Relationship between visuoperceptual functions and parietal structural abnormalities in temporal lobe epilepsy

Progressive gray matter volume reductions beyond the epileptogenic area has been described in temporal lobe epilepsy. There is less evidence regarding correlations between gray and white matter volume changepres and multi-domain cognitive performance in this setting. We aimed to investigate correlations between volume changes in parietal structures and visuospatial performance in temporal lobe epilepsy patients. we performed a cross-sectional study comparing global and regional brain volume data from 34 temporal lobe epilepsy patients and 30 healthy controls. 3D T1-weighted sequences were obtained on a 3.0 T magnet, and data were analyzed using age and sex-adjusted linear regression models. Global and regional brain volumes and cortical thickness in patients were correlated with standardized visual memory, visuoperceptual, visuospatial, and visuoconstructive parameters obtained in a per-protocol neuropsychological assessment. temporal lobe epilepsy patients had smaller volume fractions of the deep gray matter structures, putamen and nucleus accumbens, and larger cerebrospinal fluid volume fraction than controls. Correlations were found between: 1) visual memory and precuneus and inferior parietal cortical thickness; 2) visuoperceptual performance and precuneus and supramarginal white matter volumes; 3) visuospatial skills and precuneus, postcentral, and inferior and superior parietal white matter volumes; 4) visuoconstructive performance and inferior parietal white matter volume. Brain volume loss is widespread in temporal lobe epilepsy. Volumetric reductions in parietal lobe structures were associated with visuoperceptual cognitive performance.


Introduction
Temporal lobe epilepsy (TLE), one the most common forms of focal epilepsy in adults (Téllez-Zenteno and Hernández-Ronquillo 2012), is associated with volume loss in the temporal regions, particularly the hippocampus (Bonilha et al. 2003;Blümcke et al. 2013). Recent studies highlight the progressive volume loss associated with hippocampal sclerosis, which seems to correlate with epilepsy duration (Lopez et al. 2022).
Quantitative magnetic resonance imaging (MRI) provide objective, reproducible biomarkers of brain atrophy (Caciagli et al. 2017;Fischl and Dale 2000). Automated MRI segmentation tools yield quantitative measures of global and regional brain volumes and cortical thickness (Caciagli et al. 2017;Jber et al. 2021). Analyses with these tools have shown that brain volume loss in TLE affects also extratemporal regions beyond the seizure-onset focus (Bonilha and Keller 2015;Whelan et al. 2018), which seems to be a progressive phenomenon (Galovic et al. 2019). It is uncertain whether the volume loss is directly associated with the severity of epilepsy (Coan et al. 2009;Bernhardt et al. 2010). In addition, widespread abnormalities in white matter (WM) integrity seem to correlate with disease duration (Liu et al. 2015;Tsuda et al. 2018;Chang et al. 2019).
Cognitive decline has also been well described in longterm TLE. Memory is classically affected (Helmstaedter and Kockelmann 2006), although impairment of other domains such as attention, language, and executive functions, has been reported (Hermann et al. 2006;Allone et al. 2017). Visuospatial and visuoperceptual functions have received less attention in TLE. Some authors have reported impaired perceptual ability in these patients (Grant et al. 2005), whereas others describe normal visuospatial performance (Grant et al. 2008). Recent evidence has shown that visuospatial memory impairment in TLE appears in parallel to preserved non-memory visuospatial functions (Tallarita et al. 2019). However, there is an important variability in the tests used, especially when detecting right hemispheric dysfunction, which may have contributed to the disparity of results (Barr et al. 1997;Bentvelzen et al. 2021). Whether cognitive performance in these particular domains reflects extratemporal structural abnormalities remains to be determined.
The cognitive abnormalities and neuroimaging findings in TLE suggest that the structural brain damage extends beyond the seizure onset zone and causes diffuse brain dysfunction. It is important to elucidate the structural abnormalities leading to cognitive decline in TLE patients to better understand the mechanisms of epileptogenesis and underlying brain damage. With this study, we aimed to evaluate changes in regional brain volumes, focusing on the parietal structures, and explore their correlation with visuoperceptual impairment in adult patients with TLE.

