Introduction:
Alzheimer’s disease (AD) and ambient particulate matter 10 (PM10) have been associated in epidemiological studies. However, the relationship between PM10 and risk of AD has not been proven to be causal. Thus we used two-sample multivariable Mendelian randomization (MR) to examine this relationship.

Material and methods:
Genome-wide association studies (GWAS) for PM10 from UK Biobank, AD from EBI GWAS and IEU OpenGWAS were used for discovery and replication, respectively. Pooled meta-analysis of the inverse variance weighted (IVW) method was the main method. Sensitivity analyses included MR-Egger regression, weighted median, weighted mode and leave-one-out methods. The multivariable MR model adjusted for education. The MR estimates of causality association were expressed as odds ratios (OR) and corresponding 95% confidence intervals (CI).

Results:
There were in total 17 and 19 genetic variants associated with PM10 in the discovery and replication steps, respectively. In the univariate MR, pooled meta-analysis of genetically predicted PM10 was associated with a 99% increased risk of AD (95% CI: 1.25, 3.15, p = 0.004) per 1 standard deviation (SD) increment of PM10 by IVW, and in the multivariable MR with pooled meta-analysis, we found that each SD increase in PM10 was associated with a 127% increase in the risk of AD (95% CI: 1.33, 3.86, p = 0.002) after accounting for education levels.

Conclusions:
Increased PM10 levels were found to be significantly related to an increased risk of AD. This study provided evidence of genetic prediction of a causal relationship between PM10 and the risk of AD, suggesting that air pollution control may have significant implications for the prevention of AD.