 8).

5. Discussion

In this study, we systematically evaluated the causal relationship between multiple dietary habit phenotypes and ovarian cancer using large-scale genetic data. The results demonstrated a robust causal association between cherry preference and ovarian cancer risk, with this phenotype consistently showing a potential protective effect across multiple analytic approaches. By integrating data from 2 independent databases, the findings gained greater reliability and generalizability, suggesting that dietary behaviors, as modifiable lifestyle factors, may play an important role in the prevention of ovarian cancer. Furthermore, reverse MR analysis did not reveal evidence that dietary habits were influenced by disease risk, thereby strengthening the causal inference. Overall, this study provides new insights into the causal link between diet and ovarian cancer and lays the groundwork for future mechanistic exploration and personalized nutritional interventions.

A substantial body of epidemiological evidence suggests that dietary habits may play an important role in the onset and progression of ovarian cancer. Multiple cohort and case–control studies have shown that dietary patterns rich in vegetables, fruits, whole grains, and low-fat intake are generally associated with reduced ovarian cancer risk, whereas high-calorie diets, excessive red meat consumption, and high intake of sugar-sweetened beverages or refined carbohydrates may increase risk.[24,25] Specific nutrients such as dietary fiber, vitamin C, vitamin E, and polyphenols have been reported to exert protective effects through antioxidant, anti-inflammatory, and hormone-regulating pathways, while high-fat and high-cholesterol diets may promote inflammation and oxidative stress, thereby increasing susceptibility.[26] In addition, associations of alcohol consumption, dairy products, and coffee or tea intake with ovarian cancer risk have also been reported, but the results remain inconsistent across studies.

Nevertheless, despite the relatively large number of studies, current evidence remains inconclusive. Differences in dietary assessment tools, sample sizes, follow-up duration, and statistical methodologies may account for inconsistent results, while genetic background, lifestyle factors, and environmental exposures may also influence the associations. More importantly, traditional observational studies are often subject to confounding and reverse causality, limiting the ability to draw definitive causal inferences. In recent years, increasing attention has been given to genetic epidemiology approaches, particularly MR, which provides a more rigorous framework to investigate the causal relationship between dietary habits and ovarian cancer. Such methods help overcome confounding and measurement error, thereby offering novel perspectives and more reliable evidence regarding the role of diet in ovarian cancer prevention.[23,27,28]

Research directly addressing the relationship between cherries and ovarian cancer remains limited, but existing evidence suggests that cherries and their bioactive components may exert potential antitumor effects through biologically plausible pathways.[29,30] In the context of the present findings, 2 mechanism-oriented hypotheses merit emphasis. First, oxidative stress and chronic inflammatory signaling are closely involved in ovarian carcinogenesis, and nuclear factor kappa-light-chain-enhancer of activated B cells signaling is a recognized pathway promoting ovarian cancer cell survival, metastasis, and immune evasion. Because cherries are rich in anthocyanins and other polyphenols with antioxidant and anti-inflammatory activity, one plausible explanation is that a greater tendency toward cherry-related dietary behavior may be linked to reduced oxidative and inflammatory pressure and, consequently, to a less tumor-promoting microenvironment.[29,31]

Second, flavonoid-rich foods may influence cell proliferation and apoptosis. Anthocyanin-containing foods have been reported to modulate proliferation- and apoptosis-related pathways in cancer biology, and epidemiological evidence further suggests that higher dietary flavonoid intake is associated with a lower risk of ovarian cancer.[30,32–34] Although our MR phenotype represents cherry preference rather than measured cherry intake, the observed protective association is directionally consistent with the hypothesis that preference for flavonoid-rich foods may mark dietary patterns that are less favorable to ovarian tumor development. These mechanistic interpretations remain hypothesis-generating and should be tested in future functional and translational studies.

The findings of this study also open new avenues for exploring the role of dietary habits in ovarian cancer prevention. Future work should move beyond preference phenotypes alone and integrate large-scale genetic data with dietary intake assessment, circulating metabolite profiling, and microbiome data to determine whether the protective signal observed for cherry preference is mediated by actual intake, downstream metabolites, or correlated dietary patterns. Cherries and their abundant bioactive components, such as anthocyanins and other polyphenols, remain plausible nutritional targets for primary prevention, but the current results should be viewed as an initial causal clue rather than definitive evidence for a specific intake recommendation.[35] Furthermore, the development of personalized dietary interventions based on genetic susceptibility may enhance preventive efficacy and reduce disease burden at the population level.

