Abstracts

Rationale: Temporal lobe epilepsy (TLE) is associated with variable degrees of mesiotemporal damage and memory impairment. To explore associations between atypical brain structure and function, the current work adopted a novel framework to map cortex-wide gradients of microstructural differentiation and examined how microstructural changes relate to episodic memory function in TLE.

Methods: A cohort of 21 drug-resistant TLE patients (9 women; mean±SD: 36.14±11.59 years) and 35 healthy controls (HCs; 17 women; 34.51±9.27 years) underwent high-resolution T1-weighted and microstructural imaging at 3T. In each subject, we sampled quantitative T1 (qT1) intensities along 14 equivolumetric surfaces between pial and white matter boundaries, yielding cortex-wide microstructural intensity profiles (Figure 1A). Cross-correlating profiles provided microstructural similarity matrices between all vertex pairs. Corresponding affinity matrices were used to derive the principal gradient of microstructural similarity with diffusion map embedding (Figure 1B). Participants also completed a functional MRI episodic memory task (Figure 2A), in which they first memorized 56 unique image pairs (encoding phase). Participants later matched paired images in a 3-alternative forced choice design (retrieval phase). After excluding participants with incomplete data (n=4) and chance accuracy (n=4), we studied how regions showing atypical microstructural differentiation were embedded within functional memory networks and related to recall accuracy. Group comparisons of surface-based features were controlled for effects of sex and age and were corrected for multiple comparisons using random field theory.

Results: Comparing gradients of cortical microstructure between groups, differentiation between sensory/motor and paralimbic anchors was reduced in TLE. Gradient reductions primarily targeted anterior temporal regions and lateral prefrontal cortices ipsilateral to the seizure focus (Fig1C; p_FWE < 0.001). Seed-based functional connectivity analysis during encoding centered on peak regions in clusters of gradient reductions showed significantly reduced connectivity between temporal and prefrontal regions (t=-3.057, p=0.004; Figure 2B). These findings were consistent with cortex-wide changes, which also revealed more extended connectivity reductions in these regions in TLE (p_FWE < 0.05). Moreover, individual recall performance was correlated with the magnitude of microstructural gradient changes in temporal (r=0.406; p=0.004), but not prefrontal regions (r=0.111; p=0.457; Figure 2C).