Study design and participants
This is a cross-sectional study conducted in a dedicated epilepsy unit, comparing neuroimaging from adult TLE patients assessed in the outpatient clinic between 2016 and 2018 and age-/sex-matched healthy controls. The protocol was approved by the local Ethics Committee [PR(AG)391/2017], and all patients provided signed informed consent. TLE patients with a 3.0T-MRI study, who had undergone perprotocol neuropsychological assessment within ≤12 months after neuroimaging, were selected from our outpatient database. The TLE diagnosis and lateralization were based on a combination of seizure semiology suggesting a temporal onset, consistent MRI findings, and temporal lobe interictal/ ictal epileptiform activity on video-electroencephalographic recordings. Seizure and epilepsy diagnoses were classified according to the latest International League Against Epilepsy (ILAE) classification (Scheffer et al. 2017). Demographic and clinical data were collected and drug resistance criteria was established according to the 2010 ILAE task force (Kwan et al. 2010). Patients with an unconfirmed diagnosis, unknown seizure onset, structural abnormalities with visible extension to the parietal lobe, or suboptimal MRI scans were excluded from the study, as were patients unable to complete the neuropsychological assessment and those with severe sensory impairment or an intelligence quotient ≤70.

MRI analysis
MRI data were acquired using a 12-channel phased-array head coil on a 3.0 T magnet (MAGNETOM Trio, Siemens AG). The MRI protocol included the following sequences: transverse 2D T2-fluid-attenuated inversion recovery (FLAIR) (repetition time=9000ms, echo time=87ms, inversion time=2500ms, voxel size=0.49×0.49×3.0mm) and sagittal 3D T1-weighted magnetization prepared-rapid gradient echo (MP-RAGE) (TR=2300ms, TE=3000ms, voxel size=1.0×1.0×1.2mm). The MRI data were re-assessed by an experienced neuroradiologist (SS) for the study. All analyses were performed using FreeSurfer (version 6.1) on the 3D T1-weighted MP-RAGE images to calculate global and regional brain volume fractions and cortical thickness. MRI data were analyzed following established pipelines (Free-SurferAnalysisPipelineOverview -Free Surfer Wiki. (n.d.). Image voxels were classified as WM or non-WM based on image intensity and neighbor constraints. An initial surface was generated for each hemisphere by tiling the outside of the WM mass. This initial WM surface was then refined to follow the intensity gradients between the WM and gray matter (GM). The WM surface was then nudged to follow the intensity gradients between the GM and cerebrospinal fluid (CSF) to obtain the pial surface. The distance between the WM and pial surfaces gives the thickness of the cortex at each location. Subcortical GM (sGM) volumes and the global GM, WM, and CSF volumes were also calculated as part of the established pipeline. Lobar WM volumes were calculated as the sum of region-specific WM volumes for each lobe. In each patient, estimated volume fractions were multiplied by 100 and divided by the corresponding estimated total intracranial volume to obtain the GM, sGM, WM, and CSF volume fractions. Imaging data from TLE patients were compared with data from healthy controls in our laboratory database.

Cognitive assessment
All patients were evaluated by a trained neuropsychologist and completed a comprehensive neuropsychological assessment that included tests of attention, executive function, processing speed, verbal memory, and visuospatial abilities. However, for the purpose of the study, visuospatial tests were selected and described in detail: visual memory was assessed by immediate recall (ROCF-IR) and 30-minute delayed recall (ROCF-DR) in the Rey-Osterrieth Complex Figure test. Visuoperceptual, visuospatial, and visuoconstructive skills were evaluated using the Hooper Visual Organization Test (VOT), the Judgment of Line Orientation test (JLO), the Block Design subtest from WAIS-III (WAIS III-BD), and the ROCF immediate copy (ROCF-IC).. Visuoperceptual, visuospatial, and visuoconstructive skills were analyzed individually as separate specific functions and described in detail for the purpose of the study. All test scores were converted into T-scores based on a Spanish population of healthy controls by normalizing scores with a mean value of 50 and a standard deviation of 10. Lower scores indicate poorer cognitive performance. T-scores of 40 or higher were considered to indicate normal performance in the various tests.

