A genetic risk score is differentially associated with migraine with and without aura

DOI: https://doi.org/10.1007/s00439- 017- 1816- 5

Abstract-Summary We investigated to what extent a genetic risk score (GRS) based on recently pub- lished, novel migraine-associated SNPs is associated with migraine prevalence, subtypes and severity in a large population-based sample.

We combined the remaining 21 SNPs into a GRS and analyzed the association

with migraine using logistic regression models.

The GRS was significantly associated with migraine (OR = 1.56, p = 0.02) and migraine without aura (MWOA) (OR = 2.01, p = 0.003), but not with migraine with aura (MWA).

The GRS was not associated with migraine frequency, intensity or interference

with daily activities.

We show that a GRS combining multiple genetic risk variants is associated with MWOA but not MWA, suggesting a different genetic susceptibility background underlying the two forms of migraine.

Extended: The GRS was used as a continuous predictor in a binary logistic regression model with migraine (or migraine subtypes) as the dependent variable, adjusting for age, sex, and a lifetime diagnosis of MDD.

2.1 Genetics

253

The GRS was also coded as a dichotomous variable to compare the association between migraine and the upper/lower quartiles of the GRS distribution using a binary logistic regression model.

Introduction A recent meta-analysis pooling data from 29 genome-wide association studies (GWAS) identified 12 SNPs associated with increased migraine risk and, specifi- cally, with larger effect sizes in MWOA compared to MWA cases (Anttila and oth- ers [88]).

A cumulative genetic risk evaluation of migraine subtypes, i.e., MWA and MWOA, and additional important aspects of migraine severity, including associated disabling symptoms, migraine intensity, and interference with daily activities under consideration of very recently detected novel migraine-associated variants would add further understanding to the genotype–phenotype interplay in migraine.

To close this gap of knowledge, we investigated the association between genetic susceptibility factors, taking recently described, novel genetic susceptibility mark- ers into consideration, and migraine prevalence and severity in a large population- based cohort.

We aimed to specifically evaluate if (1) a GRS based on migraine-associated SNPs is associated with migraine and migraine-type prevalence in this large population- based sample and if (2) the GRS is associated with migraine frequency and distinct traits of migraine severity.

Methods For each subject with a diagnosis of migraine, the following variables were extracted from the interview: age at onset, intensity of migraine (coded as a dichotomous vari- able comparing light/moderate versus severe migraine), migraine-associated gastro- intestinal symptoms (loss of appetite, hunger, nausea, vomit, diarrhea and abdominal pain), duration of the interval between two attacks, duration of the attack, and inter- ference with daily activities (evaluated with a score from 0 to 100).

The association between individual SNPs and migraine was tested using logistic

regression analyses adjusting for age, sex and a lifetime diagnosis of MDD.

The GRS was used as a continuous predictor in a binary logistic regression model with migraine (or migraine subtypes) as the dependent variable, adjusting for age, sex, and a lifetime diagnosis of MDD.

To test the association between the GRS and continuous or dichotomous clinical characteristics of migraine, we constructed linear or logistic regression models, respectively, adjusting for age, sex and lifetime MDD.

Results One genetic variant was associated with migraine at a nominal level (rs1024905, OR = 1.19, p = 0.016).

Analyses conducted in migraine subtypes also showed an association between rs1024905 and MWOA (unadjusted OR = 1.41, p = 0.0001; adjusted OR = 1.42, p = 0.0001), and a nominal association between rs4814864 and MWOA (unadjusted OR = 1.27, p = 0.017).

254

2 Mechanisms

When the GRS was dichotomized to compare migraine prevalence between the bottom and top quartiles of the GRS distribution, the GRS was still associated with migraine [OR: 1.45 (95% confidence interval = 1.09–1.93), p = 0.011] and MWOA [OR: 1.76 (95% confidence interval = 1.24–2.50), p = 0.002] but not MWA (OR = 1.03, p  =  0.89) in a regression logistic analysis adjusted for sex, age, and life- time MDD.

Discussion We demonstrate that the GRS based on SNPs previously identified in GWAS studies or meta-analyses to be associated with migraine specifically impacts MWOA, but not the risk for MWA.

Our study was not powered enough to investigate the association between

migraine and rare genetic variants, which might especially play a role in MWA.

Since available studies, including the largest meta-analysis conducted to date (Gormley and others [58]), have not been able to identify genetic variants signifi- cantly associated with MWA, further studies characterized by larger sample size and powered enough to identify rare genetic variants could be of help to construct a GRS that could prove to be more specific for MWA.

Our study included a lower number of participants with MWA compared to MWOA, which could have limited our statistical power to identify a significant association between the GRS and this subtype.

Conclusion The present study suggests that a GRS combining the effect of multiple loci is asso- ciated with MWOA but not MWA in a large population-based sample.

Genetic risk variants so far detected to play a role in migraine are not able to

explain a comprehensive set of clinical characteristics of migraine severity.

Acknowledgement A machine generated summary based on the work of Pisanu, Claudia; Preisig, Martin; Castelao, Enrique; Glaus, Jennifer; Pistis, Giorgio; Squassina, Alessio; Del Zompo, Maria; Merikangas, Kathleen R.; Waeber, Gérard; Vollenweider, Peter; Mwinyi, Jessica; Schiöth, Helgi B. 2017 in Human Genetics.

Identifying genetic variants for age of migraine onset in a Han Chinese population in Taiwan