Fetal temporal sulcus depth asymmetry has prognostic value for language development

The study explores whether fetal asymmetry in the depth of the superior temporal sulcus (STS)—a robust, early-emerging structural brain feature—predicts later language development. Prior work shows humans uniquely exhibit a deeper right STS and longer left STS, detectable already in fetuses around the second trimester. Given the left temporal cortex’s role in speech perception and language processing, the authors hypothesized that fetal STS depth asymmetry relates to subsequent language abilities, language lateralization/localization, and organization of language-related white matter tracts. The research addresses a key gap by employing a longitudinal design from fetal MRI to language outcomes in childhood, aiming to identify an early anatomical biomarker with potential clinical utility for fetal counseling and prognosis of language development.

The article reviews cross-species and human evidence of structural and functional brain asymmetries. In mammals and primates, hemispheric differences exist, with humans showing distinctive temporal lobe asymmetries: a larger left planum temporale and deeper right STS. Fetal and neonatal studies demonstrate that right STS develops earlier and is deeper in over 90% of fetuses, suggesting prenatal origins of temporal asymmetry. Prior cross-sectional studies linked temporal asymmetry to language skills in children and to language deficits in dyslexia, autism, and specific language impairment. However, evidence from longitudinal cohorts following individuals from fetal stages to childhood has been lacking. Genetic studies suggest polygenic influences on STS asymmetry that are largely independent of visceral laterality. Functional imaging work implicates left STS regions in speech/phonology/semantics and right STS in social cognition, and some studies show anatomo-functional correspondence between STS morphology and auditory voice processing in adults.

Design: Preregistered, prospective longitudinal study of children (ages 6–13) whose fetal MRIs were normal. Fetal MRI data (1.5T Philips Ingenia) acquired 6–13 years prior were reanalyzed; current assessments included neuropsychological language testing, fMRI language localization, and DTI for white matter tracts. Participants: From 155 eligible records, 38 families consented. Inclusion: normal/corrected vision and hearing; no neurological/psychiatric disease; no MR contraindications; both parents native German speakers. Final analytic sample for fetal STS quantification: 29 children (STS visible from ≥25 weeks gestation). Ethics approved; informed consent/assent obtained. Fetal MRI acquisition: 1.5T; T2-weighted turbo spin-echo sequences in 3 planes (in-plane 0.62–1.17 mm; slice 2.0–4.5 mm; matrix 256×256; FOV 200–230 mm; TR ≤ 20,000 ms; TE 100–140 ms). STS depth measurement followed Kasprian et al.: baseline connecting vertices of superior and middle temporal gyri; depth measured from sulcal pit to baseline. Super-resolution fetal MRI aided pit definition. Two blinded raters; excellent inter-rater reliability (ICC=0.998; 95% CI 0.993–0.999). Laterality Index (LI) = (left−right)/(left+right)×100; categorized as left (≥20), bilateral (−20 to +20), right (≤−20). Alternative asymmetry measure per Bullmore et al. computed (see Supplementary). fMRI acquisition: Language fMRI at 3T (Siemens TIM Trio) or 3T Philips Elition after scanner failure; scanner type modeled as nuisance. T1 MPRAGE structural; EPI parameters provided (Siemens: TR 2000 ms, TE 42 ms, voxel 2.1×2.1×4 mm, 200 vols; Philips: TR 1000 ms, TE 25 ms, slice 2.5 mm, 51 slices, 400 vols). Task: auditory description definition vs reversed speech control; button press for correct definitions or tones. Three age-adjusted versions; total 6:40 min. Preprocessing in SPM12: realignment, motion parameters, age/sex-matched pediatric template (TOM toolbox), normalization, smoothing (FWHM 4 mm). First-level GLM with 6 motion regressors; contrast task>control. Lateralization indices (LI global; LI frontal/temporal/parietal) computed via LI-toolbox with bootstrapping. Group maps: one-sample t-test (cluster extent 20). Second-level multiple regression examined association of fetal STS LI with activation (adjusted for age at test and MR device; FWE p<0.05). DTI acquisition/processing: DWI (b=0,1000 s/mm²; 30 directions; 2×2×2 mm; TR 8000 ms; TE 83 ms). Structural T1 processed in FreeSurfer for segmentation. DWI preprocessing: denoising (overcomplete local PCA), skull-strip, bias-field correction. Constrained spherical deconvolution (MRtrix3) to estimate fODFs; distortion correction via rigid and nonrigid registration aligning pseudo-T1 (DWI space) to undistorted T1 (DRAMMS). Whole-brain anatomically constrained tractography (ACT) with 10 million streamlines; SIFT2 filtering. Automated dissection via White Matter Query Language (WMQL) of AF, SLF I/II/III, UF, ILF. For each tract: number of streamlines, volume, and mean FA. Tract LI = (left−right)/(left+right)×100; categorized as left (≥20), bilateral (±20), right (≤−20). Cognitive assessment: Standardized German tests with age-adjusted z-scores: Token Test for Children (language comprehension), Wortschatz- und Wortfindungstest (expressive vocabulary), Regensburger Wortflüssigkeitstest (semantic and phonemic verbal fluency; averaged), German Auditory Verbal Learning Test (VLMT; short-term recall, long-term recall, recognition averaged as verbal memory). Nonverbal IQ proxy: WISC-IV Perceptual Reasoning Index (Block Design, Matrix Reasoning, Picture Completion). Handedness: Edinburgh Handedness Inventory (−1 left to +1 right, inverted to match laterality sign). Socioeconomic status (SES): composite of parental education and household income (scaled 1–7). Statistics: Normality via Kolmogorov–Smirnov (cognitive data, fetal STS LI, DTI normal; fMRI LIs non-normal). Group comparisons via t-tests/paired t-tests; Pearson correlations for normal variables; Spearman for fMRI LIs. Multiple linear regressions tested fetal STS LI predicting language z-scores (adjusted for perceptual reasoning) and DTI measures (adjusted for age at test and MR device). Bonferroni corrections applied. Assessed multicollinearity (Tolerance/VIF) and Durbin–Watson for independence of errors. Sample sizes reported per analysis (e.g., STS LI n=29; fMRI LI n=23; DTI n=25).

