Investigating the genetic architecture of Dementia with Lewy bodies: a genome-wide association study

Background: Dementia with Lewy Bodies (DLB) is the second most common form of dementia in the elderly but has been overshadowed in the research field, in part due to similarities between DLB, Parkinson's (PD) and Alzheimer’s diseases (AD). So far, no large-scale genetic study of DLB has been performed. To better understand the genetic basis of DLB, we have performed a genome-wide association study, with the aim of identifying genetic risk factors for this disorder Methods: Here we have performed the first genome-wide association study of DLB in a combined cohort of 1,743 DLB patients and 4,454 controls. To reduce genetic heterogeneity, all samples were of European ancestry. All cases were diagnosed according to established criteria for clinical or pathological DLB. In the discovery stage (comprising 1,216 cases and 3,791 controls, the latter were part of two publicly available dbGaP studies (phs000404.v1.p1 and phs000982.v1.p1)) we performed genotyping and exploited the recently established Haplotype Reference Consortium panel as the basis for imputation. Association analysis was performed in all cases as well as in only those with pathological diagnosis (974 cases). In the replication stage (comprising 527 DLB cases and 663 controls) we performed genotyping of significant and suggestive results. Lastly, we conducted a meta-analysis of both stages. Genotyping was undertaken at three locations: UCL, NIH and the Mayo Clinic. Findings: Results confirm previously reported associations: APOE (rs429358; OR=2.4 [95%CI 2.14-2.70]; p=1.05x10-48), SNCA (rs7681440; OR=0.73 [0.660.81]; p=6.39x10-10) and GBA (rs35749011; OR=2.55 [1.88-3.46]; p=1.78x10-09). They also provide novel candidate loci, namely CNTN1 (rs7314908; OR=1.53 [1.29-1.81]; p=1.02x10-06); further replication of findings at these two loci will be important. Additionally, we estimate the heritable component of DLB to be approximately 36%. Interpretation: Despite the relatively modest sample size for a GWAS, and acknowledging the potential biases from ascertaining samples from multiple locations, we present the most comprehensive and well-powered genetic study in DLB to date. These data unequivocally show that common genetic variability plays a role in this disease. Funding Funded by the Alzheimer’s Society, and the Lewy Body Society. Introduction Dementia with Lewy Bodies (DLB) is the second most common form of dementia following Alzheimer’s disease (AD) 1. Despite this fact, very little attention has been devoted to understanding the pathogenesis of this disorder, particularly when compared with the other common neurodegenerative diseases such as AD and Parkinson’s disease (PD). So far, the only fully penetrant genetic variability that has been identified and replicated as a specific cause of DLB are SNCA point mutations and gene dosage. Three major factors may have contributed to this low number of causative mutations: first, DLB, often a disease of old age, is not commonly seen in multiplex kindreds, meaning that successful linkage studies have been rare 2; second, the accurate clinical diagnosis of DLB is complex, with a relatively high rate of misdiagnosis 3; and third, because even the largest cohorts of DLB samples have been generally small, in many instances including as little as 100 patients 4,5. However, it is currently indisputable that DLB has a strong genetic component. The epsilon-4 allele of APOE 6,7 is recognized to be a strong risk factor, as are heterozygous mutations and common polymorphisms in the glucocerebrosidase gene (GBA)8. Both of these results have stemmed from candidate gene association studies; it was known that APOE was strongly associated with AD and GBA was a strong risk factor for PD/Lewy body disorders. In addition to these genetic associations with susceptibility, we have recently provided evidence that DLB has a heritable component 9. It has been shown that there is no overlap in common genetic risk between PD and AD 10, a fact that is not entirely surprising given the differences in phenotype. However, it is reasonable to hypothesize that the overlaps and differences in clinical and pathological presentation between DLB with both PD and AD stem, at least in part, from aspects in their underlying genetic architecture and, consequently, disease pathobiology. Specific genes/loci associated with disease as well as strength of association are factors that can be expected to modulate these phenotypic overlaps and differences. However, despite these encouraging findings, large-scale unbiased genetic studies in DLB have not yet been performed, which is likely due to the difficulty in identifying large, homogeneous cohorts of cases. To address the need for more powerful and comprehensive genetic studies of DLB, we performed the first large-scale genome-wide association study in this disease, using a total of 1,743 cases and 4,454 controls. The majority of cases were neuropathologically assessed, providing a greater level of diagnostic detail. Controls used were derived from two publicly available datasets and from the Mayo Clinic Florida control database. We performed imputation using the most recent imputation panel provided by the Haplotype Reference Consortium enabling us to have a detailed overview of common and intermediate frequency genetic variability.