Annals of Neurology

Objective: Predicting specific cognitive, psychiatric, and health-related sequelae in patients after acute traumatic brain injury (TBI) remains an important but challenging clinical problem. Acute phase computed tomography (CT) scans acquired show hemorrhagic contusions, a common type of traumatic pathology. However, whether CT-measured contusions predict long-term sequelae is uncertain. Methods: We established a Screening Cohort of patients with acute TBI who received care at a single TBI Model Systems (TBIMS) inpatient rehabilitation facility. Regions of hemorrhagic contusion and edema were labeled on acute brain CT scans using the fully-automated Brain Lesion Analysis and Segmentation Tool (BLAST-CT). We screened 198 outcome variables at 1-year post-injury for association with acute hemorrhagic contusion volume using the Harrell's Concordance index (C-index), controlling for multiple comparisons using 5,000 outcome permutations. Finally, we tested whether the significant associations in the TBIMS database replicated in acute (Transforming Research and Clinical Knowledge in TBI [TRACK-TBI]) and chronic (Vietnam Head Injury Study [VHIS]) external validation cohorts. Results: The TBIMS Screening Cohort included 345 participants (mean {+/-} SD age: 55.7 {+/-} 21.5 years) with median [IQR] contusion volume 2.3 cc [0.1, 14.6]. Among 198 candidate outcome variables, only delayed seizures were significantly associated with acute hemorrhagic contusion volume (C-index = 0.81; PFWE = 0.007). Contusion volume was not significantly associated with commonly-used measures of global functioning like the Glasgow Outcome Scale Extended, (C-index = 0.55; PFWE = 1). Within the screening cohort, 30 ccs was the optimal volume threshold for discriminating patients with versus without delayed seizures (OR 12.6, 95% CI: [4.6, 34.3]). Contusions larger than 30 cc remained significantly associated with delayed seizures in two external cohorts: (TRACK-TBI OR 4.1 [1.5, 11.2]; VHIS OR 3.2 [1.7, 6.2]). Interpretation: Across three cohorts of patients with TBI, CT-derived contusion volume is robustly associated with the development of delayed seizures, in contrast to commonly-used outcomes measuring global functioning. A 30-cc volume threshold can be used to improve epilepsy prediction models and enrich populations for clinical trials.

Objective. Continuous, unsupervised monitoring of cognitive brain responses has long been constrained by the demands of laboratory EEG. Whether the P300 event-related potential, an established marker of attention and cognitive processing, can be elicited as an incidental byproduct of genuine gameplay, recorded with a minimal wearable EEG system under unsupervised home conditions, has not been established. Approach. Ten healthy adults played a gamified visual oddball task in which infrequent target stimuli (green gates) were embedded among frequent non-targets (red gates) within a continuous third-person running game. EEG was recorded with a four-channel dry-electrode headband (EEG channels: O1, O2, T3, T4; forehead reference; 250Hz) with self-mounted electrodes in a home setting, without experimenter supervision. Group-level effects were assessed with cluster-based permutation tests and peak-amplitude tests. Single-trial classification used linear discriminant analysis (LDA) with four features per channel (16 total). Additional analyses included a within-subject comparison with a classical visual oddball paradigm using identical hardware, pilot data from a patient with relapsing-remitting multiple sclerosis, within-subject stability across 48 sessions, and pilot recordings with a headphone form factor. Main results. A robust P300-like difference wave emerged on all four channels at the group level (cluster-based permutation tests, p < 0.05), with individual-level detection in 8 of 10 participants (exact binomial p < 0.001). Single-trial LDA yielded a median cross-validated AUC of 0.730 (95% CI 0.672-0.820), with 9 of 10 participants exceeding chance. In a within-subject comparison, waveform morphology was closely preserved relative to a classical laboratory oddball, and classification performance was markedly higher in the game condition (AUC 0.820 versus 0.555). A patient with relapsing-remitting multiple sclerosis produced a clear P300 (AUC 0.853) with latencies within the healthy range. Within-session split-half reliability was high (r > 0.70 on three of four channels), though between-session reliability was near zero across 48 sessions in one participant, with a declining classification trend over time. Pilot recordings with a headphone form factor also yielded a P300-like deflection. Significance. These results demonstrate that the P300 can be elicited as a gameplay-integrated neural readout during genuine gameplay with a wearable, dry-electrode EEG system under unsupervised conditions. Gamification does not compromise P300 elicitation; in the within-subject comparison, it enhanced single-trial discriminability. The findings indicate that gamified, home-based P300 monitoring is achievable with minimal hardware and provide preliminary evidence for applicability in clinical populations, most notably multiple sclerosis, where P300 has established biomarker value but where the logistical burden of laboratory assessment currently precludes longitudinal use.

