Fiber-specific structural properties relate to reading skills in children and adolescents

Abstract

Recent studies suggest that the cross-sectional relationship between reading skills and white matter microstructure, as indexed by fractional anisotropy, is not as robust as previously thought. Fixel-based analyses yield fiber-specific micro- and macrostructural measures, overcoming several shortcomings of the traditional diffusion tensor model. We ran a whole-brain analysis investigating whether the product of fiber density and cross-section (FDC) related to single-word reading skills in a large, open, quality-controlled dataset of 983 children and adolescents ages 6-18. We also compared FDC between participants with (n = 102) and without (n = 570) reading disabilities. We found that FDC positively related to reading skills throughout the brain, especially in left temporoparietal and cerebellar white matter, but did not differ between reading proficiency groups. Exploratory analyses revealed that among metrics from other diffusion models - diffusion tensor imaging, diffusion kurtosis imaging, and neurite orientation dispersion and density imaging - only the orientation dispersion and neurite density indexes from NODDI were associated (inversely) with reading skills. The present findings further support the importance of left-hemisphere dorsal temporoparietal white matter tracts in reading. Additionally, these results suggest that future DWI studies of reading and dyslexia should be designed to benefit from advanced diffusion models, include cerebellar coverage, and consider continuous analyses that account for individual differences in reading skill.

Keywords: DWI; children; dyslexia; fixels; human; neuroscience; reading.

Figure 1.. Methodological overview of the study.

Top: description of primary and secondary analyses. Bottom: schematic depicting interpretations of changes in examined metrics. Depictions of bundles, axons, and neurites are not drawn to scale. DWI, diffusion-weighted imaging; DTI, diffusion tensor imaging; DKI, diffusion kurtosis imaging; NODDI, neurite orientation density and dispersion index; FA, fractional anisotropy; KFA, kurtosis fractional anisotropy; MD, mean diffusivity; MK, mean kurtosis; NDI, neurite density index; ODI, orientation dispersion index; FODF, fiber orientation distribution function; FD, fiber density; FC, fiber cross-section; FDC, fiber density and cross-section product.

Figure 2.. Age-standardized TOWRE subscores of all participants.

Each dot represents a participant, color-coded by group assignment. Dashed lines mark the score cutoffs for the two reading proficiency groups. Since scores are discrete and not unique, some dots may overlap with each other. Kernel density estimation plots along the perimeter show the distribution of reading scores in each group. TR, typically reading group; RD, reading disability group; TOWRE, Tests of Word Reading Efficiency.

Figure 2—figure supplement 1.. Correlations between continuous phenotypic and neuroimaging variables.

Correlation coefficients are reported as Spearman’s $\rho$. p-values were FDR Benjamini–Hochberg adjusted across tests. *p<0.05; **p<0.001; ***p<1e-5. TOWRE, Tests of Word Reading Efficiency composite score, age-normalized; WISC VSI, Wechsler Intelligence Scale for Children visuospatial index, age-normalized; WISC VCI, Wechsler Intelligence Scale for Children verbal comprehension index, age-normalized; SES, socioeconomic status; ICV, intracranial volume; N Corr, neighbor correlation; gFD, globally averaged fiber density; gFC, globally averaged fiber cross-section.

Figure 2—figure supplement 2.. ANOVA results for site-wise comparisons between phenotypic and neuroimaging metrics.

Only metrics associated with a significant between-sites ANOVA (p<0.05) are plotted. For t-tests between sites, *p<0.05, **p<0.01, ***p<1e-03, and ****p<1e-04. SES, socioeconomic status; ICV, intracranial volume; WISC VSI, Wechsler Intelligence Scale for Children visuospatial index, age-normalized; WISC VCI, Wechsler Intelligence Scale for Children verbal comprehension index, age-normalized; N Corr, neighbor correlation; gFD, globally averaged fiber density; gFC, globally averaged fiber cross-section.

Figure 3.. Significant fixels (qFDR<0.05) for relating fiber density and cross-section product (FDC) to raw composite Tests of Word Reading Efficiency (TOWRE) scores, colored by effect size (Δ⁢Ra⁢d⁢j2).

Model confounds included a spline fit for age and linear fits for sex, site, neighbor correlation, and log(ICV). Top and bottom panels are left and right hemispheres, respectively. Sagittal slices go from lateral-to-medial. Blue arrows point to larger clusters of fixels in bilateral temporoparietal and cerebellar white matter that were associated with higher effect sizes relative to fixels in the rest of the hemisphere. The template fiber orientation distribution (FOD) image was used as the background image.

Figure 3—figure supplement 1.. Significant fixels (qFDR<0.05) relating fiber density and cross-section product (FDC) to raw composite Tests of Word Reading Efficiency (TOWRE) scores, colored by the beta estimates (top) and direction (bottom; red, LR; green, AP; blue, SI).

Model confounds included a spline fit for age and linear fits for sex, site, neighbor correlation, and log(ICV). Only the left hemisphere is shown. Sagittal slices go from lateral-to-medial. The template fiber orientation distribution (FOD) image was used as the background image.

Figure 3—figure supplement 2.. Plots of the set of tracts in which the strongest effect sizes (ΔRadj2>0.028) were achieved for relating fiber density and cross-section product (FDC) to Tests of Word Reading Efficiency (TOWRE) scores (see Table 2).

All tracts were in the left hemisphere. Sagittal slices go from lateral-to-medial. Tracts were segmented from and are plotted on top of the fiber orientation distribution (FOD) template.

Figure 3—figure supplement 3.. Significant fixels (qFDR<0.05) relating fiber cross-section (FC; top), and fiber density (FD; bottom) to raw composite Tests of Word Reading Efficiency (TOWRE) scores, colored by direction (red, LR; green, AP; blue, SI).

Model confounds included a spline fit for age and linear fits for sex, site, and neighbor correlation. Additionally, FC included an additional regressor for log(ICV). Only the left hemisphere is shown. Sagittal slices go from lateral-to-medial. Blue arrows point to larger clusters of significant fixels in temporoparietal and cerebellar white matter that overlapped with significant results in the main fiber density and cross-section product (FDC) analysis. The template fiber orientation distribution (FOD) image was used as the background image.

Figure 3—figure supplement 4.. Significant fixels (qFDR<0.05) relating fiber density and cross-section product (FDC) to raw Sight Word Efficiency (SWE; top) and Phonemic Decoding Efficiency (PDE; bottom) subscores, colored by effect size (Δ⁢Ra⁢d⁢j2).

Model confounds included a spline fit for age and linear fits for sex, site, neighbor correlation, and log(ICV). Only the left hemisphere is shown. Sagittal slices go from lateral-to-medial. The template fiber orientation distribution (FOD) image was used as the background image.

Figure 4.. Significant fixels (qFDR<0.05) for relating neurite orientation density and dispersion index (NODDI) metrics to raw composite Tests of Word Reading Efficiency (TOWRE) scores, colored by direction (red, LR; green, AP; blue, SI).

Model confounds included a spline fit for age and linear fits for sex, site, neighbor correlation, and log(ICV). Top and bottom panels are the indexes for orientation dispersion (ODI) and neurite density (NDI), respectively. Only the left hemisphere is shown. Sagittal slices go from lateral-to-medial. Blue arrows and circles indicate significant fixels. The template fiber orientation distribution (FOD) image was used as the background image.