Evaluation of performance of genoSCORETM for predicting onset of Alzheimer’s disease in a Japanese population

Abstract

AbstractBackgroundThe development of diagnostic tools to identify disease risk is critical to enable selection of suitable individuals for inclusion in clinical trials, cohort studies and ultimately to identify patients most likely to benefit from newly approved therapies. Cytox has developed and validated a polygenic risk scoring (PRS) algorithm, genoSCORETM, for predicting the Late Onset Alzheimer’s disease (AD) based on the work pioneered by Escott‐Price, Williams et al 1. However, both training and testing has been performed in study populations containing primarily individuals with Caucasian ethnicity. To ensure that genoSCORE can equally be applicable to all ethnic groups, we have conducted an initial evaluation to understand predictive accuracy for the onset of AD in a Japanese population.MethodsThe PRS approach used in this study is built from a sum of weighted contributions of over 100,000s SNPs associated with disease2. Due to the limited SNP coverage of existing genetic data, we only used 63,499 SNPs and utilized the core genoSCORE algorithm to calculate PRS in a cohort of 2,081 AD and 2,327 CN Japanese subjects from J‐ADNI3 and Niigata University.ResultsThe preliminary result shows an AUC of 0.659 (0.643‐0.675, 95% CI) between AD and CN subjects while PRS presents a bimodal distribution, but the main distribution trend is comparable to a normal distribution of PRS of the ADNI (US) subjects. Although this is significantly less than that seen for Caucasian individuals in the ADNI study, it compares well with the results generated when the reduced number of SNPs are used only in the ADNI cohort subjects (AUC = 0.691).ConclusionsThe preliminary results show the potential use of genoSCORE in the Japanese population. However, the results are likely affected by the limited SNP coverage from the available genetic data for the Japanese population. To determine the true performance of the genoSCORE algorithm in a Japanese population, genotyping samples from this study using our proprietary variaTECTTM II array providing full SNP coverage is in progress. We will then determine the performance and make adjustments to the algorithm if required in a Japanese population of suitably characterized Alzheimer’s disease patients and age‐matched healthy controls.