Background and Objectives: Older adults with epilepsy are at increased risk for Alzheimer's disease (AD), yet the mechanisms underlying this association remain poorly understood. We applied a validated AD neuroimaging signature to older adults with epilepsy to examine 1) whether older adults with epilepsy mirror AD-related changes, 2) associations with clinical, cognitive, and plasma biomarker outcomes, and 3) utility for identifying subgroups at heightened risk for cognitive decline. Our multicenter, prospectively enrolled cohort allowed for direct examination of differences in AD signatures between those with early-onset and late-onset unexplained epilepsy. Methods: Participants included 449 older adults: 87 with focal epilepsy from the multicenter Brain Aging and Cognition in Epilepsy (BrACE) cohort (age=66.10 [SD=6.86], including early-onset (<55 years at seizure onset) and late-onset ([&ge;]55 years at seizure onset) epilepsy); 362 from the Alzheimer's Disease Neuroimaging Initiative (ADNI), including cognitively unimpaired (CU) healthy controls and individuals with mild cognitive impairment (MCI) or AD dementia. An AD signature was derived from regional cortical thickness and hippocampal volume weighted by their sensitivity to AD-related neurodegeneration in prior work. Associations between the AD signature, epilepsy characteristics, plasma biomarkers ({beta}-amyloid 42/40, phosphorylated tau [pTau217, pTau181], neurofilament light chain [NfL]), and cognition were evaluated in BrACE. Results: Participants with epilepsy demonstrated more AD-like signatures compared to ADNI CU controls ({beta}= -0.43, p<0.001), reflecting reduced thickness/volume in AD-vulnerable regions. This effect was stronger among early-onset ({beta}= -0.57) versus late-onset ({beta}= -0.26) epilepsy. In BrACE, the AD signature correlated with NfL ({beta}= -0.30, p=0.050), memory performance ({beta}= 0.30, p=0.006), and predicted greater odds of cognitive impairment specifically among those with early-onset, but not late-onset, epilepsy (interaction p=0.043). Further, among those with early-onset epilepsy, the AD signature significantly improved identification of cognitive impairment over and beyond the effects of plasma AD biomarkers (p=0.041). Findings were similar when examining the effects of epilepsy duration rather than epilepsy onset age. Discussion: AD neuroimaging signatures may help identify clinically meaningful subgroups among older adults with epilepsy, particularly when integrated with AD biomarkers. Findings support a multimodal framework for assessing AD-related risk in epilepsy and highlight interactive effects of epilepsy chronicity and AD-related processes that can influence cognitive outcomes.

Objective Memory impairment is a frequent comorbidity of focal epilepsy, incompletely explained by seizure frequency or structural pathology. Ictal and postictal hippocampal dysfunction disrupt memory processes, but their cumulative impact remains poorly quantified. This study introduces cumulative hippocampal seizure-related burden metrics and examines their association with long-term memory consolidation. Methods Twenty consecutive patients undergoing stereo-EEG in Marseille (2016-2018) were prospectively included. Continuous stereo-EEG recordings between two memory assessments (30 minutes and one week post-encoding) were analysed. Hippocampal ictal involvement and durations were assessed using epileptogenicity markers and visual stereo-EEG analysis. The postictal period was quantified using permutation entropy. Cumulative hippocampal seizure-related burden metrics (ictal, postictal and combined: c-HipSZB) were computed across hippocampus-involving ictal events. Verbal and visual memory were assessed using standardized recall and recognition tasks. Associations were examined using univariate and multivariate analyses. Results Higher dominant-hemisphere hippocampal burden was associated with poorer one-week verbal memory (performance and retention), independently of most covariates. Higher c-HipSZB was associated with lower total recall performance (RT; free + cued) and RT retention ({beta} = -25.04 and -23.88; R2 = 0.57 and 0.53; p < 0.05) and accounted for the greatest variance in both outcomes (adjusted R2= 0.59 and 0.53; {beta} = -25.45 and -24.27; p < 0.01), particularly when adjusting for epilepsy duration. No robust associations were observed between non-dominant-hemisphere hippocampal seizure-related burden metrics and visual memory. Effects predominantly involved recall. Interpretation Cumulative ictal-postictal hippocampal dysfunction is a major determinant of impaired long-term verbal memory consolidation in focal epilepsy.

