MRC Integrative Epidemiology Unit (IEU) Seminar Series

Title: Using 23andMe data to understand the features and causes of Parkinson’s disease

Abstract: Using Mendelian randomisation (MR), we sought to determine the causal relationship between three unhealthy traits, namely tobacco smoking, alcohol intake, and having a higher body mass index (BMI), and Parkinson’s disease (PD) risk. Observational studies have flagged smoking and alcohol intake as potential protective factors against developing PD; associations that merit careful verification given their detrimental effects on most other health outcomes. Low BMI appears to be associated with higher PD risk in some case-control studies, albeit inconsistently. Because of the possibility of residual confounding and reverse causation, it is unclear whether such epidemiological associations are truly causal. MR involves the use of genetic variants to explore causal effects of exposures or traits on outcomes, thus minimising these sources of bias. We performed genome-wide association studies for each phenotype to identify single nucleotide polymorphisms (SNPs) associated with the traits of interest. MR analysis of the relationship between each trait and PD was undertaken using a split-sample design. The inverse weighted (IVW) method was used to combine the SNP-specific effect estimates to determine the effect of genetically-related traits on PD. Two additional MR methods, MR-Egger and a weighted mean-based approach, were used to check for bias in the IVW estimate, arising through violations of MR assumptions. In the IVW analysis, ever-smoking caused a significant reduction in PD risk (OR 0.96; 95% confidence interval [CI] 0.92-0.99; p=0.01). Sensitivity analyses did not suggest bias from horizontal pleiotropy or invalid instruments. Alcohol intake and was causally linked with increased risk of PD (IVW OR 1.13, 95% CI 1.03-1.24; p=0.01) and a 1 kg/m2 increase in BMI was causally linked with reduced risk of PD (OR 0.99, 95% CI 0.98-1.00; p=0.01). Using a split-sample MR approach in a cohort of over 2.4 million participants, we observed a protective effect of smoking on risk of PD. Although the mechanisms underlying the apparent protective effect of smoking on PD risk remain speculative, these findings warrant the prioritisation of related therapeutic targets, such as nicotine agonists, in clinical trials. Conversely, alcohol consumption appeared to causally increase risk of PD. In keeping with previous studies, genetically-estimated higher BMI had a protective effect on PD, but the effect was marginal.

Biography: 

Karl Heilbron is a Scientist I, Statistical Genetics at 23andMe. His main focus is on studying the genetics of Parkinson's disease (PD) and other neurological diseases. In addition to genome-wide association studies, Dr. Heilbron has characterized the spectrum of phenotypes associated with PD--the PD phenome--using 23andMe's wide variety of online survey data. He is also interested in using Mendelian randomization to infer causal relationships between associated phenotypes. Prior to his current position, Dr. Heilbron completed his postdoc at 23andMe, obtained a DPhil in Zoology from the University of Oxford, and obtained a BSc in Evolutionary Science from the University of Western Ontario.

All welcome