Authors: Seonjeong Kim (Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea), Semina Jung (Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea), Seunghoon Lee (Department of Neurosurgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea), Eun Yeon Joo (Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea), Dae‐Won Seo (Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea), Young‐Min Shon (Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea; Department of Medical Device Management and Research, Samsung Advanced Institute for Health Science and Technology, Sungkyunkwan University, Seoul, Republic of Korea; Smart Healthcare Research Institute, Samsung Medical Center, Seoul, Republic of Korea)
Categories: Original Article, anterior thalamic nucleus DBS, cognition, deep brain stimulation, drug‐resistant epilepsy, hippocampal DBS
Source: Epilepsia Open
Doi: 10.1002/epi4.70107
Authors: Seonjeong Kim, Semina Jung, Seunghoon Lee, Eun Yeon Joo, Dae‐Won Seo, Young‐Min Shon
Deep brain stimulation (DBS) of the anterior thalamic nucleus (ATN) and hippocampus is an emerging therapy for drug‐resistant epilepsy (DRE) when resective surgery is not feasible. We aimed to evaluate the long‐term cognitive outcomes of these two DBS targets, hypothesizing that both interventions preserve cognitive function.
We conducted a retrospective analysis of DRE patients who underwent ATN‐DBS (n = 12) or hippocampal DBS (n = 10) and completed comprehensive neuropsychological testing before surgery and after ≥18 months of stimulation. Testing assessed general cognition, intellectual function, memory, language, and executive function. Changes in performance (post‐ minus pre‐DBS) were analyzed within each group and between groups, with appropriate correction for multiple comparisons.
At median follow‐ups of 44.33 months (ATN‐DBS) and 28.60 months (Hip‐DBS), both groups achieved comparable seizure reductions, particularly for disabling seizures (ATN‐DBS: 73.05%; Hip‐DBS: 76.76%). No significant cognitive decline was observed in either group across any domain. After FDR correction, no cognitive measure showed significant change (p > 0.05 for all). Unadjusted comparisons revealed modest improvement trends in visuospatial reasoning with ATN‐DBS (Perceptual Reasoning Index, p = 0.047 uncorrected) and semantic verbal fluency with hippocampal DBS (p = 0.041 uncorrected), but these did not remain significant after correction. Between‐group comparisons identified no significant cognitive differences.
ATN and hippocampal stimulation demonstrated comparable efficacy for seizure control while maintaining cognitive stability over long‐term follow‐up. Despite targeting key structures in memory and cognitive networks, neither approach produced measurable cognitive deterioration. These findings provide critical reassurance regarding the cognitive safety of DBS in epilepsy and suggest that target selection can prioritize optimizing seizure control without compromising cognitive function.
This study compared cognitive outcomes in patients with difficult‐to‐treat epilepsy who received deep brain stimulation (DBS) in one of two brain the anterior thalamic nucleus or the hippocampus. Both treatments effectively reduced seizures by over 70%. Importantly, after 2–4 years of continuous stimulation, neither treatment caused any decline in thinking abilities such as memory, language, nor problem‐solving – despite targeting brain regions critical for these functions. This provides reassurance that DBS represents a cognitively safe option for epilepsy patients who cannot undergo traditional surgery.
Key points DBS in ATN and hippocampus showed similar long‐term seizure reduction.No significant cognitive decline occurred with either DBS target.Minor cognitive improvements were observed but not statistically significant.Target selection can focus on seizure control without risking cognition.
Deep Brain Stimulation (DBS) has emerged as a transformative therapeutic option for patients with drug‐resistant epilepsy (DRE) who are unsuitable for resective surgery. Among various stimulation targets, the anterior thalamic nucleus (ATN) and hippocampus have garnered particular attention, supported by accumulating evidence of their dual capacity to modulate both seizure networks and cognitive functioning. These targets are of special interest due to their integral roles in distinct but interconnected neural circuits mediating consciousness, memory, and seizure propagation.
