Admixture mapping of cognitive function in diverse Hispanic and Latino adults
Myriam Fornage, Tamar Sofer, Claudia L. Satizabal, Sudha Seshadri, Thomas H. Mosley, Charles Decarli, Wassim Tarraf, Timothy A Thornton, Hector M González- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
Genome‐wide association studies (GWAS) conducted primarily in populations of European ancestry have revealed the polygenic nature of cognitive abilities [1, 2] but genetic discoveries in other populations, including Hispanics/Latinos, are lagging. Genetically, US Hispanics/Latinos have uniquely admixed genomes, encompassing African, Amerindian, and European ancestries.[3] Admixture patterns in genomic regions (local ancestry) can be leveraged in gene mapping when frequencies of trait‐associated genetic variants located within these regions differ between ancestries.
Method
We conducted genome‐wide admixture mapping analyses followed by fine mapping to identify genetic variants associated with cognitive function in 7,140 Hispanic and Latino adults (mean age 55 years; 60.5% women) from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). Five measures of cognitive function were investigated: Six‐Item Screener (SIS), Brief Spanish English Verbal Learning Test (SEVLT), Word Fluency Test (WFT), Digit Symbol Substitution Test (DSST), and a measure of general cognitive function (PC1). We tested the identified variants for replication in independent samples with similar neurocognitive measures.
Result
We identified 9 significant local ancestry‐associated regions for the 5 cognitive tests. Counts of African ancestry on at 13q12.11, 4q14, 10p12.2 and 9p22.1 were associated with DSST, PC1, SIS, and WFT, respectively. Counts of Amerindian ancestry at 1q25.2 were associated with DSST; and at 12q14.2 and 12q15 with SIS. Counts of European ancestry at 8p22 were associated with SEVLT; and at 13q12.13 with, both, DSST and PC1. Fine mapping via SNP association testing in each region identified genetic variants that partially or totally explained the admixture signal. While statistical evidence of independent replication of the prioritized variants was generally weak, with only two replicated associations at chr8p22 and chr13q12.13, there was strong biological support for the observed associations to cognitive function. For example, the prioritized variant at 8q22 is an eQTL for CNOT7, which regulates synaptic plasticity and is essential for hippocampal‐dependent learning and memory [4].
Conclusion
Our admixture mapping analyses identified multiple novel loci associated with measures of cognitive function. This demonstrates the effectiveness of admixture mapping in complementing GWAS [5] and provides an opportunity to identify genetic loci that may be particularly relevant in underrepresented populations.