Premium
Phenome‐wide associations with an Alzheimer’s disease genetic risk score in UK Biobank: Identifying early indicators of disease
Author(s) -
Zimmerman Scott C,
Ackley Sarah F,
Brenowitz Willa D,
Graff Rebecca E,
Glymour M Maria
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.040114
Subject(s) - phenome , biobank , disease , dementia , bonferroni correction , medicine , neurocognitive , single nucleotide polymorphism , gerontology , bioinformatics , phenotype , biology , cognition , genotype , genetics , psychiatry , gene , statistics , mathematics
Background There is a pressing need to identify early indicators of Alzheimer’s disease and Alzheimer’s disease‐related dementias (AD/ADRD) to understand the time course of disease and help target potential secondary prevention measures. A phenome‐wide association study (PHeWAS) could be used to rapidly screen numerous possible leading indicators of AD/ADRD based on phenotypes predicted by an Alzheimer’s Disease genetic risk score (ADGRS). We conducted a preliminary PheWAS to evaluate associations with an ADGRS, considering multiple phenotypic domains relevant to AD/ADRD. Method We calculated an ADGRS as a weighted sum of 23 previously identified AD‐related single nucleotide polymorphisms in individuals of European ancestry. We conducted a PheWAS of 373,225 UK Biobank participants of European ancestry aged 50‐69 years at enrollment. Using the open‐source tool PHEnome Scan ANalysis Tool ( github.com/MRCIEU/PHESANT ), we tested the ADGRS against 434 phenotypes: 183 neuroimaging, 100 cognitive, 84 mental health, 31 psychosocial, and 14 behavioral. PHESANT regresses the ADGRS on each phenotype adjusted for baseline age, sex, genotype chip, and 10 genetic ancestry principal components. Result Using a global Bonferroni threshold (p<1.15E‐4) significance criterion, we found 34 phenotype‐ADGRS associations. The three most significant associations were with cognitive functioning from a symbol digit substitution test: Duration to entering value (0.029, 95%CI=[0.023, 0.035]; p=2.58E‐06), number of matches made correctly (‐0.029, 95%CI=[‐0.035,‐0.022]; p=7.56E‐05), and number of matches attempted (‐0.028, 95%CI=[‐0.033,‐0.021]; p=2.63E‐08). Overall, 11 behavioral (69%), 21 cognitive (21%), 1 psychosocial (2%), and 1 mental health (2%) variables were significantly associated with the ADGRS. None of the neuroimaging variables were significant, but sample sizes were small (18,314‐20,827). Using category‐specific Bonferroni‐adjustments, 13 behavioral (81%), 23 cognitive (23%), 2 psychosocial (2%), and 2 mental health (2%) variables were significant. Conclusion In a largely healthy and cognitively normal sample, the PheWAS identified cognitive assessments that may be leading indicators of AD/ADRD and identified novel associations of ADGRS with 1 mental health, 1 psychosocial, and 11 behavioral variables. PHeWAS using AD/ADRD genetic profiles may have potential for identifying early indicators of AD/ADRD.