White matter but not grey matter predicts change in reading skills after intervention

Abstract

This study examined changes in white matter microstructure and grey matter volume, cortical thickness, and cortical surface area before and after reading intervention. Participants included 22 average readers and 13 dyslexic readers (8–9 years old in third grade); the dyslexic readers were enrolled in reading intervention programs at their elementary school. Participants completed scans of diffusion tensor imaging and T1‐weighted MRI before and after 3 months of instruction. An a priori region of interest (ROI) analysis was used. Dyslexic readers, compared to average readers, showed higher mean diffusivity in white matter ROIs including bilateral inferior frontal, bilateral insula, left superior temporal, and right supramarginal gyri across time points. Dyslexic readers also had thicker cortex in left fusiform and bilateral supramarginal gyri; whereas, average readers had greater surface area in right fusiform across time. There were no significant changes in white or grey matter following intervention; however, mean diffusivity in the right hemisphere was associated with reading gains over time. White matter organization in the right hemisphere predicts reading changes, and dyslexic readers may have persistent differences in white and grey matter due to ongoing reading deficits.

Genetic and Demographic Predictors of Latent Reading Ability in Two Cohorts

“…We used a linear regression model to assess the fit of Analysis_1 and Analysis_2. Analysis_1 fit the data significantly better than a null model (F (1, 83) = 8.22, p < .0001) and explained roughly 14 percent of the variance in reading ability (Adjusted R 2 = 0.141). The positively associated SNPs were located within 16 genes with the majority representing intron variants.…”

“…Analysis_2 fit the data significantly better (F (1, 75) = 10.07, p < .0001) than a null model and explained roughly 14 percent of the variance in the reading score (Adjusted R 2 = 0.138). The positively associated SNPs were located within 11 genes while the negatively associated SNPs were located within 24 genes.…”

“…We used robust modeling to account for missing data. Two goodness-of-fit metrics were highly acceptable (TLI = 0.909, SRMS = 0.031), whereas chi-square ( χ 2 (df = 20) = 730.21, p < .001) and RMSEA (RMSEA = 0.159, 95% confidence interval = 0.149, 0.168, p < .001) did not indicate good fit. The standardized path values ranged from 0.528 (CTOPP – Blending Words) to 0.913 (SRI WR).…”

“…We fitted a single factor model to the reading variables after z-score transformation. Two measures of goodness of fit indicated highly acceptable fit (TLI = 0.925, SRMR = 0.033), whereas chi-square ( χ 2 (df = 27) = 1307.97, p < .001) and RMSEA (RMSEA = 0.124, 95% confidence interval = 0.118, 0.131, p = < .001) did not meet the guidelines for fit. All manifest paths were significant and had standardized path values over 0.70 (range = 0.745, 0.929).…”

“…Our findings also indicate high generalizability of genetic findings between cohorts, using our approach. Efficient and adequate reading is the result of several neurological processes and pathways 1,2 that enable critical component skills, which include phonological awareness 3,4 , decoding 5,6 , fluency 7,8 , and comprehension. There is substantial evidence that reading ability has a strong genetic component [9][10][11] .…”