BackgroundAccurate diagnosis of multiple system atrophy (MSA) is critical for clinical management and efficient trial designs, yet remains challenging, particularly distinguishing MSA (especially cerebellar-subtype [MSA-C]) from sporadic adult-onset ataxia (SAOA). Combining a marker of neuroaxonal degeneration, neurofilament light chain (NfL), with a marker of the pathogenic MSA hallmark, -synuclein seeding activity, may define a mechanistically-informed CSF signature of MSA, enabling sensitive and specific differentiation from SAOA even in early disease. MethodsWe analyzed 60 cross-sectional patient CSF samples (n=32 clinically diagnosed MSA [MSAclin] 22/32 MSA-C; n=28 SAOA) for NfL (Simoa) and -synuclein seeding activity (seed amplification assay [synSAA], Piperazine-N,N-bis(2-ethanesulfonic acid)-based), and assessed diagnostic accuracy, disease-duration correlations, and trial power using biomarker-based stratification. ResultsAge-adjusted NfL was higher in MSAclin than SAOA (3859 vs. 997pg/mL), yielding 96.9% sensitivity and 85.7% specificity. SynSAA was concordant with clinical diagnosis (25/32 MSAclin synSAA-positive; 23/28 SAOA synSAA-negative), with 78.1% sensitivity and 85.2% specificity (all confirmed in MSA-C subgroup). Both biomarkers displayed divergent trajectories with disease duration: NfL peaked early before declining (r=-0.45, p=0.01); whereas synSAA maximum fluorescence intensity increased (r=0.42, p=0.016), suggesting greater synSAA signal with accumulating MSA burden. Integrating both biomarkers in MSA treatment trials allows sample-size reduction by 20% versus NfL alone. ConclusionsCSF NfL and synSAA capture complementary aspects of MSA biology: while NfL provides high diagnostic accuracy for MSAclin, peaking early, synSAA adds mechanistic specificity for -synuclein seeding activity and might allow target engagement assessment. Combined, they might enable biological diagnostic frameworks, molecular trial stratification, and treatment monitoring in MSA. Key messagesO_ST_ABSWhat is already known on this topicC_ST_ABSWhile highly warranted for clinical management and efficient treatment trial design, accurate diagnosis of multiple system atrophy (MSA) against overlapping and reciprocally mimicking conditions such as sporadic adult-onset ataxia (SAOA) remains clinically challenging, especially in early disease stages. A mechanistically informed biofluid signature of MSA might enable sensitive and specific differentiation from SAOA, even in early disease stage. Recently merging molecular markers reflecting neuroaxonal damage (NfL) and -synuclein seeding activity (measured by the seed amplification assay; synSAA) might here show particular promise. What this study addsThis is the first study to systematically assess the ability of both CSF NfL and CSF -synuclein seeding activity to distinguish clinically diagnosed MSA (MSAclin) from SAOA, thereby offering a window into underlying MSA biology in patients in vivo. Our findings suggest that the rate of axonal degeneration is most pronounced in early MSA disease stages but decreases with longer disease duration; whereas -synuclein seeding signal activity increases as MSA-related disease burden accumulates. Finally, it demonstrates the impact of a combined molecular fluid signature of MSA for improving trial design: a biomarker-based stratification of MSA subjects in future MSA treatment trials combining NfL plus -synuclein seeding activity allows to reduce sample sizes by 20% compared to NfL alone. How this study might affect research, practice or policyThe findings from this study may help to molecularly diagnose patients with MSA against overlapping and reciprocally mimicking conditions such as SAOA, in particular and even in early disease stages. Moreover, they might lay the foundation for a future biologically-informed diagnostic framework of MSA; support trial stratification for more efficient upcoming MSA treatment trials; and might facilitate molecular treatment effect monitoring in MSA, in particular in synuclein-targeted treatment trials.

Background: Progressive multiple sclerosis (MS) is characterized by ongoing neurodegeneration and limited therapeutic options. Circulating metabolites provide insight into disease biology, yet biomarkers that predict disability progression and reflect treatment response are lacking. We aimed to identify metabolomic signatures associated with longitudinal MRI measures of brain atrophy and to evaluate whether ibudilast treatment was associated with metabolite trajectories over time. Methods: We repeatedly profiled 1,726 plasma metabolites using untargeted UPLC-MS/MS in 244 participants from the 96-week SPRINT-MS randomized trial of oral ibudilast, up to 100 mg daily, versus placebo. Weighted gene co-expression network analysis was used to derive groups of related metabolites. Associations between baseline metabolite groups and longitudinal MRI outcomes were evaluated using linear mixed-effects models adjusted for demographic, clinical, and treatment covariates. The primary outcome was the rate of whole-brain atrophy measured by brain parenchymal fraction (BPF), defined as the proportion of intracranial volume occupied by brain tissue. Secondary outcomes included white matter fraction (WMF), gray matter fraction (GMF), and cortical thickness (CTH). Metabolite groups nominally associated with MRI outcomes, defined as p < 0.05, were followed by individual metabolite analyses to identify potential drivers. Significant metabolites were tested for replication in a comparable real-world observational HEAL-MS cohort with longitudinal MRI data. Lastly, we tested whether ibudilast treatment was associated with metabolite trajectories and performed metabolite set enrichment analysis. Findings: Higher baseline levels of glycerophospholipids were associated with slower decline in both BPF and WMF, and sphingomyelins were similarly associated with slower BPF decline. For example, higher 1-palmityl-2-stearoyl-GPC (O-16:0/18:0) levels were associated with slower BPF decline in SPRINT-MS (beta = 0.016 [95% CI: 0.008, 0.024]; p = 4.35 x 10^-5) and replicated in HEAL-MS (beta = 0.108 [95% CI: 0.006, 0.211]; p = 3.90 x 10^-2). Metabolites associated with GMF preservation were enriched in androgenic steroids and steroid sulfates, with consistent positive associations observed in the replication cohort, whereas metabolites inversely associated with CTH were predominantly xenobiotic-related. Ibudilast treatment was associated with increased sphingomyelin species, such as palmitoyl sphingomyelin (d18:1/16:0; beta = 0.185 [95% CI: 0.085, 0.286]; FDR = 1.79 x 10^-2), and decreased levels of amino acid-related metabolites, such as anthranilate (beta = -0.270 [95% CI: -0.403, -0.137]; FDR = 3.87 x 10^-2). Pathway-based analyses corroborated these findings, highlighting glycerophospholipid and sphingolipid metabolism as key pathways implicated in brain atrophy in MS. Interpretation: Distinct lipid subsets were associated with slower brain atrophy in people with MS, and ibudilast treatment was associated with metabolite alterations in potentially neuroprotective directions. Metabolomics may provide prognostic and pharmacodynamic biomarkers for progressive MS.

