Exome-wide association study of levodopa-induced dyskinesia in Parkinson's disease

Abstract

Levodopa is the standard long-term dopamine replacement therapy to treat Parkinson's disease (PD) symptoms. With time, levodopa may induce debilitating dyskinesias (LID), the treatment of which represents a large clinically unmet need. However, time-to-LID onset varies between patients, reflecting a possible genetic component. We performed an hypothesis-free whole-exome sequencing (WES)-based screening of time-to-LID onset and attempted replication of previously published candidate gene studies. A WES association analysis was carried out in 134 PD patients in a meta-analytical framework. Replication was attempted in an independent study of 97 PD patients. Variants from previously reported candidate genes (OPRM1, COMT, BDNF) were also specifically examined. We significantly replicated, for the first time, an association of variant rs1799971 in the OPRM1 gene with time-to-LID onset. Furthermore, we identified two novel potentially functional variants, in the MAD2L2 (rs2233019) and MAP7 (rs35350783) genes, which were significantly associated at the discovery stage. In the replication study, the two variants showed direction-consistent effects but did not achieve the replication significance threshold. Our study provides the first WES results for time-to-LID onset, where we replicate association at OPRM1, and suggest new variants in MAD2L2 and MAP7 genes that are significant in discovery, but require larger datasets for replication. The results are being made publicly available to allow for independent external validation.

Figure 1

Flow chart summarizing the methodology and results of this study. Two main analyses were performed: an hypothesis-free WES-wide scan of variants associated with time to LID onset and an hypothesis-driven lookup of known candidate variants in our data.

Figure 2

(A) Odds and hazard ratios of the replicated OPRM1 variant for the effect allele G (genotypes AG or GG) versus the other allele A with genotype AA. (B) Forest plot of the Hazard ratio of rs2233019 in gene MAD2L2 in the two discovery studies DISPGP and FRAGAMP, the meta-analysis, and the replication study MNI-PD. (C) Forest plot of the Hazard ratio of rs35350783 in gene MAP7 in the two discovery studies DISPGP and FRAGAMP, the meta-analysis, and the replication study MNI-PD. For (B,C), the point/center of the triangle corresponds to the Hazard ratio, the line/dimension of the triangle corresponds to the 95% confidence interval. GT genotype, HR hazard ratio, CI confidence interval.