Marcio Almeida, PhD

Research Focus

Search for novel endophenotypes predicting risk of dementia in Mexican Americans

This study is designed to identify novel-invasive endophenotypes, biomarkers that index genetic risk for a disorder but which are not dependent on disease state. It will use extensive, available whole genome sequence, transcriptomic, and neurocognitive assessment data from 2000 members of large Mexican American families, living in San Antonio, Texas who were recruited as part of the Genetics of Brain Structure (GOBS) project, itself a component of the larger 35-year old San Antonio Family Study. Seventy of these individuals exhibit dementia, as diagnosed by clinical assessments and/or death certificate data. Extensive data also are available on socioeconomic status, educational history, physical activity, dietary behavior, acculturation, and variable related to the place of residence. 

Patterns of relatedness in these large extended pedigrees can be exploited to detect and characterize genetic contributions to variation in complex traits like ADRD risk; and, in fact, a large proportion of variation in ADRD risk in these families already is known to be due to the effects of genes. This will facilitate the selection of endophenotypes based on their genetic correlations with ADRD risk.

Like the disease state, the selected endophenotypes are expected to have important environmental determinants, and their genetic components are likely to interact with environmental factors. Therefore, the identification of informative endophenotypes will be followed by a systematic search for evidence of genotype-by-environment (GxE) interaction. The GxE interaction analyses will utilize the existing measures of socioeconomic, socio-behavioral, and socio-environmental domains as environmental indicators. An Endophenotype Ranking Value procedure, developed by study mentors [Drs. Blangero and Glahn], will be used to identify relevant transcriptomic endophenotypes. Formal tests for genotype-by-environment interactions will be done using variance component modeling approaches that include covariance kernels reflecting overall socio-environmental similarity between pairs of individuals.