Background. Conventional ICU severity scores - SOFA, qSOFA, and APACHE-II - use additive integer weightings that cannot capture non-linear organ failure interactions; prospective validations consistently report AUC near 0.73. None quantifies prediction uncertainty, evaluates demographic equity, or acknowledges that their key biomarkers (albumin, creatinine, BUN, lactate, GCS) are also primary confounders of emerging Alzheimer's disease (AD) blood biomarkers p-tau217 and neurofilament light chain (NfL). Methods. Fourteen classifiers were trained on a SOFA-calibrated synthetic ICU cohort (N = 90,000; mortality 29.2%), including an FT-Transformer, XGBoost, and LightGBM tuned by Bayesian optimisation. Seven composite features were engineered from clinical first principles; the novel lactate/albumin ratio (rLA) mirrors the albumin-adjusted p-tau217 correction formula. Post-hoc analyses included nine-method aggregated permutation importance, Monte Carlo Dropout uncertainty decomposition (T = 50), distribution-free conformal prediction, a three-zone triage system, formal ablation, survival analysis, temporal deployment validation, and demographic fairness evaluation. Results. On a natural-distribution held-out cohort (n = 18,000; mortality 29.3%), XGBoost achieved AUC = 0.967 (95% CI 0.965-0.970), surpassing SOFA (AUC = 0.731) by +0.236 (DeLong z = 55.8, p < 0.001; NRI = +0.740). Selective prediction raised FT-Transformer AUC from 0.917 to 0.980 at 50% abstention. Removing neurodegeneration-proxy features reduced AUC by 9.51 percentage points. ML probability was the sole significant covariate in adjusted Cox regression (HR = 6.19, p < 0.001); SOFA, age, lactate, and albumin were non-significant. Temporal AUC range was 0.003 across four deployment windows; sex and age AUC gaps were 0.005 each. Conclusions. This framework delivers well-calibrated, uncertainty-aware ICU mortality prediction with formal coverage guarantees and demographic equity. Ablation-confirmed contributions of neurodegeneration-proxy features, with PDP inflection points aligning with established clinical thresholds, provide a hypothesis-generating quantitative link between routine ICU biomarkers and the AD neurodegeneration pathway warranting prospective validation.

Objective: To determine whether absolute ictal energy on intracranial EEG identifies brain regions whose epileptogenic involvement is attenuated under existing baseline-normalized, dynamic-systems, and event-based frameworks. Approach: Intracranial EEG from 56 patients (five centers; 21 SEEG, 35 ECoG) was analyzed using the Teager-Kaiser Energy Operator computed as z-scored and raw envelopes; energy-dominant network regions (EDNRs) were defined as electrodes whose raw-energy rank exceeded their z-score rank by at least 2 positions. Hilbert decomposition characterized instantaneous amplitude and frequency. Main results: EDNRs were identified in 51 of 56 patients (91%; mean 3.4). Hilbert decomposition revealed elevated baseline amplitude in EDNRs relative to both non-involved regions (p < 0.001) and potential seizure onset zones (PSOZs, the top-ranked electrodes under both metrics; p = 0.029), with EDNRs participating in seizure-frequency dynamics comparable to PSOZs (mean ictal frequency shift +3.7 versus +4.1 Hz). EDNR detectability correlated directly with electrode count (Spearman r = 0.899, p < 0.001) without plateau. Significance: Absolute ictal energy identifies an epileptogenic network component with elevated baseline amplitude attenuated under baseline-normalized metrics. The dual-metric framework defines a complementary energy-based axis and establishes the second layer of a two-layer approach with seizure onset and propagation mapping as the first layer. EDNR detectability scales with electrode count, directly relevant to SEEG implantation strategy and to network-level inferences from heterogeneously covered cohorts.

KCNT1-related disorders represent clinically heterogeneous severe epilepsies associated with profound neurodevelopmental impairment. The full phenotypic spectrum and longitudinal disease trajectory remain incompletely characterized, which is a critical gap limiting the establishment of quantifiable endpoints necessary for future clinical trials. Compounding this challenge, identical pathogenic variants result in phenotypically distinct syndromes, including early infantile developmental and epileptic encephalopathy (EIDEE) and autosomal dominant sleep-related hypermotor epilepsy (ADSHE), underscoring unresolved genotype-phenotype relationships. To address these gaps, we performed a comprehensive analysis of 159 individuals with KCNT1-related disorders, including a longitudinally characterized subgroup of 62 individuals across 390 patient years, systematically defining disease progression, seizure trajectories, developmental outcomes, and treatment response across the full spectrum of the disorder. Seizures were nearly universal, affecting 157 of 159 individuals, with 81% (n=126/156) having seizure onset within the first year of life. Stratification by clinical subgroup revealed divergent seizure onset patterns. Recurrent variants did not significantly differ in age of seizure onset yet exhibited variant-specific clinical fingerprints, such as the preponderance of focal clonic seizures (OR=5.03, 95% CI 1.60-15.7, f=0.47) in those with the p.Gly288Ser variant. Comparison with a broader cohort of 14,893 individuals with neurodevelopmental disorders revealed phenotypic features such as migrating focal seizures (OR=21716, 95% CI 2409-Inf, f=0.42) and hypertonia (OR=26.5, 95% CI 18.2-38.3, f=0.45) to be more common in EIDEE, and nocturnal seizures (OR=29787, 95% CI 3062-Inf, f=0.5) and hyperactivity (OR=13.7, 95% CI 4.70-35.9, f=0.32) to be more common in ADSHE. These findings corroborate and extend those reported in the existing literature. Developmental milestones revealed marked delays across all domains. Analysis of longitudinal medication prescription patterns exposed striking therapeutic variability, reflecting the absence of a consistent treatment framework. Several anti-seizure medications frequently cited as beneficial, quinidine and cannabidiol, were not associated with seizure improvement or sustained seizure freedom in our cohort. In contrast, clobazam (OR=1.39, 95% CI 1.12-1.72, f=0.85), ketogenic diet (OR=1.30, 95% CI 1.07-1.57, f=0.75), and lacosamide (OR=2.03, 95% CI 1.54-2.66, f=0.59) demonstrated positive comparative effectiveness. Quantitative EEG analysis distinguished individuals with KCNT1-related disorders from age-matched controls with high accuracy (AUC=0.906), with key discriminating spectral features, including alpha power in the central and parietal regions, demonstrating significant reduction across childhood and adolescence. Collectively, these findings expand the phenotypic and genotypic landscape of KCNT1-related disorders through large-scale real-world clinical data, establish quantifiable longitudinal clinical endpoints, and provide actionable insights into genotype-phenotype relationships and differential treatment response. Together, these findings will help identify outcome measures and biomarkers to inform future clinical trial design.

