Genetic link between migraines and breast cancer risk uncovered

By Vijay Kumar MalesuReviewed by Benedette Cuffari, M.Sc.Sep 24 2023

In a recent study published in BMC Cancer, researchers assess the association between migraine susceptibility and breast cancer risk using Mendelian randomization (MR) analysis.

Study: Effect of genetic liability to migraine and its subtypes on breast cancer: a mendelian randomization study. Image Credit: Studio Romantic / Shutterstock.com

Background 

Migraine, a prevalent neurological disorder, frequently affects women between 25 and 55 years of age and is linked to various health issues, including dementia, cardiovascular disease, and potentially cancer.

Current research on the relationship between migraine and breast cancer has provided inconsistent findings. For example, while the Nurses' Health Study found no association between the two conditions, two other studies utilizing MR analysis and genome-wide association studies (GWAS) data reported a relationship between migraines and breast cancer.

Additional research is needed to elucidate the association between migraines and breast cancer due to inconsistent findings in previous studies, coupled with the limitations of traditional observational methods. 

About the study 

To explore the causal relationship between outcome and exposure, a two-sample MR study was conducted. This method relies on three fundamental assumptions that genetic variations are robustly associated with exposure, are unconnected with potential confounders, and do not have a direct influence on outcomes. In the current study, extensive sets of GWAS data were procured from participants of European descent, thus ensuring adherence to these foundational principles.

The data regarding any migraine (AM) was extracted from a recent publication that comprised 102,804 cases and 771,257 controls. From these, specific subsets of migraine with aura (MA) and without aura (MO) were isolated for further analysis. GWAS statistics for breast cancer were obtained from the Breast Cancer Association Consortium (BCAC) with a large sample of both cases and controls.

The selection of instrumental variables, particularly single nucleotide polymorphisms (SNPs), followed a meticulous procedure. SNPs were carefully screened for strong associations with migraine, minor allele frequency, and direct relation to outcomes. The researchers also excluded any potential confounding factors like body mass index (BMI) and age of menopause.

The Wald ratio was utilized to gauge the migraine's effect on breast cancer for each SNP. The main results were determined by the inverse-variance weighted (IVW) method, and multiple tests were used to assess any heterogeneity in the data. Sensitivity analyses like MR-Egger, weighted median (WM) methods, and the Robust Adjusted Profile Score (MR-RAPS) were conducted to account for possible inaccuracies or biases in the data.

A structured three-step process was applied to mitigate potential bias due to heterogeneity in the results. This involved a series of MR analyses, with each stage addressing potential outliers and reassessing the MR findings.

Study findings 

In the present study, detailed information was provided regarding select SNPs, for which the observed F statistics vary from 29.9 to 314.8. For associations relating to AM and MO, most were strongly evidenced, while for MA, there was a noted decrease in statistical power, which was attributed to its low explained variance and having only three instrumental variables (IVs). This suggests that care should be taken when interpreting any relationship between MA and breast cancer.

Four SNPs were removed from the analysis for AM due to their association with confounding factors. Among them, three were associated with BMI, and one was related to the age at which menstruation begins. Another SNP, rs10828247, was excluded from the AM exposure analysis due to its direct relationship with the resulting outcome.

An increase in genetically predicted AM was correlated with an increased risk of overall breast cancer. Although the risk of estrogen receptor-positive (ER+) breast cancer increased, there was no discernible effect of AM on ER- breast cancer. After adjusting for the False Discovery Rate (FDR), AM remained a significant risk factor for both overall and ER+ breast cancer.

During the analysis between MA and breast cancer, only three SNPs were chosen as IVs. These results did not identify a causal effect of MA on b overall, ER+, or ER- breast cancer.

An increase in MO was associated with a higher risk of overall breast cancer and a significantly increased risk for ER- breast cancer; however, no causal relationship was identified between MO and ER+ breast cancer. While the results from different methods were consistent, MO was only suggestively associated with overall breast cancer risk.

Further analysis did not identify any issues related to horizontal pleiotropy or heterogeneity. However, some potentially influential SNPs were detected, thus emphasizing a cautious interpretation of the findings.