Penetrance estimation of SORL1 loss‐of‐function variants adjusted on APOE genotypes suggest a non‐monogenic inheritance

Abstract

AbstractBackground SORL1 rare loss‐of‐function (LoF) variants are a strong risk factor for Alzheimer disease (AD) with odds ratios above 30 for protein‐truncating variants (PTVs). Accordingly, some pedigrees suggest a high penetrance with individuals developing AD before 65 years. However, the age‐dependent penetrance of these variants is unknown, precluding an accurate use for genetic counselling or inclusion in preventive trials. Importantly, both rarity of these variants and co‐occurrence with other risk factors, including the common APOE4 allele, make penetrance estimations difficult based on classical strategies.MethodWe selected LoF SORL1 rare variants (PTVs and missense variants with a LoF effect following in vitro analyses) from exome sequencing data of >1,300 EOAD cases from France (Rouen‐CNRMAJ). To evaluate the age‐related penetrance of AD associated with SORL1 LoF variants, we first built a model based on family information. To ensure an accurate adjustment for APOE genotypes, our model combines a literature‐based baseline for SORL1 non carriers to an additive effect of SORL1 LoF variants computed from our pedigrees. Since pedigrees include relatives with missing genotypes, we applied an Expectation‐Maximization (EM) algorithm to take full advantage of the available information. Probands’ phenotype information was excluded to avoid ascertainment biases. We applied this first method to 34 pedigrees (69 affected relatives, 270 unaffected relatives, 45 relatives with available genotype information). Additionally, we used sequencing data from 7306 cases and 5852 controls to compute a population‐based estimate of the age‐dependent lifetime risk associated with both APOE4 allele and SORL1 rare PTVs or predicted damaging missense variants.ResultIn both models, 100% penetrance was reached before 75 years for carriers of SORL1 LoF variants who were also homozygous for APOE4. Penetrance remained incomplete for heterozygous E4 carriers, even at age 75, and was lower for non‐E4 carriers. We are providing penetrance estimation curves and 95%IC for the first time for SORL1 rare variants and at the digenic level.ConclusionOur results suggest that SORL1 rare LoF variants should be considered in a non‐monogenic model, different from APP and PSEN1/2 mutations. Such curves appear critical before these variants are used for genetic counseling and preventive clinical trials.