Abstract Background Prediction of long-term neurodevelopmental outcomes remains challenging after perinatal asphyxia. Here, we studied whether computational metrics of brain function derived from neonatal EEG are associated with long-term neurodevelopment in infants with perinatal asphyxia. Methods Total of 36 term-born infants with perinatal asphyxia with or without hypoxic-ischemic encephalopathy were studied with neonatal multichannel electroencephalography (EEG). We computed local EEG amplitudes and phase-amplitude coupling (PAC), as well as large-scale functional cortical networks estimated using amplitude-amplitude correlations (AAC) and phase-phase correlations (PPC). These EEG-derived markers were tested for associations with neurodevelopmental outcomes at two years, assessed using the Griffiths Scales of Child Development, 3rd edition (GMDS-III). Results EEG amplitudes showed positive associations with GMDS-III Foundations of Learning and General Development scores across most electrodes during quiet sleep, with the strongest effects observed at frontal and central regions (r = 0.44-0.66). PAC showed negative associations with the same scores mainly over parietal and temporal regions (r = -0.45 to -0.55). Cortical AAC networks demonstrated the most robust and widespread negative associations in all frequency bands during quiet sleep (r = -0.47 to -0.54), with 70-72% of connections significant in high delta frequency. In turn, PPC networks showed frequency-selective and more spatially constrained negative associations during quiet sleep (r = -0.48 to -0.53), involving 5-12% of the network. Conclusions Both local and network-based metrics in the newborn brain show significant association with neurodevelopmental outcome at 2 years after perinatal asphyxia.

Introduction: Variants in the polymerase gamma (POLG) gene are associated with a wide range of mitochondrial disorders. Emerging evidence suggests a potential link between POLG variants and Parkinson's disease (PD); yet, results remain inconclusive. Objectives: To investigate the genetic spectrum and prevalence of POLG variants in PD across diverse ancestries. Methods: We leveraged multi-ancestry genetic data from the Global Parkinson's Genetics Program (GP2), including genotyping data from 98,589 and short-read sequencing data from 36,022 individuals. We performed a POLG rare variant screen, case-control association, and gene-level burden analyses. Results: Five PD cases carried potentially biallelic rare pathogenic/likely pathogenic POLG variants. Additionally, 228 individuals (<1%; 161 PD cases, 28 individuals with other neurological disorders, and 39 controls) carried 34 distinct rare pathogenic/likely pathogenic heterozygous variants, with no significant frequency differences between cases and controls, except for the p.Ala467Thr variant in the European population. The co-inherited pathogenic variants p.Thr251Ile and p.Pro587Leu were present in <1% of both cases and controls, with no significant group differences. Burden and variant-level association analyses showed no association between rare POLG variant burden or common POLG variant enrichment and PD. Conclusions: POLG variants are overall rare in PD. The identification of rare pathogenic variants among PD cases suggests that POLG-related mitochondrial dysfunction may contribute to PD in isolated instances, particularly under recessive inheritance. Our findings support a role for POLG variants in select cases and underscore the need for larger-scale sequencing and functional studies.