The landmark Stimulation of the Anterior Nuclei of Thalamus for Epilepsy (SANTE) trial provided pivotal insights into the cognitive impact of ATN‐DBS. ^1^ , ^2^ , ^3^ This multicenter investigation demonstrated remarkable stability – and in some cases, enhancement – of cognitive functions over a seven‐year follow‐up period, with patients exhibiting improvements in visual attention, executive function, visual construction, and immediate visual recall, without significant cognitive deterioration or depression. These findings have been further corroborated by subsequent studies, which revealed specific enhancements in verbal memory and verbal fluency following ATN‐DBS treatment over periods exceeding 12 months. ^4^ , ^5^ Recent evidence supports the long‐term efficacy and safety of Hip‐DBS. A previous study reported significant seizure reduction at 48‐month follow‐up without cognitive decline in drug‐resistant temporal lobe epilepsy patients. ^6^ Subsequently, our eight‐year review of 15 patients (11 with bilateral temporal lobe epilepsy and four with posterior epilepsy) demonstrated sustained seizure control, accompanied by stable or improved cognitive functions, particularly in nonverbal and visuospatial domains. ^7^
Despite the expanding body of evidence supporting DBS efficacy in epilepsy management, a critical knowledge gap persists in our understanding of target‐specific cognitive outcomes. Current literature has focused on isolated analyses of individual stimulation targets, leaving the comparative cognitive impacts of different DBS approaches unexplored. This limitation is particularly significant given that the neural networks underlying learning, memory, and executive functions are differentially engaged by distinct stimulation targets. The intricate interconnections between the anterior thalamic nucleus and hippocampus within the Papez circuit suggest potentially distinct yet complementary effects on cognitive domains, warranting systematic comparative investigation.
To address these critical knowledge gaps, we conducted a comprehensive comparative analysis of cognitive outcomes between ATN‐DBS and Hip‐DBS in DRE patients. This study represents the first direct comparison of target‐specific cognitive trajectories across multiple domains in patients receiving either ATN or Hip‐DBS over an extended period. We aimed to determine whether the stimulation target influences long‐term cognitive changes. Our findings are intended to guide a more personalized approach to DBS therapy in epilepsy – optimizing seizure control while monitoring cognitive outcomes based on individual patient profiles.
We conducted a retrospective analysis of prospectively collected data from patients who underwent DBS implantation for drug‐resistant epilepsy between January 2018 and December 2022 at Samsung Medical Center, Seoul, Korea. From an initial cohort of 30 patients (18 ATN‐DBS, 12 Hip‐DBS), 27 patients (16 ATN‐DBS, 11 Hip‐DBS) completed both baseline and follow‐up neuropsychological evaluations with a minimum follow‐up period of 18 months. To control for potential confounding effects of mood disorders on cognitive assessment, patients were stratified based on the presence of depression. This study focused on the non‐depressed cohort, comprising 22 patients (12 ATN‐DBS, 10 Hip‐DBS).
Inclusion (1) Confirmed diagnosis of drug‐resistant epilepsy according to International League Against Epilepsy (ILAE) criteria; (2) age ≥ 18 years at the time of DBS implantation; (3) failed response to at least three appropriate antiseizure medications; (4) unsuitable for resective surgery based on comprehensive presurgical evaluation; (5) completion of standardized neuropsychological testing before and after DBS implantation.
Exclusion (1) Presence of clinically significant depression (Beck Depression Inventory score > 20); (2) major psychiatric comorbidities; (3) progressive neurological disorders; (4) inability to complete comprehensive neuropsychological testing; (5) significant structural brain abnormalities that could interfere with DBS placement or efficacy.
Comprehensive data were collected through review of medical records and seizure diaries. All patients underwent detailed presurgical evaluations, including video‐EEG monitoring, high‐resolution 3 T brain MRI, [18F]‐FDG PET, and ictal/interictal SPECT, to characterize seizure focus and network. Baseline and follow‐up neuropsychological test results, clinical outcomes, surgical records, and post‐operative imaging were also reviewed.
A comprehensive neuropsychological evaluation was conducted at baseline (1–2 weeks before DBS implantation) and during follow‐up visits (mean follow‐up: ATN‐DBS 44.33 ± 30.25 months; Hip‐DBS 28.60 ± 15.74 months).
All assessments were administered by qualified neuropsychologists blinded to stimulation target and clinical outcomes. The test battery evaluated general cognitive function (K‐MMSE), intellectual function (Full‐Scale IQ and index scores from the K‐WAIS‐IV, including Verbal Comprehension Index [VCI], Perceptual Reasoning Index [PRI], Working Memory Index [WMI], and Processing Speed Index [PSI]), language (Korean‐Boston Naming Test, K‐BNT), verbal memory (Korean California Verbal Learning Test, K‐CVLT, including immediate recall and delayed recall tasks), visuospatial memory (Rey–Osterrieth Complex Figure Test, RCFT, including copy and memory trials with delayed recall and recognition), attention and working memory (forward/backward digit span and visual memory span), executive function (Controlled Oral Word Association Test [COWAT] for verbal fluency, Wisconsin Card Sorting Test [WCST], Stroop color‐word test, and Trail Making Test [TMT] Parts A and B), and mood (Beck Depression Inventory [BDI], Beck Anxiety Inventory [BAI], and Beck Hopelessness Scale [BHS], Table 1).