Statistical analysis
Descriptive and frequency statistics were obtained and compared using the SPSS Statistics 22.0 software package. As quantitative data were normally distributed (checked with QQ plot), parametric tests were performed for all comparisons involving continuous variables. The pairedsample t test was used to assess differences in volume fractions between the left and the right hemisphere. The asymmetry index (AI) was calculated as a percentage ratio; that is, the difference in volume values between the left and right hemisphere over the sum of these (AI= [leftright]*100/[left+right]). A positive AI indicates larger volume in the left hemisphere, whereas a negative index indicates larger volume in the right hemisphere. Patient results were compared with those of healthy controls using Pearson's chi-square test for categorical variables and the t test for continuous variables. Correlations between continuous variables were assessed with Pearson's correlation coefficient. Scatterplots with the most representative correlations between cognitive performance and volume fractions are shown in Supplementary Material. Age-and sexadjusted linear regression models were used to compare patient characteristics and cognitive performance with the estimated brain volume fractions. B coefficients were used to calculate relative volume differences with respect to the reference group.
Bonferroni was applied to manage multiple comparison correction using the Dubey/Armitage-Parmar (D/AP) method (AJ et al. 1997). Thus, to achieve statistical significance at a level of 0.05, it was necessary to obtain a p-value of <0.0091 when comparing global brain volumes between TLE patients and healthy controls, and <0.0027 when comparing brain volume/cortical thickness with cognitive assessment. Significant p-values after correction for multiple comparisons are marked with an asterisk throughout the results. Given the exploratory nature of the study, we also reported results with raw p-values of <0.05 as findings of potential interest.

Results
Thirty-eight patients were initially selected from the outpatient database. After reassessment, four patients were excluded because of an imprecise TLE diagnosis or incomplete neuropsychological assessment due to sensory impairment. Thirty-four TLE patients (mean ± SD age 39.7 ± 14.4 years; range 19-68 years; 55.9% women) were ultimately recruited. The patients' imaging data were compared with data from 30 healthy controls comparable for age (mean age 40.5 ± 16.8 years; range 18-69) and sex (51.6% women). The patients' demographic and clinical data are summarized in Table 1.
In specific visual processing and perceptual skills, TLE patients had low T-scores in both the ROCF-IR and ROCF-DR (Fig. 1).

TLE patients compared with healthy controls
TLE patients had a smaller GM (-3.6%, p=0.015) and a larger CSF (+10.7%, p=0.006*) volume fraction compared to controls. Regarding the regional volumes, smaller fractions were found in the frontal lobe WM (-4.8%, p=0.027) and the sGM (-6.4%, p=0.003*), where several subcortical nuclei were affected (Fig. 2). There were no significant differences in cortical thickness or AI between TLE patients and controls.
As to the specific parietal volumes, the posterior cingulate WM tended to be smaller in TLE patients (-4.6%; p=0.067), although not statistically significant. No other differences were found in the specific parietal WM volumes or cortical thickness between TLE patients and controls.

Correlation between parietal cortical thickness and visuoperceptual functioning (Fig. 3)
No significant correlations were found between cortical thickness and visuoperceptual, performance in the overall group of TLE patients. On separate analysis of these correlations according to the hemispheric seizure onset zone, an inverse correlation was found in right-TLE patients between visual memory (ROCF-IR) and right superior parietal cortical thickness (r= -0.463, p=0.044), and between visuospatial performance and right posterior cingulate (r= -0.522, p=0.037) and post-central gyrus (r= -0.566, p=0.036) cortical thickness. In addition, a direct correlation was found between visuoconstructive functions (ROCF-IC) and left precuneus cortical thickness (r= 0.625, p=0.045).
In left-TLE patients, an inverse correlation was found between visuoconstructive functioning and left postcentral WM volume (r= -0.216; p=0.047). No significant correlations were found between visuoperceptive functioning and WM volume status in the right hemisphere in these patients.