Background and Purpose: Drug resistant epilepsy (DRE) affects approximately 15 million people worldwide, and surgery remains the only curative option. A key challenge in predicting outcomes is the lack of standardized, quantitative tools to help distinguish seizure driver regions from responder regions during stereoelectroencephalography (sEEG) recordings. We validated the CN Suite, a computational platform that uses causal network mapping and machine learning to assign criticality scores to sEEG contacts, testing whether higher scores correspond to surgically treated tissue in patients with favorable outcomes. Methods: We analyzed deidentified clinical data from 60 patients (aged 2 years and older) with focal or multifocal DRE who underwent sEEG monitoring and proceed to surgery at four U.S. Level 4 epilepsy centers. The algorithm was trained on an independent cohort (N=37) and locked prior to validation. The primary outcome was the standardized effect size (Cohens d) of the patient level surgical zone enrichment ratio between more favorable (Engel I or II) and less favorable (Engel III or IV) outcome groups. Contact level sensitivity, specificity, PPV, and NPV were evaluated at a prelocked threshold. Results: The findings support our hypothesis: the algorithm results showed significantly higher criticality values for surgically treated tissue in favorable outcome patients (d=0.74, 95% CI: 0.39 to 1.06, p=0.003). Three potentially clinically actionable findings emerged. First, high-criticality contacts formed spatially compact clusters (~9 mm nearest-neighbor distance vs. 17mm expected by chance), consistent with focal targets amenable to minimally invasive ablation. Second, sensitivity was highest in small focal procedures (80% at 10 or fewer treated contacts) and decreased with resection size. Third, in patients whose surgery failed, high-critical tissue remained outside the resection boundary, suggesting incomplete treatment coverage of the epileptogenic zone rather than mislocalization. Prediction specificity was 84% at the contact level. For adult and pediatric cases (n=28), 88% of contacts that were identified as seizure free were in fact seizure free. Conclusions: Causal network mapping of sEEG identifies compact epileptogenic targets that correspond to surgically treated tissue in patients with more favorable outcomes. CN-Suite performed best in focal procedures and may be best suited for LITT and other minimally invasive approaches. In addition, low-criticality regions were infrequently associated with seizure-generating tissue, particularly in the pediatric cohort although our sample size was small. When surgery failed, residual high-critical tissue outside the resection boundary offered both a mechanistic explanation for less favorable surgical outcome as well as a potential target for reoperation.

Objective: Cognitive impairment is common among older adults with epilepsy, although efficient screening tools suitable for routine use are lacking. Here we examine, for the first time, the utility of the Alzheimers Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) as a screening tool to identify cognitive impairment in older adults with epilepsy. Methods: Participants included 83 adults (ages over 55) with epilepsy from the Brain, Aging, and Cognition in Epilepsy (BrACE) study and 83 age-, sex-, and education-matched cognitively healthy controls from the Alzheimers Disease Neuroimaging Initiative (ADNI-3). All completed the ADAS-Cog and a comprehensive neuropsychological battery to identify cognitive phenotypes (intact vs impaired). Performance on individual ADAS-Cog items and the total score was assessed, and diagnostic efficiency statistics were determined. Results: Epilepsy participants (mean age=66.4 years) performed significantly worse across the ADAS-Cog total score and 8 of the 13 individual test items compared to controls. The largest effect sizes were observed on verbal learning and memory tasks, particularly word recall (d=0.87) and delayed word recall (d=1.06). An ADAS-Cog total score of at or exceeding 15 yielded optimal diagnostic efficiency (67.5% accuracy, 68.8% sensitivity, 66.7% specificity) for identifying cognitive impairment. Significance: The ADAS-Cog is sensitive to detecting cognitive impairment in older adults with epilepsy and may represent a scalable screening option in this population. Additional comparative studies in older epilepsy populations are needed to determine the sensitivity of this measure to longitudinal change, cross-cultural applicability, and availability across languages. Plain language summary: Cognitive decline is common among older adults with epilepsy, although sufficient evidence supporting the use of screening tools to identify cognitive impairment in this population is lacking. The ADAS-Cog may be a useful screening option in epilepsy research and clinical care, although additional studies are needed to compare it with other cognitive screening tests and to confirm its applicability for clinical care and across cultures and healthcare settings.

BackgroundDischarge destination after acute ischemic stroke has implications for functional recovery and healthcare costs. Individuals discharged to inpatient rehabilitation facilities (IRFs) achieve better outcomes than those discharged to skilled nursing facilities (SNFs); however, many patients discharged to IRFs and SNFs have similar clinical profiles. We examined non-clinical factors associated with discharge location after acute ischemic stroke. MethodsPopulation: 236 adults hospitalized with acute ischemic stroke, living independently in the community prior to admission, and discharged to either an IRF (n=171) or SNF (n=65). Clinical variables: NIHSS, Charlson Comorbidity Index (CCI), acute care length of stay (LOS), functional status (AM-PAC "6-Clicks"), and neglect. Non-clinical variables: age, sex, race, marital status, insurance, home layout, living status, and available assistance. Associations with discharge location were evaluated using univariable and multivariable logistic regression and reported as odds ratios (OR) with 95% confidence intervals (CI). ResultsIndividuals discharged to IRFs were younger, more likely to cohabitate, and had shorter LOS than those discharged to SNFs. Functional status (AM-PAC) and comorbidity burden (CCI) did not differ significantly between groups despite differences in discharge destination. In univariable models, younger age, cohabitating marital status, living with family, available assistance, shorter LOS, private insurance, and higher NIHSS were associated with greater odds of IRF discharge. In multivariable analysis, younger age (OR 0.94, 95% CI 0.91-0.98), cohabitating marital status (OR 2.46, 95% CI 1.13-5.48), and shorter LOS (OR 0.88, 95% CI 0.82-0.93) remained independently associated with IRF discharge. ConclusionsIndividuals with comparable pre-stroke independence and similar clinical severity, discharge to IRF versus SNF was independently associated with non-clinical factors; age, marital status, and LOS, whereas stroke severity and functional status were not significant predictors. These findings underscore the importance of evidence-informed discharge criteria integrating clinical indicators and social context to support equitable access to intensive rehabilitation after stroke.