All tests were administered in a standardized environment, with appropriate rest periods to minimize fatigue. Raw scores were converted to age‐adjusted standard scores where applicable. To ensure consistency, all follow‐up assessments were conducted at similar times of day as baseline testing, and patients were tested during periods of seizure freedom (minimum 24 h post‐ictal).
The choice of DBS target was individualized based on a comprehensive presurgical evaluation, incorporating seizure semiology, long‐term video‐EEG monitoring, and multimodal neuroimaging (MRI, PET, SPECT). Hippocampal DBS was primarily indicated for patients with drug‐resistant epilepsy presumed to originate from a bitemporal network, particularly when evidence pointed to the mesial temporal lobes as the primary epileptic focus. In contrast, ATN‐DBS was selected for patients with more diffuse or multifocal seizure onsets, including those with focal seizures that rapidly generalize, or when a discrete epileptogenic zone could not be lateralized or localized.
All DBS procedures were performed under general anesthesia using a Leksell stereotactic frame system. Target selection and trajectory planning were conducted using preoperative volumetric MRI merged with stereotactic Computed Tomography (CT) images.
Bilateral ATN targets were identified using direct targeting methods based on 3 T MRI visualization. The target coordinates 5–6 mm lateral to the midline, 12 mm superior to the anterior commissure–posterior commissure (AC‐PC) line, and 2–3 mm anterior to the PC. Microelectrode recording was performed to confirm appropriate targeting, followed by permanent electrode implantation (Medtronic 3389 lead). Final electrode positions were verified using intraoperative fluoroscopy and post‐operative CT merged with preoperative MRI.
For hippocampal targeting, electrodes (Medtronic 3391 lead) were implanted bilaterally along the anterior–posterior axis of the hippocampus. Entry points were planned at the occipital region, with trajectories following the longitudinal axis of the hippocampus. The deepest contact was positioned in the hippocampal head, with the remaining contacts spanning the hippocampal body.
Following a two–four‐week recovery period, stimulation was initiated with the following ATN‐DBS: Frequency 130 Hz, pulse width 90 μs, voltage 1.5 V, continuous stimulation. ^8^
Hip‐DBS: Frequency 130 Hz, pulse width 120 μs, voltage 1.5 V, continuous stimulation. ^7^
Parameters were individually optimized based on clinical response and side effects during follow‐up visits, with most titration adjustments occurring within the first 3 to 6 months of therapy. At the time of the follow‐up neuropsychological assessment, all patients had been on stable stimulation parameters for at least 12 months, ensuring that the cognitive data reflect the effects of chronic, rather than acute, neuromodulation.
All statistical analyses were conducted using R software (version 4.4.1). We first assessed the normality of data distributions using the Shapiro–Wilk test to determine the appropriate use of parametric or non‐parametric tests. Continuous demographic and clinical variables were compared between groups using independent samples t‐tests or Mann–Whitney U tests (for non‐normally distributed data), and categorical variables were compared with chi‐square or Fisher's exact tests.
For cognitive outcomes, we compared pre‐ and post‐DBS scores within each group using paired t‐tests (or Wilcoxon signed‐rank tests if data were non‐normally distributed). This focus on intra‐individual change scores was chosen specifically to account for baseline differences in cognitive function and clinical characteristics between the two treatment groups, thereby allowing for a valid comparison of the longitudinal effects of each DBS target.
Between‐group differences in change scores were evaluated using independent t‐tests (or Mann–Whitney tests as appropriate) with significance established at p < 0.05 (two‐tailed). To account for multiple comparisons across cognitive measures, a false discovery rate (FDR) correction using the Benjamini–Hochberg procedure was employed to balance between Type I error control and statistical power. For all other analyses, statistical significance was set at p < 0.05 (two‐tailed). Data are presented as mean ± standard deviation (SD). Analyses were reviewed by a biostatistician to ensure appropriate methodology.