Despite the novel findings, several limitations should be acknowledged. First, the data used in this study were primarily derived from European populations, and thus the generalizability of the findings to other ethnicities and regions remains to be validated. Second, the dietary habit phenotypes were derived from GWASs of food liking and preference rather than direct measures of actual intake amount or frequency. Accordingly, our results should be interpreted as evidence for genetically proxied preference tendencies, not as proof that a specific quantity of cherry consumption reduces ovarian cancer risk. Such phenotypes may capture sensory preference, reward-related behavior, and correlated lifestyle patterns, which could attenuate or complicate biological interpretation. Third, although MR methods offer advantages in reducing confounding and reverse causation, the results may still be affected by horizontal pleiotropy and genetic heterogeneity. In addition, the primary IV selection threshold of P < 5 × 10−6 represented a balance between instrument strength and instrument number. Although all retained SNPs had F-statistics > 10, applying the stricter threshold of P < 5 × 10−8 left only 5 instruments for the key phenotype, which may have reduced statistical power and limited the stability of causal estimation. Future research should include larger prospective studies across diverse populations and incorporate functional experiments to elucidate the biological mechanisms of cherry bioactive compounds in ovarian cancer prevention at the molecular and cellular levels. Moreover, further investigations into the combined effects of different dietary components within overall dietary patterns are warranted to provide stronger evidence for the development of personalized dietary interventions and public health strategies.

6. Conclusion

The study systematically evaluated the causal relationship between dietary habit phenotypes and ovarian cancer risk. The results demonstrated a robust protective association between cherry preference and ovarian cancer in the primary analysis using P < 5 × 10−6 for IV selection, supported across multiple analytic approaches and independent datasets. A supplementary stricter-threshold analysis using P < 5 × 10−8 was also conducted, but the number of available SNPs for the key phenotype decreased substantially, underscoring the trade-off between instrument stringency and statistical power. These findings provide novel genetic evidence for the role of diet in ovarian cancer prevention and suggest that dietary habits may serve as modifiable preventive factors. Future research across diverse populations and multi-omics levels is warranted to further elucidate the underlying mechanisms and explore the translational potential of dietary interventions in precision nutrition and public health.

Acknowledgments

Firstly, we express our profound thanks to all individuals and researchers who contributed to the GWAS data used in this research. Additionally, we extend our sincere gratitude and respect to the personnel involved with the associated public databases. Finally, our heartfelt appreciation goes out to every author who played a role in contributing to this study.

Author contributions

Conceptualization: Chengying Yu, Zhang Ruoxi, Jun Xu, Zhaifang Ye.

Data curation: Chengying Yu, Zhaifang Ye.

Formal analysis: Chengying Yu, Zhang Ruoxi, Jun Xu, Zhaifang Ye.

Investigation: Chengying Yu, Zhang Ruoxi, Jun Xu.

Methodology: Chengying Yu, Zhang Ruoxi, Jun Xu, Zhaifang Ye.

Project administration: Chengying Yu, Jun Xu, Zhaifang Ye.

Resources: Chengying Yu.

Software: Chengying Yu, Zhang Ruoxi.

Supervision: Zhaifang Ye.

Validation: Chengying Yu, Zhang Ruoxi, Jun Xu.

Writing – original draft: Chengying Yu, Zhang Ruoxi, Jun Xu.

Writing – review & editing: Zhaifang Ye.

medi-105-e49642-s001.xlsx (155.1KB, xlsx)
medi-105-e49642-s002.xlsx (1.2MB, xlsx)
medi-105-e49642-s003.xlsx (30.4KB, xlsx)
medi-105-e49642-s004.xlsx (47.4KB, xlsx)
medi-105-e49642-s005.xlsx (112.4KB, xlsx)
medi-105-e49642-s006.xlsx (50.2KB, xlsx)
medi-105-e49642-s007.xlsx (8.5KB, xlsx)
medi-105-e49642-s008.xlsx (9.1KB, xlsx)

Abbreviations:

CI
confidence interval
EAF
effect allele frequency
GWAS
genome-wide association study
HEI
Healthy Eating Index
IV
instrumental variable
IVW
inverse-variance weighted
Kb
kilobase pairs
MAF
minor allele frequency
MR
Mendelian randomization
MR-PRESSO
Mendelian Randomization Pleiotropy RESidual Sum and Outlier
OR
odds ratio
SNP
single nucleotide polymorphism

Participants in FinnGen provided informed consent for biobank research on the basis of the Finnish Biobank Act. Alternatively, separate research cohorts collected before the Finnish Biobank Act came into effect (in September 2013) and the start of FinnGen (August 2017) were collected on the basis of study-specific consent and later transferred to the Finnish biobanks after approval by Fimea, the National Supervisory Authority for Welfare and Health. Recruitment protocols followed the biobank protocols approved by Fimea. The Coordinating Ethics Committee of the Hospital District of Helsinki and Uusimaa (HUS) approved the FinnGen study protocol (No. HUS/990/2017).

