Using Genetic Similarity Quantified by Kinship Coefficients to Investigate Familial Contributions to Reading Disorder

项目总结

PROJECT SUMMARY Decoding-based reading disorder (RD, or developmental dyslexia) is one of the most prevalent specific learning disorders in the population. Previous literature suggests multiple familial and environmental factors play a role in the manifestations of RD. RD individuals often have a family history; children from families where at least one first-degree relative exhibits history of the disorder have up to a sixfold increase in RD occurrence compared to controls. However, studies on RD that use family history (FH) measures typically use it as a binary categorical (family history present/absent) variable to examine group differences based on whether at least one relative has an official or likely diagnosis of RD. This means the spectrum of variation in FH is not captured. FH is typically determined based on parents and often does not include information from extended family (aunts/uncles, grandparents, cousins, etc.); this means important familial data that is less subject to the confounds of shared environment is being discarded. In addition, neuroimaging is valuable when used jointly with behavioral and FH measures, which may together improve early identification of RD; my advisor has shown that neuroimaging data reflects non-redundant metrics and mechanisms for behavior. The goal of this project is to determine the predictive ability of (1) continuous FH, and (2) FH from extended families in relation to the likelihood/severity of RD characteristics in children. In a large group of children (N = 841) ages 5-12.5 with varying levels of reading ability, I construct a novel factor, known as the RD kinship index (RDKI) which describes how many biologically related family members have likely had RD and how genetically close they are to the children. I will first replicate prior findings using categorical FH as a predictor for reading ability and cortical morphology (gray matter [GM] thickness and cortical surface area [SA] in reading-related regions) (Aim 1A). I will then replicate this analysis using a continuous measure of FH (Adult Reading History Questionnaire [ARHQ] scores from parents) as a predictor (Aim 1B), then compare the predictive ability of categorical vs. continuous FH on all outcomes in a single regression model (Aim 1C). I also propose using the novel RDKI to measure FH inclusive of extended family members and examine its ability to predict reading ability and structural neuroanatomy using analogous methods from Aims 1A-B (Aim 2A). The goal is to examine how additional familial factors for which the RDKI serves as a proxy may account for variance in outcomes; the utility of the RDKI will be compared to the model predictors from Aim 1 (Aim 2B). Finally, I will construct a supervised machine learning classifier trained on RDKI and GM thickness / cortical SA (Aim 3) to predict binary RD diagnoses. This proposal therefore aims to characterize the multifaceted relationship between familial predisposition to RD, diagnostic risk, and phenotypic severity, and links to its neural mechanisms. Results from this research will directly compare categorical and continuous FH measures and indicate the utility of the RDKI as an FH measure/early identifier for RD that can inform public policy.