Our analysis included 22 patients with drug‐resistant epilepsy, comprising 12 patients in the ATN‐DBS group and 10 in the Hip‐DBS group (Table 2). The ATN‐DBS group underwent neuropsychological follow‐up for a mean of 44.33 ± 30.25 months, while the Hip‐DBS group was followed for 28.60 ± 15.74 months. The ATN‐DBS cohort was predominantly male (83%) with a mean age of 33.17 ± 9.41 years, whereas the Hip‐DBS cohort had equal gender distribution and a mean age of 39.80 ± 8.57 years. Educational level was comparable between groups. Both groups were on similar numbers of antiseizure medications at baseline (ATN‐DBS: 4.67 ± 1.67; Hip‐DBS: 4.80 ± 1.62 concomitant medications). Cognitive‐impairing antiseizure medications such as topiramate and zonisamide were used by half of the patients in each group (ATN‐DBS: 6/12; Hip‐DBS: 5/10) at baseline. While some medication adjustments occurred during follow‐up, these specific cognitive‐affecting drugs remained unchanged, ensuring stable medication‐related cognitive effects throughout the study.
All patients experienced focal impaired awareness seizures (FIAS), and most had focal to bilateral tonic–clonic seizures (FBTCS) as well (ATN‐DBS: 11/12 patients; Hip‐DBS: 9/10 patients). A subset of ATN‐DBS patients (3/12) also had focal aware seizures, whereas none in the Hip‐DBS group reported focal aware seizures. The predominant epilepsy syndrome differed between bilateral frontotemporal lobe epilepsy (5/12 patients) was most common in the ATN‐DBS group, whereas bilateral temporal lobe epilepsy (5/10 patients) was most common in the Hip‐DBS group.
Structural abnormalities on MRI were identified in 50% (6/12) of ATN‐DBS patients and 80% (8/10) of Hip‐DBS patients. The ATN‐DBS group included cases of HSV‐related bilateral hippocampal sclerosis and post‐traumatic encephalomalacia. The Hip‐DBS group showed a higher prevalence of bilateral hippocampal pathology, with five patients exhibiting changes due to various etiologies, including HSV encephalitis, unknown encephalitis, and neonatal hemorrhage.
Two patients in the ATN‐DBS group had prior neurostimulation treatments. One patient with right frontoparietal cortical dysplasia underwent stereoelectroencephalography (SEEG) after surface EEG showed non‐lateralized ictal onsets; SEEG revealed bilateral frontal onsets, leading to DBS therapy. Another patient had previously undergone a right anterior temporal lobectomy (with incomplete amygdalo‐hippocampectomy) for right hippocampal sclerosis; after seizure recurrence with bilateral temporal onsets, this patient tried vagus nerve stimulation (VNS) with minimal benefit before proceeding to ATN‐DBS.
Both DBS targets produced marked reductions in seizure frequency. We categorized seizures into non‐disabling seizures (NDS: primarily focal aware seizures) and disabling seizures (DS: focal impaired awareness and focal to bilateral tonic–clonic seizures). The ATN‐DBS group showed substantial decreases in both mean monthly NDS frequency fell from 34.56 to 11.71, and DS frequency from 18.33 to 2.46. Similarly, the Hip‐DBS group's monthly NDS dropped from 34.00 to 9.38, and DS from 9.38 to 1.70 (Table 3). The overall seizure reduction ratios were comparable between groups, with both showing more pronounced improvement in disabling seizures (ATN‐DBS: 73.05 ± 29.11% reduction in DS; Hip‐DBS: 76.76 ± 19.22% reduction in DS) than in NDS (ATN‐DBS: 69.38 ± 24.55%; Hip‐DBS: 59.35 ± 29.13%). These results indicate that both stimulation targets effectively suppress seizure activity, especially for more severe seizure types, with no significant difference in seizure control efficacy between ATN and hippocampal stimulation.
At baseline, there were some differences in cognitive test scores between groups. Notably, the ATN‐DBS group had a lower initial Full‐Scale IQ (66.17 ± 15.69) compared to the Hip‐DBS group (74.00 ± 12.04).