The FinnGen study was approved by the THL (approval number THL/2031/6.02.00/2017; amendments THL/1101/5.05.00/2017, THL/341/6.02.00/2018, THL/2222/6.02.00/2018, THL/283/6.02.00/2019, and THL/1721/5.05.00/2019), the Digital and Population Data Services Agency (VRK43431/2017-3, VRK/6909/2018-3, and VRK/4415/2019-3), the Social Insurance Institution (KELA) (KELA 58/522/2017, KELA 131/522/2018, KELA 70/522/2019, and KELA 98/522/2019), and Statistics Finland (TK-53-1041-17).

The Biobank Access Decisions for FinnGen samples and data utilized in FinnGen Data Freeze 5 include the following datasets: THL Biobank BB2017_55, BB2017_111, BB2018_19, BB_2018_34, BB_2018_67, BB2018_71, BB2019_7, BB2019_8, and BB2019_26; Finnish Red Cross Blood Service Biobank 7.12.2017; Helsinki Biobank HUS/359/2017; Auria Biobank AB17-5154; Biobank Borealis of Northern Finland_2017_1013; Biobank of Eastern Finland 1186/2018; Finnish Clinical Biobank Tampere MH0004; Central Finland Biobank 1-2017; and Terveystalo Biobank STB 2018001.

The application number for the research on the 46 antibody immune responses is 27449.

CY and ZR contributed to this article equally.

The authors have no funding and conflicts of interest to disclose.

All data generated or analyzed during this study are included in this published article (and its supplementary information files).

Supplemental Digital Content is available in the online version of this article (http://dx.doi.org/10.1097/MD.0000000000049642).

How to cite this article: Yu C, Ruoxi Z, Xu J, Ye Z. Causal assessment of 187 dietary habits with ovarian cancer from multiple sources via Mendelian randomization and meta-analysis. Medicine 2026;105:29(e49642).

Contributor Information

Zhang Ruoxi, Email: ZRX2112@shtrhospital.com.

Jun Xu, Email: xujunly198111@163.com.

