The association between migraine genetic risk and dementia: A Mendelian randomization study

Abstract

AbstractBackgroundAlthough some studies have suggested a possible link between migraine and dementia, there have been concerns that these associations may be due to diagnostic bias or unmeasured confounding. We tested whether migraine causes an increased risk of dementia using Mendelian randomization, which leverages genetic variability in migraine risk to estimate effects independent of confounding.MethodIn a sample of 259,839 individuals of European Ancestry aged 55 years or older from the UK Biobank, we calculated a migraine genetic risk score (MGRS) for each individual based on 38 known genome‐wide significant risk single nucleotide polymorphisms. Using algorithmically defined dementia diagnosis and migraine diagnosis from hospital and primary care sources, 2,069 and 11,357 individuals were diagnosed with dementia and migraine, respectively. Fluid intelligence score (score on a 13‐item test to assess logic and reasoning ability) and total brain volume normalized for skull size (TBV) were used as outcomes in addition to dementia diagnosis and were available for 174,609 and 10,229 individuals, respectively. Linear, logistic, and linear mixed‐effects models were used to assess the associations between the MGRS and TBV, dementia diagnosis, and fluid intelligence, respectively. Using instrumental variable regression with the MGRS as the instrument, we estimated the causal effect of migraine on dementia diagnosis, fluid intelligence, and TBV. All analyses were adjusted for the first 10 genetic ancestry principal components, orthogonal polynomials of degrees 1‐4 for age at time of outcome assessment, and genetic sex.ResultThe MGRS was strongly associated with migraine (p < 0.0001), but not dementia diagnosis (p = 0.22), fluid intelligence (p = 0.15), or TBV (p = 0.12). The estimated causal effect of migraine was a 3% (95% CI: [‐2%, 8%]) increase in absolute risk of dementia, a 1.1 (95% CI: [‐1.0, 3.1]) point increase in total fluid intelligence score, and a 123 (95% CI: [‐50, 296]) cubic centimeter decrease in TBV.ConclusionThe MGRS was not significantly associated with dementia diagnosis, fluid intelligence, or TBV. Using a Mendelian randomization approach, we did not find evidence that migraine causes dementia.