Background: Sleep fragmentation and reduced sleep efficiency are markers of disrupted sleep architecture linked to cognitive and age-related decline. Current assessments rely on subjective reports prone to recall bias, limiting their effectiveness for longitudinal monitoring. Data-driven analysis of sleep using physiological signals such as EEG and EMG remains underutilised, particularly in mid-to-older adults. Objective: We present a deep learning pipeline for automated sleep staging and label-free abnormality scoring, with the primary objective of quantifying deviations in sleep architecture to capture progressive sleep disruption and longitudinal change. Methods: Temporal and attention-based models were benchmarked using datasets from the National Sleep Research Resource and PhysioBank. To improve class-specific performance, we introduce a stacking-based ensemble of sleep stage classifiers, each trained to specialise in a different stage. For longitudinal scoring, we develop a reconstruction loss-based abnormality metric using a temporal convolutional autoencoder trained on hypnograms generated by the sleep staging models. Results: Attention-based models, particularly AttnSleep, achieved the highest performance in both multimodal and single-channel settings (accuracy: 0.85 and 0.83; F1: 0.79 and 0.74, respectively). The encoder-decoder ensemble model improved overall classification accuracy by 3% compared to the best-performing biased base classifier, with a modest gain in N1-stage F1 score (0.444). The proposed abnormality score correlated with Pittsburgh Sleep Quality Index components and showed sensitivity to synthetic hypnogram degradation, highlighting its potential as a label-free indicator of sleep disruption. Conclusion: Automated classification and annotation-free scoring enable an end-to-end multimodal pipeline that supports scalable, objective sleep health monitoring, with relevance for future clinical deployment.

Mild malformation of cortical development with oligodendroglial hyperplasia and epilepsy (MOGHE) is a recently recognized cause of drug-resistant focal epilepsy. It is often MRI-negative or shows imaging features mimicking focal cortical dysplasias, which makes recognition difficult and limits presurgical counseling. We aimed to identify an intracranial EEG (iEEG) biomarker that distinguishes MOGHE from other developmental brain lesions encountered in epilepsy surgery. In a retrospective multicenter test cohort of 38 patients (18 MOGHE, 20 non-MOGHE), we analyzed long-term stereo-EEG and subdural recordings. Only MOGHE patients showed highly stereotyped clusters of very brief low-voltage fast activity (LVFA) events, organized into status-like 3 to 12-minute episodes that often lacked clear clinical symptoms. LVFA clusters were present in 16/18 MOGHE and 0/22 non-MOGHE patients. We then tested diagnostic performance in an independent, blinded single-center validation cohort of 22 patients (11 MOGHE, 11 non-MOGHE), in which visual identification of LVFA clusters correctly classified 10/11 MOGHE and 10/11 non-MOGHE cases (Cohens kappa=0.82). Penalized logistic regression further confirmed MOGHE histology as the strongest predictor of LVFA clusters, independent of age and lobe localization. Because LVFA clusters can be recognized visually on routine intracranial EEG recordings without specialized software, this biomarker is readily applicable in clinical practice and may improve presurgical identification of MOGHE. Future prospective studies should determine whether its recognition influences surgical planning, improves outcome prediction, or facilitates selection of patients for mechanism-based therapies.

Lesion anatomy has been widely used to study post stroke cognitive outcomes, but it is unclear whether lesion-based measures provide clinically meaningful prognostic information beyond established predictors. Stroke survivors (n = 408) completed the Oxford Cognitive Screen (OCS) during acute hospitalisation and at chronic (6-month) follow-up. Lesion characteristics and structural disconnection profiles associated with chronic OCS scores were identified using ROI-level, voxel-level and structural network disconnection lesion mapping approaches. The incremental predictive value of these measures, relative to acute behaviour and pre-morbid brain health, was evaluated using regression analyses, receiver operating curve (ROC) and support vector regression (SVR) models predicting continuous chronic scores. Significant lesion and disconnection correlates of chronic cognitive impairment were identified for 9/10 OCS subtests. The extent of damage to these correlates was significantly associated with chronic cognitive scores, but their diagnostic utility for identifying persistent impairment was low under conventional thresholds (AUC mean = 0.59, range= 0.46-0.66). Acute cognitive task performance was the single best predictor of chronic cognition (AUC mean = 0.66, range = 0.4-0.95). In multivariate analyses, SVR models trained on acute cognitive performance and regional atrophy severity scores both outperformed models trained on lesion anatomy or structural disconnection across most cognitive domains. SVR models combining anatomical, disconnection and behavioural predictors did not improve predictions accuracy relative to behaviour or atrophy-only models. Together, these findings demonstrate that statistically significant lesion-outcome relationships do not necessarily translate into clinically useful prognostic indicators. In a large, clinically representative stroke cohort, detailed lesion-based measures provided limited incremental prognostic value beyond acute cognitive assessment and coarse brain health markers. These results highlight the importance of explicitly evaluating predictive utility when developing prognostic models for post-stroke cognitive outcomes.

