The investigators analyzed resected middle temporal gyrus specimens from 117 individuals who underwent surgery for temporal lobe epilepsy (TLE). The cohort included 74 cases with hippocampal sclerosis (HS) and 12 cases with temporal lobe sclerosis (TLS, defined as FCD IIIa). Preoperative cortical thickness of the middle temporal gyrus was quantified using CAT12 relative to healthy controls to classify cases by presence or absence of MRI-defined cortical thinning. The source did not report details of MRI threshold values used for categorical classification beyond the CAT12 pipeline description.
On NeuN-immunolabelled sections, the study team applied automated image analysis to measure neuronal spatial distributions, cell-size distributions, and neuronal packing density across cortical layers. They also quantified relative cortical layer thickness and sulcal–gyral cytoarchitectural variation. Data indicated that cases with MRI cortical thinning did not show diffuse neuronal loss. Instead, the thinning phenotype correlated with increased neuronal density and altered cell-size spatial organisation in both superficial and deep layers. The investigators reported this pattern most prominently in right-sided TLE with HS. The source did not provide longitudinal imaging data or progressive atrophy time courses; therefore, the relationship to temporal progression could not be assessed from the preprint.
Layer-specific analyses produced a nuanced picture. The study team observed limited evidence of laminar thinning by histology, confined to layer VI in the right TLE/HS subgroup. Superficial layers did not show consistent laminar loss that would explain MRI thinning. Instead, altered neuronal packing and inter-layer cell-size redistribution emerged as the dominant histological correlates of the MRI signal-based thinning metric. Clinical–pathological correlations were lateralised; the investigators noted differences linked to hemisphere and to age of onset, implying that processes driving cortical thinning may differ between left and right TLE.
Temporal lobe sclerosis (TLS/FCD IIIa) cases did not show cortical thinning on MRI nor laminar thinning on histology when compared to non-TLS cases. Yet TLS tissue demonstrated altered neuronal densities and changes in neuronal composition that extended beyond superficial cortex. Conversely, some cases with hippocampal sclerosis but without MRI cortical thinning exhibited loss of the expected inter-hemispheric cytoarchitectural asymmetry. The authors reported this as a distinct pathological pattern separate from the thinning phenotype, suggesting at least two different substrate classes detectable by combined MRI and histology.
MRI cortical thinning aligned with attenuation of normal sulcal–gyral cytoarchitectural variation. The study team observed less pronounced microarchitectural gradients across sulcus and gyrus in thinned cortex, a pattern consistent with disruption of typical cortical architecture rather than simple tissue loss. The investigators interpreted these findings as evidence for altered neuropil volumes—potentially reduced myelin or dendritic arborisation—rather than a primary loss of neuronal cell bodies. The article did not present direct myelin or dendritic quantitation, so those interpretations are inferential from neuronal packing and cell-size distributions reported.
Authors noted that the prevailing interpretation of MRI cortical thinning as cumulative neuronal degeneration is not supported by their morphometric data. Instead, cortical thinning in this surgical TLE cohort associated with microstructural reorganisation: higher neuronal density, changed cell-size distributions, and disrupted cytoarchitectural gradients. These features point toward alterations in neuropil volume—including myelin content and dendritic remodelling—and to network- or seizure-related dynamic processes that could be reversible. The preprint did not provide direct measures of neuropil components (for example, quantitative myelin staining or dendritic spine counts), so those hypothesized substrates remain to be directly tested.
This work uses a surgical cohort, which may not represent the full spectrum of TLE; the source did not report population prevalence data or prospective longitudinal imaging correlations. The preprint did not supply specific MRI thresholding parameters used for classifying cortical thinning beyond CAT12 processing, nor did it include quantitative measures of myelin or dendritic structure to confirm the inferred neuropil changes. The authors raised hemisphere-specific mechanisms and clinical-pathological correlations (including age of onset) but did not resolve causal links between seizures, network alterations, and the observed microstructural features. These unresolved questions define next steps: targeted myelin and dendritic quantitation, longitudinal imaging–histology correlation, and replication in non-surgical cohorts to test generalisability.