In‐depth evaluation of protein domains improves classification of SORL1 variant pathogenicity in Alzheimer’s Disease patients

Abstract

AbstractBackgroundRare pathogenic SORL1 gene variants have been identified as strong risk‐increasing factors in Alzheimer’s disease (AD). Recent analysis indicated that SORL1 loss‐of‐function (LoF) variants associated with a ∼40‐fold increased risk for early‐onset AD, while missense variants that were prioritized using in silico variant pathogenicity‐prediction algorithms associated only with a 2‐fold risk increase. However, current risk‐prediction strategies do not take SORL1‐specific features into account. We aimed to design a variant classification strategy that adds in‐depth knowledge of SORLA structural folding and function to the variant risk prediction.MethodWe identified coding SORL1 variants in a discovery/replication sample of whole exome sequencing data from the ADES/ADSP studies comprising 18,959 cases and 21,893 controls after quality control. We selected only the extremely rare variants (MAF<0.05%) and further divided these into LoF(n = 77) and missense variants(n = 539). With an in‐depth evaluation of protein functional domains (informed by pathogenic variants in homologous genes) combined with their predicted pathogenicity (REVEL algorithm) we prioritized missense variants into ‘high priority’, ‘moderate priority’ and ‘low priority’. We compared the age‐dependent penetrance of missense variants for each priority‐category and LoF variants, in context of APOE genotype.ResultCarriers of prioritized missense variants had a similar age‐dependent penetrance as LoF variants, with median age at AD‐onset of ∼65 years which is significantly lower compared to non‐carriers (∼72 years). In addition, carriers of prioritized SORL1 missense variants have an expedited age at onset by ∼7 years independent of their APOE genotype. If we look at a 80% penetrance level, SORL1 carriers with homozygous ε4 background had AD by age 61 (95%CI 60‐NA, n = 6) compared to age 73 (95%CI 72‐73, n = 1091) for SORL1 WT carriers. Carriers with a heterozygous background had AD by age 71 (95%CI 68‐76, n = 62) compared to 78 (95%CI 78‐78, n = 4816) for WT. For a non‐ε4 background the age at onset for carriers was 83 years (95%CI 78‐90, n = 54) compared to 85 years (95%CI 85‐85, n = 5773) for WT. The moderate and low priority group showed no significant difference compared to WT.ConclusionOur prioritization model can successfully identify likely pathogenic missense SORL1 variants, allowing for better identification of patients with possible SORL1‐associated‐AD.