Correlation with epilepsy features
Longer epilepsy duration correlated with smaller right supramarginal WM volume (r= -0.330; p=0.031). Patients with focal seizures evolving to bilateral tonic-clonic status showed a tendency to have larger left precuneus (0.61±0.07 vs 0.55±0.10; p=0.054) and right posterior cingulum (0.29±0.03 vs 0.27±0.03; p=0.079) WM volumes. No significant correlations were found for either brain volume or cognitive performance with respect to drug resistance, seizure type, or epilepsy duration.

Discussion
In this study, we systematically investigated regional brain volume abnormalities occurring in TLE patients, and the relationship between parietal lobe volumes and visuoperceptual cognitive performance. Our results show significant volume decreases in several brain structures involving both GM and WM regions in this population, as well as close correlations between parietal lobe volume status and visuoperceptual functioning, which have not been strictly established in TLE.
These data provide further evidence that TLE is associated with brain damage beyond the temporal lobe (Jber et al. 2021;Keller and Roberts 2008). The structural changes may indicate a neural network dysfunction that is reflected in cognitive functioning classically understood to be extratemporal (Neal et al. 2019).
The percentage of drug-resistant patients enrolled was slightly higher than that reported in the literature (Currie et al. 1971;Kalilani et al. 2018; Téllez-Zenteno and shown at the top of the graph. Regional volume fractions (GM and specific subcortical nuclei, followed by lobar and specific WM) are shown below. WM-specific volumes are represented as frontal, parietal, occipital, temporal, and cingulum volumes. The volume of each structure is represented as their total volume (considering the sum of both hemispheres in the case of bilateral structures). Relative differences are shown as 95% confidence intervals with respect to the average volume of each structure in the control group. TLE patients had smaller GM, subcortical GM, putamen, hippocampal, accumbens, and frontal WM volumes, as well as a larger total CSF volume. *raw p-value <0.05. CC, corpus callosum; GM, gray matter; WM, white matter Hernández-Ronquillo 2012). All participants were assessed using a high-quality, standardized MRI protocol that is commonly performed in epilepsy patients, facilitating reproducibility and interpretation of the results.
TLE patients are expected to show lower scores in several cognitive domains (Allone et al. 2017;Helmstaedter and Kockelmann 2006;Sheldon et al. 2020). Mean T-scores in visuoperceptual, visuospatial, and visuoconstructive skills were in line with previous evidence indicating normal performance in these cognitive domains (Grant et al. 2008). It is likely that these findings are influenced by the variability and type of tests used, which can make it difficult to generalize and establish a cognitive phenotype in the overall TLE population (Bentvelzen et al. 2021;McDonald et al. 2022). However, T-scores in the patients included showed a wide distribution, which allowed us to establish more reliable correlations between these cognitive domains and parietal volumes, and interestingly, although performance in most visuoperceptual functions was considered normal in our sample, several associations were still found between cognitive performance and imaging markers.
Our results are consistent with previously reported volume differences in GM and WM structures and CSF in TLE patients. Although a progressive reduction in cortical thickness has been described in TLE (Galovic et al. 2019)(B. C. Bernhardt et al. 2009), we found no large differences in this parameter between patients and the control group. Patterns of cortical thickness loss seem to vary depending on the type of epilepsy and its lateralization (Galovic et al. 2019); hence, it could be more difficult to detect cortical thinning in smaller and more heterogeneous groups of patients. As to the sGM structures, our results are consistent with previous evidence showing hippocampal atrophy in TLE patients (Riederer et al. 2020). The findings also provide further evidence of demonstrable atrophy in extratemporal regions, such as the putamen (Kim et al. 2016). There is currently very little evidence of accumbens volume loss in TLE patients (Wang et al. 2017) as was seen here, although there is some data from animal models (Fu et al. 2018).
Several correlations were found between the regional WM volume loss and poorer visuoperceptual performance. In contrast, cortical thickness showed weaker and even inverse correlations with cognitive function. Evidence in Alzheimer's disease has suggested that cortical atrophy in itself may be a poorer predictor of global cognitive decline than other biomarkers (Hedderich et al. 2020). Disrupted WM integrity has also been widely described in epilepsy patients (Tsuda et al. 2018), and our results suggest that the loss of white matter volume could also be a structural manifestation of this neural network disturbance underlying the global cognitive dysfunction observed in TLE (Chang et al. 2019).
Also as expected, a larger percentage of correlations between parietal WM volume loss and poorer visuoperceptual functioning were found in patients with right-onset TLE. Thus, WM status in this region seems to be more highly implicated in these cognitive functions (De Schotten et al. 2011). Previous functional neuroimaging studies have shown that the superior and inferior parietal cortex are involved in visuospatial skills (Seydell-Greenwald et al. 2017), but specific correlations have not been assessed to date in TLE patients.
In line with previous studies, the right posterior cingulate correlated with visuospatial and visuoconstructive skills (Leech and Sharp 2014). We also found that the precuneal region correlated with visuoperceptual, visuospatial, and visuoconstructive skills. Interestingly, structural and metabolic abnormalities in this region have been associated with poorer performance in constructional praxis skills in patients with early Alzheimer's disease (Hedderich et al. 2020). Our results provide support that the cingulum and precuneus are both implicated in the visuoperceptual network of these patients. Furthermore, these correlations could represent an interindividual diversity in the established damage in parietal regions, which could underlie the contrasting findings for visuoperceptual impairment in previous analyses (Grant et al. 2005(Grant et al. , 2008.
Our study has limitations. The relatively small sample size compared to recent studies investigating cognitive performance and imaging markers, and the cross-sectional design may have prevented us from elucidating certain findings in further detail. In addition, our sample included TLE of several etiologies and a high percentage of patients were receiving two or more ASDs, which might have influenced the structural and cognitive correlations described. Further studies including larger samples of non-lesional and strictly unilateral epilepsies would be useful to confirm these results and eventually allow establishing common patterns of cortical thinning and white matter atrophy and/or dysfunction in patients with TLE, which could be related to the spread of epileptiform activity and patterns of cognitive impairment