Following DBS therapy, both groups demonstrated comparable patterns of cognitive change, with no statistically significant between‐group differences after correcting for multiple comparisons (Table 4). In the ATN‐DBS group, we observed improvements or stable performance across multiple cognitive domains, but none of the changes reached significance under the stringent corrected threshold. For example, Full‐Scale IQ showed an upward trend post‐DBS, and the Perceptual Reasoning Index (PRI) increased (pre‐DBS 71.42 ± 12.09 to post‐DBS 77.17 ± 14.55, p = 0.047 uncorrected). However, this improvement in PRI did not remain significant after FDR correction. The Verbal Comprehension Index (VCI) showed a slight decline on average (81.33 ± 17.57 to 80.50 ± 18.88, p = 0.360 after adjusting test method), which was not statistically significant. The Working Memory Index (WMI) and Processing Speed Index (PSI) in the ATN group changed minimally (both indices increased by only ~1 point on average, p > 0.7). Thus, overall cognitive ability in the ATN‐DBS group was preserved post‐DBS, with a slight non‐significant improvement in certain nonverbal reasoning skills.
The Hip‐DBS group displayed a generally similar cognitive profile over time. Most IQ subscales remained stable or showed minor, non‐significant declines. Notably, the PRI in the Hip‐DBS group remained essentially unchanged (82.56 ± 9.81 at baseline vs. 82.50 ± 12.03 at follow‐up, p = 0.205). We observed a modest improvement in semantic verbal fluency in the Hip‐DBS group (COWAT semantic fluency score increased from 19.10 ± 7.81 to 22.70 ± 7.70, p = 0.041 uncorrected). However, like the ATN group's PRI change, this improvement did not meet the corrected significance threshold (p > 0.05 after correction). Other cognitive measures in the Hip‐DBS group, such as VCI and executive function tests, showed no significant changes (for instance, VCI declined slightly from 82.56 ± 17.77 to 79.00 ± 17.66, p = 0.874). In summary, the Hip‐DBS group also exhibited no statistically significant cognitive changes between baseline and follow‐up when accounting for multiple comparisons, indicating cognitive stability.
Importantly, both groups showed a trend toward improvement in visuospatial memory performance. On the Rey–Osterrieth Complex Figure Test (RCFT) recognition trial, which assesses visuospatial recognition memory, scores increased in both the ATN‐DBS group (from 18.92 ± 3.00 to 19.58 ± 3.03) and the Hip‐DBS group (17.20 ± 4.08 to 19.30 ± 1.64). These represented notable gains for individual patients; however, in our analysis, the changes did not reach statistical significance (ATN‐DBS p = 0.314; Hip‐DBS p = 0.316 uncorrected).
Performance on other neuropsychological tests – including language (K‐BNT), verbal learning (K‐CVLT immediate recall and short‐delay recall), and executive function tasks (WCST, Stroop test, and TMT) – showed mixed patterns of minor increases or decreases within each group, but none of these changes were statistically significant (all corrected p values > 0.05). In summary, after FDR correction, neither the ATN‐DBS nor Hip‐DBS group demonstrated any significant within‐group cognitive change, and there were no significant differences between the two groups in cognitive outcomes.
Regarding psychological measures, both groups exhibited favorable trends in mood. Depression scores (BDI) decreased in both ATN‐DBS (mean BDI 18.42 to 15.92) and Hip‐DBS (17.70 to 12.80) patients, and anxiety levels (BAI) also declined slightly (ATN‐DBS 12.17 to 9.83; Hip‐DBS 9.80 to 8.80). Feelings of hopelessness (BHS) showed a minor increase on average in both groups (ATN‐DBS 8.75 to 9.55; Hip‐DBS 9.60 to 10.44). None of these changes in BDI, BAI, or BHS were statistically significant (p > 0.1 for all, uncorrected). Thus, there was no indication of a worsening in mood or psychiatric status following DBS; if anything, depression and anxiety tended to improve slightly in both groups, although not to a significant degree.
Our investigation indicates that DBS targeting either the anterior thalamic nucleus or the hippocampus yields comparable cognitive outcomes in patients with chronic, drug‐resistant epilepsy, while providing similar efficacy in seizure control. In contrast to potential concerns about cognitive decline, we found no evidence of cognitive deterioration in either group across a comprehensive battery of neuropsychological tests, despite chronic stimulation of brain structures central to memory and cognitive processing. Both groups experienced substantial seizure reduction (particularly for disabling seizures) while maintaining cognitive stability throughout the extended follow‐up period. Although the ATN‐DBS group exhibited a lower baseline FSIQ, likely reflecting more diffuse underlying epilepsy syndromes, the analysis of change scores demonstrates that the cognitive trajectory over time did not significantly differ between targets, suggesting comparable long‐term cognitive safety.