Objective: To systematically evaluate pathway-informed polygenic risk score (PRS) strategies and determine which approaches most effectively leverage biological annotations for risk prediction, using brain amyloid-beta positivity as a case study. Methods: We systematically benchmarked approaches for integrating pathway information into PRS construction to predict brain A{beta} positivity. Using two cohorts, the Alzheimer's Disease Neuroimaging Initiative (ADNI, n = 969) and Australian Imaging, Biomarkers and Lifestyle (AIBL, n = 251), we compared Apolipoprotein E (APOE) genetic risk score (GRS), clumping and thresholding (C+T) PRS, pathway-guided single nucleotide polymorphism (SNP) selection PRS, and pathway-specific PRSs ensembled via machine learning. Pathways were derived from manually curated literature or from pathway databases via Functional Mapping and Annotation (FUMA). Results: In cross-validation on the ADNI cohort, pathway-informed PRS using a narrow-set of pathways to guide SNP selection (PathPRS-SNPLit without APOE locus) significantly outperformed the standard PRS model (median AUC = 0.742, p = 0.006) and the APOE locus model (median AUC = 0.736, p = 5.1 x 10-5) based on the Mann-Whitney U test, achieving a median AUC of 0.763. This model showed enhanced ability to identify subgroups within the 10% lowest- and highest-risk groups compared to the current standard of APOE locus alone (odds ratio = 0.67, 95% CI: 0.56-0.81; and OR = 13.23, 95% CI: 10.23-17.11), highlighting its clinical potential. Using a focused set of literature-curated pathways outperformed using a broader set of database-derived pathways across configurations. When contrasting strategies for aggregating information across pathways, we observed that using pathways to guide selection of SNPs and then building a single PRS performed comparably to building PRS for each pathway and using machine learning (ML) to aggregate these, though the latter enabled pathway-level interpretability. Similar trends were observed in the external AIBL validation dataset. Interpretation: Pathway-informed PRS can meaningfully improve genetic risk enrichment for A{beta} positivity beyond APOE and standard C+T approaches, provided pathway definitions are carefully curated. The choice of pathway source has the strongest impact on predictive performance, with aggregation strategies or ML model choice having far less impact. Our findings highlight the utility of literature-curated, pathway-informed PRSs for A{beta} prediction and offer practical guidance for pathway-informed PRS construction in other polygenic traits.

Background: Gait variability is a hallmark of Parkinson's disease (PD) and has been linked to cognitive deficits and fall risk. Rapid eye movement sleep behavior disorder (RBD) is a strong predictor of synucleinopathies, yet evidence for gait changes in RBD is inconsistent. Performing a dual task increases gait variability, an effect that can be quantified using a cost function. Objective: Determine the degree to which dual task cost differs between control, RBD, and PD participants at baseline, and between RBD converters versus non-converters at follow-up. Methods: 46 RBD, 23 control, and 14 PD participants completed standardized gait analysis at baseline. Parameters chosen for analysis included enhanced gait variability index (eGVI), functional ambulation performance (FAP), velocity, step length, cadence, base of support, and double support time. Medical records were surveilled for 3 years following participant enrollment, determining that 6 RBD participants converted to PD or dementia. Baseline gait indices and dual task costs were compared between control, RBD, and PD groups at enrollment, and between RBD stable and RBD converters at follow-up. Results: The PD group had greater eGVI, as well as greater dual task cost for FAP, cadence, width, and double support time. No differences in gait variability were identified between RBD and control groups at baseline. Compared to the stable group, RBD converters had greater dual task cost for FAP, velocity, cadence, and double support time. Conclusions: Increased gait variability during dual task may identify RBD patients at imminent risk of phenoconversion.

Background: Mono-allelic Dehydrodolichyl Diphosphate Synthase (DHDDS) variants are associated with juvenile Parkinsonism, developmental delay and seizures. Symptoms are progressive, and various mechanisms, such as defective glycosylation, lysosomal dysfunction and cholesterol accumulation have been hypothesized to underlie disease symptoms. There is no treatment for DHDDS-related disease. Methods: Patient-derived cortical forebrain organoids were created to elucidate disease mechanisms and evaluate potential treatments. In these neuronal models, glycosylation, lipidomics, proteomics, cholesterol/ganglioside accumulation, mitochondrial function and electrophysiological activity were assessed. Finally, we investigated the effects of nicotinamide mononucleotide (NMN), identified through a yeast-based drug screen, in neuronal cell models and in six patients in an off-label, N-of-1, observational series. Results: DHDDS-patient derived organoids showed visual signs of degeneration after four months of culturing. This was accompanied by significant cholesterol accumulation in astrocytes, decreased mitochondrial respiration and loss of deep-layer neurons. In addition, we identified glycosylation abnormalities, showing for the first time that glycosylation in human tissue is affected by monoallelic DHDDS variants. Proteomic analysis revealed altered protein expression of proteins involved in lipid metabolism, cytoskeletal organization and neuronal development. We found that oral Nicotinamide Mononucleotide supplementation led to significant improvement in mitochondrial respiration and electrophysiological parameters in organoids, concurring with clinical improvements in all of the treated patients, particularly regarding their ataxia and tremor. Conclusion: Our findings reveal a progressive phenotype in DHDDS-patient-derived brain organoids, with mitochondrial dysfunction and astrocyte-specific metabolic alterations contributing to disease pathology. Notably, NMN treatment led to clinical improvements in patients with heterozygous DHDDS variants, highlighting its potential as a therapeutic strategy.