Conclusions
Widespread GM loss affecting several brain regions is common in TLE and is reflected by an increase in CSF volume. In this setting, WM volume differences in parietal lobe structures could be an indicator of visuoperceptual cognitive performance. These results suggest that TLE is not focal, but instead a more complex neurological disorder causing structural changes related to the epileptogenic network and cognitive dysfunction beyond verbal memory and language.
revising it critically for important intellectual content (EF, SS, DP, MT, AR, MT) and approval of final version to be published and agreement to be accountable for the integrity and accuracy of all aspects of the work (all authors).
Funding This study was supported by a predoctoral grant in Neuroscience by Fundación Tatiana Pérez de Guzmán el Bueno. This funding source was not involved in the study design, collection, analysis or interpretation of the data, writing of the report, or submission for publication.
Data availability Anonymized data that support the findings of this study are available from the authors on reasonable request.
Code availability Not applicable.

Declarations
Conflicts of interest E. Fonseca declares research funding and honoraria from UCB Pharma, Esteve laboratorios, Eisai Inc, GW Pharmaceuticals, Angelini Pharma and Sanofi Genzyme. E. Santamarina declares research funding and speaking fees from UCB Pharma, BIAL Pharmaceutical, EISAI Inc. and Esteve laboratorios. L. Abraira declares research funding and speaking fees from UCB Pharma, BIAL Pharmaceutical, EISAI Inc., Sanofi Genzyme and Esteve laboratorios. I. Seijo declares research funding from UCB Pharma, Neuraxpharm and GW pharmaceuticals. M. Toledo declares research funding and speaking fees from UCB Pharma, BIAL Pharmaceutical, EISAI Inc., GW Pharmaceuticals, Arvelle Therapeutics, Angelini Pharma and Esteve laboratorios. S. Sarria, D. Pareto, M. Turon, M. Quintana, C. Tortajada and A. Rovira have no conflicts of interest to declare.

Ethics approval
We confirm that we have read the journal's position on issues concerning ethical publication and that this report is consistent with those guidelines. The study protocol was approved by the Vall d'Hebron Research Institute Ethics Committee [PR(AG)391/2017], following the Ethics of the World Medical Association for experiments involving humans. The study conforms with the STROBE guidelines for observational studies.
Consent to participate All patients provided signed written informed consent prior to inclusion in the study.
Consent for publication Not applicable.