References

  • [1].Mazidimoradi A, Momenimovahed Z, Allahqoli L, et al. The global, regional and national epidemiology, incidence, mortality, and burden of ovarian cancer. Health Sci Rep. 2022;5:e936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [2].Chen G, Zhou S, Shi Q, et al. Global, regional, and national estimates of burden and risk factors of female cancers in child-bearing age: a systematic analysis for Global Burden of Disease Study and Bayesian projection to 2030. Transl Oncol. 2025;60:102473. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [3].Razaq L, Dhali A, Maity R, et al. Demographic trends in mortality due to ovarian cancer in the United States, 1999–2020. World J Clin Oncol. 2025;16:108393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [4].Webb PM, Jordan SJ. Global epidemiology of epithelial ovarian cancer. Nat Rev Clin Oncol. 2024;21:389–400. [DOI] [PubMed] [Google Scholar]
  • [5].Wei YF, Ning L, Xu YL, et al. Worldwide patterns and trends in ovarian cancer incidence by histological subtype: a population-based analysis from 1988 to 2017. EClinicalMedicine. 2025;79:102983. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [6].Meena RK, Syed NA, Sheikh ZA, et al. Patterns of treatment and outcomes in epithelial ovarian cancer: a retrospective North Indian single-institution experience. JCO Glob Oncol. 2022;8:e2200032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Sun H, Gong TT, Xia Y, et al. Diet and ovarian cancer risk: an umbrella review of systematic reviews and meta-analyses of cohort studies. Clin Nutr. 2021;40:1682–90. [DOI] [PubMed] [Google Scholar]
  • [8].Zhu L, Shu Y, Ran J, Zhang C. Glycemic load, but not glycemic index, is associated with an increased risk of ovarian cancer: a systematic review and meta-analysis. Nutr Res. 2024;123:67–79. [DOI] [PubMed] [Google Scholar]
  • [9].Merritt MA, Cramer DW, Vitonis AF, Titus LJ, Terry KL. Dairy foods and nutrients in relation to risk of ovarian cancer and major histological subtypes. Int J Cancer. 2013;132:1114–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Cao A, Esserman DA, Cartmel B, Irwin ML, Ferrucci LM. Association between diet quality and ovarian cancer risk and survival. J Natl Cancer Inst. 2024;116:1095–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [11].Qin Y, Chen XY, Cao F, et al. Pre- and post-diagnosis dietary patterns and overall survival in patients with epithelial ovarian cancer: a prospective cohort study. BMC Cancer. 2025;25:363. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [12].Liu S, Ding D, Liu F, Guo Y, Xie L, Han FJ. Exploring the causal role of multiple metabolites on ovarian cancer: a two sample Mendelian randomization study. J Ovarian Res. 2024;17:22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [13].Dobbie LJ, Pittam B, Zhao SS, et al. Childhood, adolescent, and adulthood adiposity are associated with risk of PCOS: a Mendelian randomization study with meta-analysis. Hum Reprod. 2023;38:1168–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [14].May-Wilson S, Matoba N, Wade KH, et al. Large-scale GWAS of food liking reveals genetic determinants and genetic correlations with distinct neurophysiological traits. Nat Commun. 2022;13:2743. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Kurki MI, Karjalainen J, Palta P, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613:508–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [16].Phelan CM, Kuchenbaecker KB, Tyrer JP, et al. Identification of 12 new susceptibility loci for different histotypes of epithelial ovarian cancer. Nat Genet. 2017;49:680–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [17].Burgess S, Thompson SG; CRP CHD Genetics Collaboration. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol. 2011;40:755–64. [DOI] [PubMed] [Google Scholar]
  • [18].Zhu Z, Zheng Z, Zhang F, et al. Causal associations between risk factors and common diseases inferred from GWAS summary data. Nat Commun. 2018;9:224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [19].Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [20].Hemani G, Zheng J, Elsworth B, et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife. 2018;7:e34408. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50:693–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [22].Burgess S, Thompson SG. Interpreting findings from Mendelian randomization using the MR-Egger method. Eur J Epidemiol. 2017;32:377–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [23].Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37:658–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [24].Schulz M, Lahmann PH, Boeing H, et al. Fruit and vegetable consumption and risk of epithelial ovarian cancer: the European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev. 2005;14:2531–5. [DOI] [PubMed] [Google Scholar]
  • [25].Grosso G, Bella F, Godos J, et al. Possible role of diet in cancer: systematic review and multiple meta-analyses of dietary patterns, lifestyle factors, and cancer risk. Nutr Rev. 2017;75:405–19. [DOI] [PubMed] [Google Scholar]
  • [26].Jordan SJ, Purdie DM, Green AC, Webb PM. Coffee, tea and caffeine and risk of epithelial ovarian cancer. Cancer Causes Control. 2004;15:359–65. [DOI] [PubMed] [Google Scholar]
  • [27].Alizadeh S, Djafarian K, Alizadeh M, Shab-Bidar S. The relation of healthy and Western dietary patterns to the risk of endometrial and ovarian cancers: a systematic review and meta-analysis. Int J Vitam Nutr Res. 2020;90:365–75. [DOI] [PubMed] [Google Scholar]
  • [28].Yarmolinsky J, Relton CL, Lophatananon A, et al. Appraising the role of previously reported risk factors in epithelial ovarian cancer risk: a Mendelian randomization analysis. PLoS Med. 2019;16:e1002893. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [29].Kelley DS, Adkins Y, Laugero KD. A review of the health benefits of cherries. Nutrients. 2018;10:368. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [30].Maaz M, Sultan MT, Noman AM, et al. Anthocyanins: from natural colorants to potent anticancer agents. Food Sci Nutr. 2025;13:e70232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Harrington BS, Annunziata CM. NF-κB signaling in ovarian cancer. Cancers (Basel). 2019;11:1182. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Hua X, Yu L, You R, et al. Association among dietary flavonoids, flavonoid subclasses and ovarian cancer risk: a meta-analysis. PLoS One. 2016;11:e0151134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].Fonseca LRS, Silva GR, Luís A, et al. Sweet cherries as anti-cancer agents: from bioactive compounds to function. Molecules. 2021;26:2941. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Cassidy A, Huang T, Rice MS, Rimm EB, Tworoger SS. Intake of dietary flavonoids and risk of epithelial ovarian cancer. Am J Clin Nutr. 2014;100:1344–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [35].Hejazi J, Amiri R, Nozarian S, Tavasolian R, Rahimlou M. Genetic determinants of food preferences: a systematic review of observational studies. BMC Nutr. 2024;10:24. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

medi-105-e49642-s001.xlsx (155.1KB, xlsx)
medi-105-e49642-s002.xlsx (1.2MB, xlsx)
medi-105-e49642-s003.xlsx (30.4KB, xlsx)
medi-105-e49642-s004.xlsx (47.4KB, xlsx)
medi-105-e49642-s005.xlsx (112.4KB, xlsx)
medi-105-e49642-s006.xlsx (50.2KB, xlsx)
medi-105-e49642-s007.xlsx (8.5KB, xlsx)
medi-105-e49642-s008.xlsx (9.1KB, xlsx)

Articles from Medicine are provided here courtesy of Wolters Kluwer Health

RESOURCES