A Kaleidoscope of Cognitive Vulnerability in Temporal Lobe Epilepsy

Abstract

Cognitive Phenotypes in Patients With Drug-Resistant Temporal Lobe Epilepsy: Relationships With Cortisol and Affectivity

Cano-López I, Catalán-Aguilar J, Lozano-García A, Hidalgo V, Hampel KG, Tormos-Pons P, Salvador A, Villanueva V, González-Bono E. Clin Neuropsychol. 2025;39(2):400-423. Epub 2024 Jul 4. PMID: 38965831.

Objective: Drug-resistant temporal lobe epilepsy (TLE) is a neurological disorder characterized by cognitive deficits. This study examined whether patients with TLE and different cognitive phenotypes differ in cortisol levels and affectivity while controlling for demographic and clinical variables. Methods: In this cross-sectional study, 79 adults with TLE underwent neuropsychological evaluation in which memory, language, attention/processing speed, executive function, and affectivity were assessed. Six saliva samples were collected in the afternoon to examine the ability of the hypothalamic-pituitary-adrenal (HPA) axis to descend according to the circadian rhythm (C1 to C6). The cortisol area under the curve concerning ground (AUCg) was computed to examine global cortisol secretion. Results: Three cognitive phenotypes were identified: memory impairment, generalized impairment, and no impairment. The memory-impairment phenotype showed higher cortisol levels at C4, C5, and C6 than the other groups (p = 0.03, η² = 0.06), higher cortisol AUCg than the generalized-impairment phenotype (p = 0.004, η² = 0.14), and a significant reduction in positive affectivity after the evaluation (p = 0.026, η² = 0.11). Higher cortisol AUCg and reductions in positive affectivity were significant predictors of the memory-impairment phenotype (p < 0.001; Cox and Snell R² = 0.47). Conclusions: Patients with memory impairment had a slower decline in cortisol levels in the afternoon, which could be interpreted as an inability of the HPA axis to inhibit itself. Thus, chronic stress may influence hippocampus-dependent cognitive function more than other cognitive functions in patients with TLE.

Commentary

When Sir William Gowers lamented the “imperfections of intellectual power” in epilepsy over a century ago, he presciently alluded to what modern science now confirms: epilepsy extends far beyond seizures alone. Early 20th-century neurology focused narrowly on seizure control and survival, but today's sharper lens captures a more nuanced landscape. Cognitive impairment is now recognized as one of the most prevalent and consequential comorbidities of epilepsy, affecting up to 70% of people with epilepsy (PWE). ¹ The relationship between epilepsy and memory is particularly precarious, as the condition directly targets memory circuits and imposes cumulative neurobiological insults over time. Longitudinal research suggests that PWE may follow an accelerated trajectory of cognitive decline, in some cases culminating in mild cognitive impairment or dementia.^2–4 So, how do we address these 70 shades of cognitive vulnerability in PWE?

In this context, the recent study by Cano-Lopez et al ⁵ offers an illuminating contribution to the field, applying data-driven clustering to parse the heterogeneity of cognitive phenotypes in drug-resistant temporal lobe epilepsy (TLE). The investigators evaluated 79 individuals undergoing presurgical workup, integrating salivary cortisol profiles and affectivity using standardized positive and negative affect scales alongside neuropsychological testing. By leveraging both biological and psychological data, the study aimed not only to identify cognitive subtypes but also to validate them mechanistically, a forward-thinking step in epilepsy phenotyping.

Cognitive heterogeneity in TLE is by no means a new concept, but it is one that continues to evolve. Landmark studies by Hermann et al ⁶ and others have identified 3 main cognitive clusters, challenging the once-dominant narrative that memory loss is the universal cognitive signature of TLE. Building on this work, researchers such as McDonald, Hermann, and colleagues have proposed a “network-based precision taxonomy” that includes cognitive and psychiatric comorbidities as integral to epilepsy classification. ⁶ In alignment with this framework, Cano-López et al ⁵ used hierarchical clustering across 4 domains, memory, language, attention/processing speed, and executive function, to identify 3 cognitive phenotypes in a Spanish-speaking cohort: globally intact cognition (42%), memory-predominant impairment (25%), and global impairment (33%). These findings reaffirm the reproducibility of cognitive phenotypes across cohorts and languages, offering future opportunities to integrate phenotypic classification into global collaborations and registries.

