Self-reported and genetically predicted coffee consumption and smoking in dementia: A Mendelian randomization study

Background and aims: Studies of self-reported coffee consumption and smoking on risk of dementia have shown results conflicting with two-sample Mendelian randomization studies. We tested the hypotheses that coffee consumption and smoking influence risk of dementia using observational and one-sample Mendelian randomization designs with individual level data.

Methods: We included 114,551 individuals from two Danish general population cohorts (median age 58 years). First, we tested whether high self-reported coffee consumption/smoking were associated with risk of dementia. Second, whether genetically predicted high coffee consumption/smoking due to variation near CYP1A1/AHR/CHRNA3 genes were associated with risk of dementia.

Results: We observed 3,784 dementia events. Moderate self-reported coffee consumption was associated with low risk of all dementia and non-Alzheimer's dementia, with a similar trend for Alzheimer's disease. Genetically predicted high coffee consumption was associated with high risk of all dementia (hazard ratio [95% confidence interval] per +1 cup/day: 1.20 [1.01-1.42]), with a similar trend for non-Alzheimer's dementia (1.23 [0.95-1.53]). High self-reported smoking was associated with high risk of non-Alzheimer's dementia. High genetically predicted smoking was associated with a trend towards high risk of all dementia and Alzheimer's disease (hazard ratios per +1 pack-year: 1.04 [0.96-1.11]) and 1.06 [0.97-1.16]).

Conclusions: Moderate self-reported coffee consumption was associated with low risk of all and non-Alzheimer's dementia, while high genetically predicted coffee consumption was associated with a trend towards the opposite. High self-reported smoking was associated with high risk of non-Alzheimer's dementia, with a similar trend for genetically predicted smoking on all dementia and Alzheimer's disease.