Although no significant target‐specific cognitive effects were confirmed after correction for multiple comparisons, we did observe small trends in different cognitive domains for each group. The ATN‐DBS group showed a modest improvement in the Perceptual Reasoning Index (a measure of nonverbal, visuospatial reasoning ability), whereas the Hip‐DBS group demonstrated a slight gain in semantic verbal fluency. Both DBS targets were associated with a shared trend of improvement in visuospatial recognition memory. However, none of these trends reached statistical significance after appropriate correction. Thus, our results underscore that cognitive function remained stable overall under chronic DBS in both targets, without a clear advantage or detriment attributable to one stimulation target versus the other.
Hitherto, clinicians and patients have had to extrapolate from these separate bodies of work, which often involve different patient populations and methodologies, making it difficult to weigh the potential cognitive risks of one target versus another. The present study uniquely advances this field by providing the first direct, head‐to‐head comparison of long‐term cognitive outcomes. By demonstrating comparable cognitive preservation across both groups, this work reinforces the general principle that neuromodulation for epilepsy can be achieved without sacrificing cognitive health. More importantly, it provides the first piece of direct comparative evidence to suggest that the choice of target can be guided primarily by the characteristics of the patient's seizure network, rather than by overriding concerns of a differential cognitive risk.
Despite substantial seizure reduction, neither group showed cognitive improvement following DBS therapy, suggesting seizure control alone is insufficient for cognitive gains. Patient factors including long‐standing epilepsy, polytherapy with antiseizure medication, and structural abnormalities may limit cognitive recovery even after successful seizure reduction. Future controlled studies and longer follow‐up are needed to determine specific cognitive benefits of neuromodulation and capture potential delayed improvements.
The preservation of cognition across both stimulation targets suggests engagement of shared neural networks despite different entry points into those networks. The ATN and hippocampus are interconnected through the Papez circuit and other limbic pathways, so stimulating either node may influence a broader memory and executive network. Our observation of cognitive stability across both ATN and hippocampal DBS supports the notion that DBS can modulate epilepsy networks without compromising cognitive networks, likely by normalizing dysfunctional neural activity.
The mechanisms underlying cognitive preservation following DBS warrant careful consideration, particularly given the distinct yet complementary roles of ATN and hippocampus within the Papez circuit. Our observation of cognitive stability across both stimulation targets suggests the engagement of shared neural networks despite different entry points into the circuit.
ATN‐DBS has been proposed to modulate epileptic activity through disruption of thalamic networks that mediate seizure propagation. Animal studies have shown that ATN stimulation increases glucose uptake in thalamic nuclei and bilateral hippocampi through efferent thalamo‐hippocampal projections. ^9^ This bilateral network modulation may contribute to the maintenance of cognitive function observed in our ATN‐DBS cohort.
Hippocampal DBS, operating more directly within the mesial temporal lobe, likely influences cognition through dual mechanisms. First, by suppressing ictal activity through direct modulation of the hippocampal formation, as suggested by inhibition of epileptogenic neurons via the perforant pathway. ^10^ Second, by modulating broader epileptogenic networks, influencing both ictal focus and propagation pathways. ^11^ , ^12^ , ^13^ , ^14^ The cognitive stability observed despite direct hippocampal stimulation suggests that successful seizure control may facilitate the maintenance of physiological network dynamics, enabling preserved cognitive processing across distributed neural systems. ^15^
Our findings provide valuable insights into the long‐term cognitive effects of DBS, with the ATN‐DBS group maintaining cognitive stability despite a longer follow‐up duration (44.33 ± 30.25 months) compared to the Hip‐DBS group (28.60 ± 15.74 months). The longer follow‐up in the ATN‐DBS group, while a point of difference with the Hip‐DBS cohort, provides reassuring evidence for the durability of cognitive safety over an extended period, well past the window for transient post‐operative or initial stimulation‐related effects. This extended observation enhances confidence that the cognitive stability observed is a lasting outcome of chronic neuromodulation.
This observation suggests that the cognitive preservation associated with DBS therapy may be maintained over extended periods – a finding with important clinical implications for the long‐term management of drug‐resistant epilepsy.
