Heritability of regional and global brain structure at the onset of puberty: a magnetic resonance imaging study in 9-year-old twin pairs

Abstract

Puberty represents the phase of sexual maturity, signaling the change from childhood into adulthood. During childhood and adolescence, prominent changes take place in the brain. Recently, variation in frontal, temporal, and parietal areas was found to be under varying genetic control between 5 and 19 years of age. However, at the onset of puberty, the extent to which variation in brain structures is influenced by genetic factors (heritability) is not known. Moreover, whether a direct link between human pubertal development and brain structure exists has not been studied. Here, we studied the heritability of brain structures at 9 years of age in 107 monozygotic and dizygotic twin pairs (N = 210 individuals) using volumetric MRI and voxel-based morphometry. Children showing the first signs of secondary sexual characteristics (N = 47 individuals) were compared with children without these signs, based on Tanner-stages. High heritabilities of intracranial, total brain, cerebellum, and gray and white matter volumes (up to 91%) were found. Regionally, the posterior fronto-occipital, corpus callosum, and superior longitudinal fascicles (up to 93%), and the amygdala, superior frontal and middle temporal cortices (up to 83%) were significantly heritable. The onset of secondary sexual characteristics of puberty was associated with decreased frontal and parietal gray matter densities. Thus, in 9-year-old children, global brain volumes, white matter density in fronto-occipital and superior longitudinal fascicles, and gray matter density of (pre-)frontal and temporal areas are highly heritable. Pubertal development may be directly involved in the decreases in gray matter areas that accompany the transition of our brains from childhood into adulthood.

Figure 1

Genetically influenced regional white matter density. (A) Superior occipitofrontal fascicle (Z = 36), (B) superior occipitofrontal fascicle, superior longitudinal fascicle (X = 27), (C) superior longitudinal fascicle (Y = −22). Images are according to neurological convention (left = left). The left side displays heritability estimates. The right side displays significant genetic effects in χ²‐values, overlaid on the model‐brain: CE versus ACE model (critical level of significance is 17.7, corrected for multiple comparisons according to the False Discovery rate, α = 0.05). For visualization purposes, this threshold is relaxed to an FDR rate of α = 0.15 (corresponding to a χ²‐value of 10.5). χ²‐maps are resampled to anatomical resolution for overlap with anatomical boundaries.

Figure 2

Genetically influenced regional gray matter density. (A) Superior frontal gyrus (Z = 57), (B) middle temporal gyrus (X = 55), (C) amygdala (Y = −9). Images are according to neurological convention (left = left). The left side displays heritability estimates. The right side displays significant genetic effects in χ²‐values, overlaid on the model‐brain: CE versus ACE model (critical level of significance is 20.5, corrected for multiple comparisons according to the False Discovery rate, α = 0.05). For visualization purposes, this threshold is relaxed to an FDR rate of α = 0.15 (corresponding to a χ²‐value of 12.2). χ²‐maps are resampled to anatomical resolution for overlap with anatomical boundaries.

Figure 3

Decreases in (pre)frontal and parietal gray matter density in pubertal children compared with nonpubertal children at 9 years. Shown here are χ²‐values of the worsening in fit of a (means) model with Tanner status versus without Tanner status, corrected for sex and handedness. Level of significance: χ² > 3.84 (α < 0.05, uncorrected for exploratory purpose). To measure decreases or increases in density, χ²‐maps were multiplied by negative and positive regression maps, respectively. Significant reductions in gray matter density are mainly located in the (pre)frontal cortex [A (X = 2), B (Y = 60), and C (Z = 41)] and parietal cortex (A and C) bilaterally with χ²‐values up to 18.5. Images are according to neurological convention (left = left) χ²‐maps are resampled to anatomical resolution for overlap with anatomical boundaries.