Bidirectional relationship between olfaction and Parkinson's disease

Abstract

Hyposmia (decreased smell function) is a common early symptom of Parkinson's disease (PD). The shared genetic architecture between hyposmia and PD is unknown. We leveraged genome-wide association study (GWAS) results for self-assessment of 'ability to smell' and PD diagnosis to determine shared genetic architecture between the two traits. Linkage disequilibrium score (LDSC) regression found that the sense of smell negatively correlated at a genome-wide level with PD. Local Analysis of [co]Variant Association (LAVA) found negative correlations in four genetic loci near GBA1, ANAPC4, SNCA, and MAPT, indicating shared genetic liability only within a subset of prominent PD risk genes. Using Mendelian randomization, we found evidence for a strong causal relationship between PD and liability towards poorer sense of smell, but weaker evidence for the reverse direction. This work highlights the heritability of olfactory function and its relationship with PD heritability and provides further insight into the association between PD and hyposmia.

Fig. 1. Study design.

Top: The study used two GWAS summary statistics of PD and self-reported “ability to smell”. Bottom-left: The LDSC and LAVA were used to estimate heritability and genetic correlation of the two traits overall and at specific genomic loci. Bottom-right: MR was used to make causal estimations between the two traits while testing for horizontal pleiotropy.

Fig. 2. LocusCompare plots of LAVA results and MR results on PD and the ability to smell.

A–D LocusCompare plots of GBA1, ANAPC4, SNCA, and MAPT loci with PD and ability to smell. Y and x-axes indicate the two-tailed p value of association for PD and the ability to smell, respectively. Lead SNP is the SNP with the lowest sum of the p value from both studies. LD r² value relative to the lead SNP is indicated by color. E, F Scatter and forest plots of MR results before removal of the heterogeneity outliers. G, H Scatter and forest plots of MR results after removal of heterogeneity outliers determined by MR-PRESSO. On the scatterplot, y and x-axes indicate the effect size of SNPs on the ability to smell and PD, respectively, in the log odds ratio. The vertical and horizontal bars indicate the 95% confidence interval for the effect size on PD risk and the ability to smell, respectively. Colored lines represent the different MR results. On the forest plot, individual SNP MR results are colored in black, while the meta-analysis results are in red. The y-axis label indicates the individual SNPs for the single SNP MR or the meta-analysis method for the multi-SNP MR. The x-axis represents the MR effect size in log odds ratio. The horizontal bars indicate the 95% confidence interval for the MR effect size.