Evaluation of SORL1 in Lewy body dementia identifies no significant associations

Lewy body dementia (LBD) is a clinically heterogeneous neurodegenerative disorder characterized by parkinsonism, visual hallucinations, fluctuating mental status, and REM sleep behavior disorder^1,2. LBD lies along a spectrum between Parkinson’s disease and Alzheimer’s disease, and recent evidence suggests that the genetic architectures of these age-related syndromes are intersecting³.

Numerous investigations into the role of amyloid precursor protein (APP) pathway genes have implicated an intracellular transmembrane protein, sortilin-related receptor 1 (encoded by the SORL1 gene), to be associated with an increased risk of Alzheimer’s disease. Pathogenic loss-of-function and missense mutations in SORL1 have been postulated to affect APP shuttling between the trans-Golgi network and early endosomes, leading to disease⁴. Interestingly, several studies have also found rare SORL1 mutations in patients with Alzheimer’s disease with concomitant LBD, suggesting that SORL1 may be a pleiotropic risk gene^4–6. Here, we assessed the possible association of SORL1 variants with risk for LBD.

We used our previously published whole-genome sequencing data generated on 2,591 European-ancestry LBD patients and 4,027 neurologically healthy individuals³, which has good coverage of the SORL1 gene (mean coverage 38x, range=22x–83x). Patients with LBD included in this dataset were diagnosed with pathologically definite or clinically probable disease according to consensus criteria.^1,3 Controls were selected based on a lack of evidence of cognitive decline in their clinical history and absence of neurological deficits on neurological examination. We first reviewed a common SORL1 risk variant for Alzheimer’s disease (rs74685827) in our previous LBD GWAS data,³ which found no association (beta=0.02, standard error=0.14, p-value=0.90). Next, we extracted 6,617 rare variants in SORL1 (minor allele frequency ≤0.01 in the European, non-Finnish population; Supplementary Table 1) and annotated them with variant effect predictor⁷, then filtered the data to only include loss-of-function and missense mutations. This step identified 167 variants on which we performed a gene burden analysis using SKAT-O, as implemented in RVTests (v.2.1.0)⁸. The model included age, sex, and four principal components as covariates. We also explored the rare single-variant association tests for each observed missense and loss-of-function mutation using Fisher’s exact tests, as implemented in PLINK (v.1.9) (Supplementary Table 2).

Our gene burden analysis did not identify a significant association of rare damaging variants with an increased risk of developing LBD (SKAT-O p-value=0.24). Additionally, none of the single-variant associations were significant after correction for multiple testing (Supplementary Table 2). We observed eight missense mutations that have been previously described in patients with mixed dementias^4–6. However, these mutations were either found in cases and controls or were more frequent in controls. None of these variants were nominally associated with disease (Table 1). Furthermore, a recent report examining SORL1 mutations in a large Greek family with Parkinson’s disease, Parkinson’s disease dementia and mixed dementia reported a rare missense variant (p.G379W) segregating with disease⁶. We observed the same mutation in one LBD patient of Greek ancestry. From our data, we cannot conclude that this mutation is disease-causing or simply a variant present in the Greek population.

Table 1.

Single-variant associations of rare SORL1 mutations reported in the literature in patients with dementia

Position (hg38)	REF	ALT	dbSNP	AA Change	gnomAD MAF (all population)	gnomAD MAF (NFE)	No. of LBD Patients	No. of Controls	p-value	Previous Reports
chr11:121514222	A	C	rs150609294	p.N371T	0.0014	0.0031	8	8	0.2458	Alzheimer’s disease with FTLD
chr11:121514245	G	T	-	p.G379W	-	-	1	0	0.3915	Parkinson’s disease and mixed dementia
chr11:121550604	G	A	rs148430425	p.D734N	0.0006	0.0002	0	3	0.2853	Alzheimer’s disease with DLB
chr11:121558767	C	T	rs143571823	p.T947M	0.0042	0.00011	0	1	1	Late-onset Alzheimer’s disease with DLB
chr11:121586253	C	G	rs1699102	p.N1246K	0.00002	0.00005	0	1	1	Alzheimer’s disease
chr11:121595662	G	A	rs776117825	p.R1470Q	0.00001	0.00004	0	2	0.5235	Alzheimer’s disease with DLB
chr11:121625107	A	T	rs140327834	p.D2065V	0.0022	0.0037	25	42	0.8029	Alzheimer’s disease, DLB, FTLD
chr11:121627591	C	T	rs142884576	p.T2134M	0.0005	0.0008	4	5	0.7441	Early-onset Alzheimer’s disease with DLB

Rare loss-of-function and missense mutations in SORL1 reported in patients with mixed dementia. The minor allele frequencies (MAF) in non-Finnish European ancestry (NFE) population are reported from the Genome Aggregation Database v3.1.2 (gnomAD)(https://gnomad.broadinstitute.org/). Abbreviations: reference allele (REF), alternate allele (ALT), amino acid change (AA change), Frontotemporal dementia (FTLD), Dementia with Lewy bodies (DLB).

In summary, we did not find a significant enrichment of rare, damaging SORL1 mutations in our well-powered LBD cohort. Our dataset is, to our knowledge, the largest genome-sequence cohort in this understudied disease. While it is possible that an association was missed due to allelic heterogeneity, our findings indicate that caution should be exercised when interpreting SORL1 mutations in LBD, as current evidence does not conclusively support an association with disease risk.