SORL1 genetic variants and Alzheimer disease risk: a literature review and meta-analysis of sequencing data.

Massive parallel sequencing recently allowed the identification of three genes carrying a higher burden of rare, protein-truncating and missense predicted damaging variants in Alzheimer disease (AD) cases as compared to controls: TREM2, SORL1, and ABCA7. SORL1 encodes SorLA, a key protein involved in the processing of the amyloid-beta (Aβ) precursor protein (APP) and the secretion of the Aβ peptide, the aggregation of which triggers AD pathophysiology. Common SORL1 single nucleotide polymorphisms had originally been associated with AD with modest odds ratios (ORs). The association of AD with rare SORL1 coding variants has been demonstrated at the gene level by aggregating protein-truncating (PTV) and rare predicted damaging missense variants. In addition to the loss of SorLA function induced by PTVs, a few missense variants were studied in vitro, showing diverse degrees of decreased SorLA function and leading to increased Aβ secretion. However, the exact functional consequences of most of the missense variants remain to be determined as well as corresponding levels of AD risk. Hereby we review the evidence of the association of SORL1 common and rare variants with AD risk and conduct a meta-analysis of published data on SORL1 rare variants in five large sequencing studies. We observe a significant enrichment in PTVs with ORs of 12.29 (95% confidence interval = [4.22-35.78]) among all AD cases and 27.50 [7.38-102.42] among early-onset cases. Rare [minor allele frequency (MAF) < 1%] and ultra-rare (MAF < 10-4) missense variants that are predicted damaging by 3/3 bioinformatics tools also show significant associations with corresponding ORs of 1.87 [1.54-2.28] and 3.14 [2.30-4.28], respectively. Per-domain analyses show significant association with the APP-binding CR cluster class A repeats and the Aβ-binding VPS10P domains, as well as the fibronectin type III domain, the function of which remains to be specified. These results further support a critical role for SORL1 rare coding variants in AD, although functional and segregation analyses are required to allow an accurate use in a clinical setting.