This temporal dimension of cognitive outcomes finds parallels in other neuromodulation approaches. In VNS, chronic stimulation spanning more than 6–12 months has been associated with subtle changes in cognitive functions, while acute stimulation demonstrates minimal impact. ^16^ , ^17^ , ^18^ Although DBS operates through distinct mechanisms, the neural networks engaged by both ATN and hippocampal stimulation share considerable overlap with VNS through their connections with the Papez circuit and broader limbic system engagement.
The comparable cognitive profiles between groups despite different follow‐up durations reinforce the durability of cognitive safety with chronic DBS therapy. This finding is particularly reassuring given the progressive cognitive decline often observed in untreated chronic epilepsy, suggesting that DBS may help stabilize cognitive trajectories in this vulnerable population.
The initial trends observed in cognitive domains, such as the modest improvement in perceptual reasoning with ATN‐DBS and semantic verbal fluency with hippocampal DBS (though not statistically significant after correction), merit consideration within the context of network‐based neuromodulation. Traditional perspectives have linked cognitive patterns to hemispheric specialization – with verbal comprehension deficits typically associated with left hemisphere dysfunction and perceptual organization deficits with right hemisphere involvement. ^19^ , ^20^
Prior research using voxel‐based lesion symptom mapping has demonstrated correlations between Wechsler Adult Intelligence Scale (WAIS) index scores and specific hemispheric regions, with PRI performance particularly associated with right parietal, occipito‐parietal, and superior temporal cortical function. ^21^ It is conceivable that ATN stimulation might influence these networks, either through direct modulation of right‐hemispheric function or through effects on interhemispheric balance, although our findings do not provide statistical evidence for such effects after appropriate correction.
The trend toward improvement in semantic verbal fluency in the Hip‐DBS group invites theoretical consideration of the hippocampus's potential role in semantic processing. Recent perspectives conceptualize semantic memory as a dynamic, constructive system that shares features with episodic memory, including flexibility and relational processing capabilities. ^22^ Neurophysiological studies using intracranial recordings have reported that hippocampal theta oscillations may encode representational distances in semantic space, suggesting that theta rhythms could support navigation through cognitive maps in the mesial temporal lobe. ^23^ , ^24^
This theoretical framework provides context for understanding potential mechanisms through which different DBS targets might interact with distinct cognitive processes. While our data do not provide statistically significant evidence for target‐specific enhancement effects, these potential mechanisms warrant further exploration in larger studies with more targeted cognitive assessments.
A pivotal finding of our study is the absence of significant cognitive deterioration across multiple domains in both ATN‐DBS and Hip‐DBS groups over extended follow‐up periods. This finding directly addresses one of the most persistent concerns regarding chronic neuromodulation of deep brain structures involved in cognitive processing, the potential for iatrogenic cognitive impairment.
The anterior thalamic nucleus and hippocampus represent critical nodes in memory and cognitive networks, with the ATN serving as a key relay within the Papez circuit and the hippocampus functioning as the cornerstone of declarative memory formation. The preservation of cognitive function despite chronic high‐frequency stimulation of these structures challenges earlier apprehensions about potential disruption of normal neural processing. While we did not observe statistically significant cognitive improvements after appropriate correction for multiple comparisons, the stability of cognitive performance across comprehensive neuropsychological assessment batteries provides reassuring evidence for the cognitive safety of both approaches.
Our findings align with emerging literature suggesting that properly targeted neuromodulation can achieve therapeutic benefits without compromising cognitive function. The landmark SANTE trial demonstrated cognitive stability – and in some domains, modest enhancement – over 7 years of ATN‐DBS. ^3^ Similarly, longitudinal studies of hippocampal stimulation have reported cognitive preservation over extended periods. ^6^ Our results extend these observations by providing direct comparative evidence across different stimulation targets, reinforcing the cognitive safety profile of both approaches despite their distinct anatomical positioning within memory networks.
Furthermore, while mood was formally assessed, a standardized questionnaire, such as the cognitive subscale of the Quality of Life in Epilepsy Inventory (QOLIE‐31‐P), was not employed to capture patients' subjective reports of cognitive function. Such a measure could provide a valuable dimension to our objective findings, allowing for a more complete clinical picture by correlating patient experience with neuropsychological performance. From a clinical perspective, these findings have substantial implications for patient counseling and therapeutic decision‐making. The cognitive safety demonstrated in our study may help alleviate patient and clinician concerns regarding potential tradeoffs between seizure control and cognitive function, potentially expanding access to neuromodulation therapies for patients who might otherwise be excluded due to cognitive concerns.
