The genomic architecture of the latent variable δ homolog (dEQ) in Mexican Americans and non-Hispanic whites

The latent variable δ homolog (dEQ) has been established as a reliable indicator of cognitive performance and a predictor of future dementia. Here, we sought to identify genetic variants underlying dEQ in both non-Hispanic White (NWH) and Mexican American (MA) populations, hypothesizing that novel genomic risk loci for dementia will be implicated in the MA cohort. Genotyping was performed on the TARCC cohort using the Illumina® MEGA Array, which includes ~1.7 million SNPs. dEQ was generated by Donald Royall and Ray Palmer (UT Health Science Center @ San Antonio) for the entire genotyped cohort (nNHW= 1572; nMA=1030). Association testing was conducted using PLINK (Purcell et al., 2007) according to standard procedures (Anderson et al., 2010). The primary genetic association with dEQ in NHW is the TOMM40/APOE locus (rs4420638, p=9.477x10-32); the association for this locus in the MA cohort was much less significant (p=3.886x10-5). Among MAs, three interesting and unique loci emerged as suggestive: XIRP2 (chr2, rs7595556 p=8.165x10-6), KIF13A (chr6, rs7766167, p= 3.404x10-6), and LINC00907 (chr18, rs237972, p=7.88x10-7). The loci discovered in the MA cohort have not previously been implicated in dementia/Alzheimer's disease; however, variants in these genes have been associated with metabolic phenotypes that are particularly relevant to the pathophysiology of cognitive decline in MAs: diabetic neuropathy, epigenetic aging, and childhood obesity. Further investigation of these genomic regions may illuminate mechanisms by which metabolic syndromes may confer risk for earlier onset of age-related cognitive decline in MAs.