Identifying individuals with high risk of Alzheimer's disease using polygenic risk scores

Abstract

Polygenic Risk Scores (PRS) for AD offer unique possibilities for reliable identification of individuals at high and low risk of AD. However, there is little agreement in the field as to what approach should be used for genetic risk score calculations, how to model the effect of APOE, what the optimal p-value threshold (pT) for SNP selection is and how to compare scores between studies and methods. We show that the best prediction accuracy is achieved with a model with two predictors (APOE and PRS excluding APOE region) with pT<0.1 for SNP selection. Prediction accuracy in a sample across different PRS approaches is similar, but individuals' scores and their associated ranking differ. We show that standardising PRS against the population mean, as opposed to the sample mean, makes the individuals' scores comparable between studies. Our work highlights the best strategies for polygenic profiling when assessing individuals for AD risk.

Fig. 1. Effects of APOE allele frequencies and age on genetic risk scores.

A Allele ε4, ε3, ε2 frequencies (red, orange and green lines respectively) in the case-control dataset (271 cases and 278 controls), B, C the mean risk score for ε4 allele frequencies, ORS.no.APOE and PRS.no.APOE (red, green and blue lines respectively) split by age groups. ORS.no.APOE includes SNPs with p-value≤ 1e-5, PRS.no.APOE includes SNPs with p-value ≤ 0.1 and both exclude the APOE region. A represents the full sample (B) represents cases only and (C) controls only. Age groups are specified as (55–65, 65–75, 75–85, 85+). For comparability of the scores in (B, C) the e4 genotypes (originally coded as 0/1/2) were also standardised. ORS Oligogenic risk score, PRS Polygenic risk score, SNP Single nucleotide polymorphism.

Fig. 2. Prediction accuracy across different PRS methods (PRS(C + T), PRSice, LDpred-Inf, PRS-CS, LDAK and SBayesR) for ORS.full, ORS.no.APOE, PRS.full, PRS.no.APOE and PRS.AD.

Bar plot for prediction accuracy (AUC and R²) across 6 PRS approaches: PRS(C + T), PRSice, LDpred-Inf, PRS-CS, LDAK and SBayesR (red, yellow, green, teal, blue, pink bars respectively). The colour of each PRS method is consistent across all plots. Upper figures represent ORS and lower figures represent PRS and PRS.AD models in the case-control dataset (271 cases and 278 controls). ORS.full includes SNPs with pT ≤ 1e-5 and PRS.full includes SNPs with pT ≤ 0.1, ORS.no.APOE and PRS.no.APOE exclude SNPs in the APOE region and PRS.AD models APOE separately and subsequently adds this to PRS.no.APOE. AUC Area Under the Curve, ORS Oligogenic risk score, PRS Polygenic risk score, AD Alzheimer’s Disease, SNP Single nucleotide polymorphism.