While our study provides novel insights into the cognitive outcomes of DBS in epilepsy, several limitations must be acknowledged. First, the modest sample size (n = 22) limits statistical power, especially for detecting subtle cognitive changes or differences between groups. We employed strict FDR correction to avoid type I errors, but this also raises the risk of type II errors (missing real effects). It is possible that with a larger cohort, some of the trends we observed might reach significance, or small differences between targets might emerge.
A post‐hoc power analysis indicates that a sample of approximately 64 patients per group would be needed to reliably detect a medium‐sized difference in cognitive outcomes between targets with 80% power. Therefore, our findings should be considered exploratory, and the existence of smaller, yet potentially meaningful, target‐specific cognitive effects cannot be ruled out. This highlights the critical need for future collaborative, multicenter studies to achieve the statistical power necessary to definitively compare these interventions.
Second, the ATN‐DBS and Hip‐DBS groups had different follow‐up durations. Although follow‐up length was not significantly different in a statistical sense, the mean difference (roughly 15 months longer for ATN‐DBS) could have influenced the cognitive trajectories as discussed. Longer follow‐up in the Hip‐DBS group might reveal additional cognitive changes, so interpretations regarding target‐specific effects should be tempered by this disparity.
Third, our study focused exclusively on patients without clinical depression to isolate the effects of DBS on cognition; while this strengthens internal validity, it may limit generalizability. Many real‐world patients have mood comorbidities, and our findings may not fully extend to those populations.
Another limitation is the comprehensive test battery itself, which increases the multiple‐comparison burden. We chose to examine a broad range of cognitive domains given the exploratory nature of this first direct comparison between targets. Future studies might streamline assessments or use composite cognitive scores to reduce the number of comparisons. Additionally, practice effects on some tests could partially mask or mimic true cognitive changes, although our long inter‐test interval (≥18 months) helps mitigate this concern.
We did not include a non‐DBS control group of DRE patients, so we cannot definitively separate DBS‐related cognitive changes from those due to natural disease progression or chronic antiseizure medication use. However, prior work suggests that without intervention, chronic epilepsy patients are more likely to experience cognitive decline rather than improvement, lending support to the positive nature of the cognitive stability we observed.
Future research should address these limitations through larger, multicenter studies with standardized cognitive follow‐up protocols. It would be valuable to integrate functional neuroimaging or electrophysiological monitoring to correlate cognitive performance with DBS‐induced network modifications. For instance, neuroimaging could reveal whether activation or connectivity patterns change in memory circuits under ATN vs. hippocampal stimulation, providing biological evidence of cognitive effects (or lack thereof).
Exploring optimal stimulation parameters specifically for cognitive outcomes is another area of interest. It remains unknown whether different stimulation settings (frequency, cycling patterns, etc.) might enhance cognitive performance while maintaining seizure control – this would require systematic investigation, possibly through adaptive DBS systems or trials adjusting parameters with cognitive endpoints in mind.
Finally, comparative studies including other neuromodulation modalities, such as responsive neurostimulation (RNS) or continued pharmacotherapy, could provide a broader context for interpreting the cognitive effects of DBS. By examining how DBS compares to these alternatives, we can better understand if the cognitive stability we observed is unique to DBS or part of a general effect of improved seizure control.
This investigation demonstrates cognitive stability following DBS therapy in drug‐resistant epilepsy, with both ATN and hippocampal stimulation achieving comparable seizure control without significant adverse effects on cognitive function. While our initial analyses suggested potential target‐specific cognitive effects, after appropriate correction for multiple comparisons, these differences did not maintain statistical significance. Nevertheless, the absence of cognitive deterioration across comprehensive neuropsychological assessment batteries despite chronic stimulation of structures central to memory and cognitive processing represents an important finding with significant clinical implications.
These results provide reassurance regarding the cognitive safety of both ATN‐DBS and Hip‐DBS over extended follow‐up periods, addressing a primary concern in neuromodulation therapies for epilepsy. The comparable efficacy in seizure control coupled with cognitive preservation suggests that both approaches represent viable therapeutic options for patients with drug‐resistant epilepsy who are unsuitable for resective surgery.
Future investigations focusing on larger cohorts with standardized long‐term follow‐up and more targeted cognitive assessments will be crucial for further characterizing the cognitive profiles associated with different stimulation targets and potentially identifying patient characteristics that might predict optimal responses to specific neuromodulation approaches.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.