A/T/N polygenic risk score for cognitive decline in old age

Abstract

Research in Context Systematic ReviewAuthors reviewed relevant literature using PubMed and Google Scholar. Key studies that generated and validated polygenic risk scores (PRS) for clinical and pathologic AD were cited. PRS scores have been increasingly used in the literature but clinical utility continues to be questioned. InterpretationIn the current research landscape concerning PRS clinical utility in the AD space, there is room for model improvement and our hypothesis was that a PRS with integrated risk for AD biomarkers could yield a better model for cognitive decline. Future DirectionsThis study serves as proof-of-concept that encourages future study of integrated PRS across disease markers and utility in taking an A/T/N (amyloidosis, tauopathy and neurodegeneration) focused approach to genetic risk for cognitive decline and AD. Abstract INTRODUCTION: We developed a novel polygenic risk score (PRS) based on the A/T/N (amyloid plaques (A), phosphorylated tau tangles (T), and neurodegeneration (N)) framework and compared a PRS based on clinical AD diagnosis to assess which was a better predictor of cognitive decline. METHODS: We used summary statistics from genome wide association studies of cerebrospinal fluid amyloid-β (Aβ42) and phosphorylated-tau (ptau181), left hippocampal volume (LHIPV), and late-onset AD dementia to calculate PRS for 1181 participants in the Alzheimer's Disease Neuroimaging Initiative (ADNI). Individual PRS were averaged to generate a composite A/T/N PRS. We assessed the association of PRS with baseline and longitudinal cognitive composites of executive function and memory. RESULTS: The A/T/N PRS showed superior predictive performance on AD biomarkers and executive function decline compared to the clinical AD PRS. DISCUSSION: Results suggest that integration of genetic risk across AD biomarkers may improve prediction of disease progression. Boldface indicates P < 0.05. *P<1.25E-03 Bonferroni threshold. N=1,182 unless noted otherwise. No APOE denotes PRS excluding APOE region results. NC/MCI and NC indicate sensitivity analysis results with the denoted diagnostic groups (as assessed at baseline).

Genetic propensity for cerebral amyloidosis and risk of mild cognitive impairment and Alzheimer's disease within a cognitive reserve framework

“…Previous studies have replicated the association of variants used for PRS construction and validated the relationship of the GWAS with CSF amyloid levels in independent cohorts, providing strong evidence for their true association. 18,51 Our study corroborates the ability of PRS for amyloid to predict AD/aMCI risk in a genetically different group than previous studies' population.…”

“…This is in accordance with previous studies that used summary statistics from the GWAS of Deming et al 28 to derive a PRS for amyloid. 18,51 The PRS was constructed excluding the APOE region, suggesting that genetic variation beyond APOE accounts for a substantial proportion of AD/aMCI genetic risk. Indeed, the addition of PRS in the adjusted model with APOE as main predictor increased the proportion of variance explained by 1.1%.…”

“…Previous studies that used summary statistics from the GWAS of Deming et al 28 to construct PRS for amyloid, excluding the APOE region, reported that PRSs explained 1.8%-4.8% of the variance in CSF Aβ42 levels. 18,51 We acknowledge that the predictive ability of a PRS alone is low and it will likely not replace actual pathology measurements in clinical care, and a more nuanced approach combining PRS, plasma-based biomarkers, and lifestyle/environmental factors may be required as more accurate pathology surrogates. 57 Furthermore, genetic predisposition to amyloid accumulation was determined based on the effects of common variants identified in GWASs for CSF Aβ42, while other biological factors such as rare variants, haplotypes, epigenetic elements, and biomarker signatures known to influence or predict amyloid burden were not taken into account.…”

“…14 The heritability of AD and amnestic mild cognitive impairment (aMCI) is particularly high, ranging between 58%-79% 15,16 and 40%-48%, 17 respectively, and polygenic risk score (PRS) analysis was used to predict AD/aMCI risk. 18 Considering the high heritability of ADrelated amyloidosis, explaining around 73% of plasma Aβ42 levels, 19 we investigated the CR hypothesis using as a proxy for AD pathology a PRS constructed for amyloid brain deposition. As a proxy, we define a metric reflecting increased susceptibility to brain amyloid accumulation.…”

Associations of Polygenic Risk Score for Late-Onset Alzheimer's Disease With Biomarkers

“…AD-PRS also showed a significant correlation with the A/T/N profiles (Ebenau et al, 2021 ). A study found that the integration of genetic risk across the AD biomarkers like A/T/N may improve the prediction of the disease progression (Moore et al, 2019 ).…”