Sample: 38 children (6–13 years; mean ≈9), all with normal fetal and childhood MRIs; STS visible and quantifiable in 29 fetuses (≥25 weeks gestation). Language abilities were broadly typical; no child met criteria for language impairment. Fetal STS asymmetry and development:

• Right STS was significantly deeper than left in fetuses with visible STS (paired t=6.494, p<0.001). Group mean LI indicated right lateralization. Individual LI ranged from −80.18 to 33.80; 20 right-lateralized, 8 bilateral, 1 left-lateralized.
• STS depth (both hemispheres) increased with gestational age (right r=0.645, p<0.001; left r=0.677, p<0.001), but fetal STS LI did not correlate with gestational age (r=0.12, p=0.551).
• Fetal STS measures did not correlate with handedness, SES, or nonverbal perceptual reasoning; no sex differences in fetal STS depths or LI. Language outcomes and lateralization:
• Language test scores did not correlate with age at test, handedness, SES; expressive vocabulary and language comprehension correlated with nonverbal perceptual reasoning (r=0.52, p=0.004; r=0.56, p=0.001).
• fMRI (usable n=23) showed typical group-level left-lateralized language activation (middle temporal including insula, inferior frontal operculum, superior frontal); right hemisphere activation in insula and medial frontal. Single-subject LI total: 17 left, 1 bilateral, 5 right. Predictive value of fetal STS LI for later language abilities (n=29; adjusted for perceptual reasoning):
• Expressive vocabulary: F(2,26)=10.55, p<0.001, R²=0.45; less right lateralization (i.e., higher LI) associated with better vocabulary.
• Verbal fluency: F(2,26)=8.85, p=0.001, R²=0.43; less right lateralization associated with better fluency.
• Verbal memory: F(2,26)=11.88, p<0.001, R²=0.48; less right lateralization associated with better memory.
• No association with language comprehension. No multicollinearity (Tolerance/VIF=1.00); Durbin–Watson values ~1.5–1.9 indicated independent errors. Association with later language localization (fMRI):
• Spearman correlations of fetal STS LI with LI measures (global/regional) were nonsignificant. However, second-level multiple regression (SPM; adjusted for age and MR device; FWE p<0.05) revealed a significant positive association between fetal STS LI and activation in the left superior temporal lobe (peak MNI −54, −8, 0; cluster size 84; T=10.97): less right fetal STS lateralization predicted more left temporal language activation 6–13 years later. Association with language-related white matter tracts (DTI; n=25):
• Fetal STS LI did not predict LIs of AF, ILF, SLF I/II/III, or UF (laterality indices). Fractional anisotropy showed little hemispheric difference and was not analyzed for laterality.
• A positive association was found with the absolute volume of the left SLF III: regression model F(3,21)=4.95, p=0.009, R²=0.41; more leftward fetal STS LI associated with greater left SLF III volume. No other tract volume or streamline count correlated with fetal STS LI. Overall, fetal STS depth asymmetry explained over 40% of the variance in multiple verbal skills and related to left superior temporal functional localization, supporting its prognostic relevance for language development.

Findings demonstrate that individual variability in fetal STS depth asymmetry, a robust prenatal structural feature, has meaningful prognostic value for later language abilities and neural organization. While most fetuses exhibited the normative right-deeper STS, those with less pronounced rightward asymmetry (i.e., relatively deeper or earlier-developing left STS) developed superior expressive vocabulary, verbal fluency, and verbal memory, and showed increased left temporal fMRI activation in childhood. The results align with theories that early cortical folding and sulcal depth reflect maturation of adjacent functional regions and their connectivity. Given that language-sensitive cortex concentrates in the depths of the left STS, earlier or more localized maturation of left temporal cortex may support more efficient early language learning. The association with left SLF III volume further suggests a structural substrate connecting frontal-parietal-temporal regions linked to language functions. Absence of associations with language comprehension may reflect its distributed neural basis extending beyond left temporal cortex. The specificity of effects to language (and not nonverbal reasoning) underscores the functional relevance of fetal temporal morphology to language development.

This longitudinal study provides evidence that fetal STS depth asymmetry is a prognostic biomarker for later language development in typically developing children. Less rightward fetal STS asymmetry predicted better expressive vocabulary, verbal fluency, and verbal memory, and more left-lateralized activation in the left superior temporal lobe years later. These findings suggest that earlier or more localized development of left temporal cortex is favorable for language acquisition. If replicated in larger cohorts and clinical populations (e.g., dyslexia, autism, specific language impairment), fetal STS asymmetry could inform early diagnostics and counseling regarding the maturity and integrity of language-related neural systems. Future work should employ larger, gestational-age–stratified samples, harmonized imaging protocols, and examine non-linear growth effects, sex-specific trajectories, and clinical relevance across neurodevelopmental disorders.

• Small final analytic sample (n=29 for fetal STS quantification; fMRI n=23; DTI n=25) limits power and detection of complex or non-linear effects.
• Wide age range at follow-up (6–13 years) and variable gestational ages at fetal MRI may introduce heterogeneity; STS not visible in 9 participants (<25 weeks), reducing sample size.
• Scanner change during study (Siemens vs Philips) could introduce variance despite statistical adjustment.
• Sex imbalance reflecting clinical referral patterns; study underpowered to assess sex-specific effects.
• Laterality indices of white matter measures may be insensitive to language abilities and to fetal STS asymmetry.
• Cohort comprised only typically developing children with normal MRIs; generalizability to clinical populations remains to be established.