Where this study truly distinguishes itself is in its integration of cortisol levels and affective responses to biologically validate the cognitive phenotypes. ⁵ Cano-López et al ⁵ identified lower education level, elevated afternoon cortisol, and reduced positive affect as key predictors of the memory-impaired phenotype. These findings echo prior studies linking stress and anxiety to worse verbal memory performance, independent of seizure frequency. ⁷ The findings underscore the relevance of psychoneuroendocrine mechanisms, particularly dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis, in shaping cognitive outcomes in epilepsy. As the field advances, future investigations should examine whether these biological signatures can be harnessed prognostically to identify patients at risk or monitor their response to stress-reduction interventions.

From a mechanistic standpoint, the hippocampus appears to be the anatomical and physiological fulcrum upon which memory vulnerability pivots. Chronic seizures, hippocampal atrophy, and mood disorders all contribute to memory dysfunction.^1,6 The high density of glucocorticoid receptors within the hippocampus renders it exquisitely sensitive to stress. Chronic dysregulation of the HPA axis, manifested as a sluggish cortisol decline, may impair this region's ability to inhibit further cortisol secretion, creating a maladaptive feedback loop. ⁵ This well-documented phenomenon may selectively compromise hippocampus-dependent memory while sparing other cognitive faculties. ⁵ Chronic stress and prolonged exposure to cortisol can damage the mesial temporal lobe. ⁵ Conversely, hippocampal injury from chronic TLE could in itself heighten basal cortisol output, perpetuating this bidirectional dysfunction. Future research would benefit from imaging-endocrine coupling studies to quantify hippocampal volume, functional connectivity, and HPA reactivity in tandem across phenotypes.

The globally impaired phenotype, by contrast, was marked by a higher burden of antiseizure medications, consistent with prior literature. ⁸ Yet intriguingly, Cano-López et al ⁵ did not find differences in affective symptoms or education across this group, diverging from earlier studies that linked global impairment to depression and lower education.^9,10 Similarly, while previous studies found an association of this phenotype with mesial temporal sclerosis, ⁹ such structural correlates were not observed in the current sample. Moreover, cortisol levels were not elevated in this group, suggesting that HPA-axis hyperactivity may not be the principal driver of global cognitive dysfunction. Importantly, these findings call for future multimodal research to examine structural, metabolic, and pharmacological contributors to this phenotype, potentially informing rational ASM selection and cognitive risk mitigation.

By contrast, the cognitively intact group demonstrated higher educational attainment and lower rates of mesial temporal sclerosis, reinforcing the concept of cognitive reserve as a protective factor. Interestingly, negative affectivity was correlated with elevated cortisol levels, but this was not phenotype-specific, suggesting that mood and endocrine responses may act as transdiagnostic modifiers rather than direct determinants of phenotype. Future studies should also explore whether strengthening cognitive reserve (eg, through cognitive stimulation or education-based interventions) could buffer against cortisol-driven cognitive decline or enhance resilience to surgical outcomes.

Crucially, 42% of this drug-resistant TLE cohort had intact cognition, making them particularly vulnerable to postoperative cognitive decline and raising the stakes for precision in presurgical risk stratification. ⁵ Cognitive phenotyping thus emerges not only as a descriptive tool but also as a practical guidepost for surgical decision-making. The integration of cortisol and affect measures may augment current presurgical workups by identifying patients with latent neuroendocrine vulnerability, offering new pathways for tailored counseling and intervention. Future directions should include prospective follow-up of these phenotypes postsurgery to evaluate neurocognitive outcomes and HPA normalization trajectories.

This study offers several key strengths. It reinforces that cognitive impairment in TLE is heterogeneous, proposes a biologically informed model of cognitive phenotyping, replicates cognitive phenotypes in a Spanish-speaking population outside North America, enhancing generalizability, and highlights that cognitive and mood profiles are integral to the disorder itself rather than mere consequences. However, some caveats are warranted. These include its cross-sectional design, which precludes causal inference, and a relatively small sample size, which may limit statistical power.

Taken together, the findings of Cano-López et al ⁵ advance a biologically informed model of cognitive heterogeneity in TLE, one in which structural pathology, stress response, affective regulation, and cognitive reserve converge. By bridging cognitive phenotypes with cortisol biomarkers, the study lays foundational groundwork for a biopsychosocial precision medicine approach to epilepsy care. Future research should capitalize on these insights by embedding biomarker-informed phenotyping into clinical trials, exploring interventions to modulate HPA function, and designing cognitive and affective therapeutics tailored to phenotype. These insights also have direct implications for personalized treatment planning and presurgical risk assessment. The goal is clear: to move beyond seizure freedom alone, and toward a fuller understanding and treatment of the cognitive toll of epilepsy.