Annals of Clinical and Translational Neurology

Background Friedreich ataxia (FRDA) is a rare neurodegenerative disorder with substantial heterogeneity in clinical presentation and progression, complicating prognosis and trial design. Neuroimaging offers objective biomarkers to track disease evolution, yet variability in progression patterns remains poorly understood. Objective To identify biologically meaningful FRDA progression subtypes using longitudinal multimodal MRI and assess their associations with demographic, genetic, and clinical factors. Methods Longitudinal structural and diffusion MRI data from 54 FRDA and 57 controls were analysed. Annualised progression rates of macrostructural (volumetric) and microstructural (diffusion) features across cerebellum, brainstem, and spinal cord regions were clustered using Gaussian Mixture Models. Cluster robustness was assessed using per-cluster Jaccard similarity and other validation metrics. Random Forest classification examined predictors of cluster membership. Results Three reproducible clusters/subtypes emerged: micro-dominant/dual progression, characterised by widespread microstructural deterioration with modest volumetric decline; macro-dominant, marked by pronounced volumetric decline with minimal microstructural change; and minimal/no progression, showing negligible change in all measures. FRDA participants predominated in the first two clusters. Random Forest prediction of cluster membership using clinical and demographic variables identified length of the trinucleotide repeat expansion in the FXN gene as key predictor. Conclusions Data-driven clustering of longitudinal MRI identified distinct FRDA subtypes with unique co-progression patterns, underscoring genetic burden as a key driver. Recognising such heterogeneity can improve patient stratification, enable personalised monitoring, and guide targeted therapeutic strategies. Future studies should validate these subtypes in larger, more diverse cohorts and integrate additional biomarkers for enhanced precision.

BackgroundChronic relapsing inflammatory optic neuropathy (CRION) is a steroid-dependent form of optic neuritis with incompletely understood pathophysiology. The identification of myelin oligodendrocyte glycoprotein antibodies (MOG-IgG) in a substantial patient subset has challenged the diagnostic and therapeutic management. The aim of this study was to investigate clinical profiles and treatment outcomes of patients with CRION, comparing MOG-IgG-positive (MOG+) and seronegative (MOG-) subgroups. MethodsPatients from six European tertiary centers fulfilling diagnostic criteria for CRION were included. All underwent cell-based autoantibody testing. Clinical outcomes (visual acuity, annualized relapse rate), laboratory and imaging findings (MRI, OCT), and treatment responses were retrospectively analyzed. ResultsSixty patients were included (median age 33 years; 70% female); 27 (45%) were MOG+. MOG+ CRION was associated with later onset, higher ARR before treatment (median [IQR] 2 [1-3] vs. 1 [1-2], p = 0.023), and a trend toward shorter inter-relapse intervals. Additional distinguishing features included higher frequencies of antinuclear antibody positivity, elevated CSF interleukin-6, and extensive optic neuritis on MRI. Relapse burden correlated with visual acuity decline and retinal thinning. In MOG+ patients, monoclonal antibody therapy reduced the ARR (n = 21; 2 [1-3] vs. 0 [0-2], p = 0.024), primarily driven by tocilizumab (n = 11; 2 [1-3] vs. 0 [0-1], p = 0.023). In MOG-patients, rituximab and azathioprine showed a trend toward ARR reduction. ConclusionCRION represents a heterogeneous syndrome encompassing distinct subgroups. MOG+ patients demonstrate higher disease activity but respond favorably to tocilizumab. Serological testing is critical for treatment stratification and preventing relapses.

BackgroundAlthough neurogenic orthostatic hypotension (nOH) is a common and debilitating feature of multiple system atrophy (MSA), little is known about the burden of symptoms in the real world. ObjectivesTo design and conduct a cross-sectional community-based research survey targeting patients with MSA, with and without nOH. MethodsWe recruited patients with MSA to complete an anonymous online survey covering three core themes: 1) timely diagnosis, 2) nOH pharmacotherapy and refractory symptoms, and 3) confidence in physician knowledge. Responses were grouped by pre-specified diagnostic certainty levels. Relationships between symptoms, function, and pharmacotherapy were assessed using univariate and multivariate methods. ResultsWe analyzed 259 respondents with a self-reported diagnosis of MSA (age: M=64.38, SD=8.09 years; 44% female). In total, 42% also had a diagnosis nOH; 40% had symptoms highly suspicious of nOH, but no diagnosis; and 21% reported having never had their blood pressure measured in the standing position at a clinical visit. Treatment with a pressor agent was independently associated with the presence of other symptoms of autonomic failure. Each additional nOH symptom reported increased the odds of requiring pharmacotherapy by 18%. Yet, despite anti-hypotensive medication use, 97% of patients reported limitations in their ability to bathe, cook, or arise from a chair/bed with 76% needing caregiver support for refractory nOH symptoms. ConclusionsThis cross-sectional representative sample shows nOH is underrecognized and undertreated in MSA patients, leading to substantial functional limitations. It is our hope that these findings are leveraged for planning future trials and advocating for better treatments.

Objective: Spinocerebellar ataxia type 1 (SCA1) is a rare, inherited neurodegenerative disease characterised by progressive deterioration of motor and cognitive function. Here, we illustrate the pattern and evolution of brain atrophy in people with SCA1 using a large multisite dataset. Methods: Structural magnetic resonance imaging data from SCA1 (n=152) and healthy control (n=131) participants from seven sites and two consortia were analyzed using voxel-based morphometry. Cross-sectional stratification and correlations were undertaken with ataxia severity and duration to profile disease evolution. Cerebrocerebellar structural covariance analysis was used to understand the relationship between cerebral and cerebellar tissue atrophy. Results: Atrophy in SCA1 first manifests in the lower brainstem and cerebellar white matter (WM), before progressing to the pons, anterior cerebellum, and cerebellar lobule IX. The midbrain and peri-thalamic WM and the remainder of the cerebellar cortex are then affected, with preferential involvement of specific motor and cognitive areas. Finally, degeneration in the striatum and cerebral WM corresponding to the corticospinal tract become apparent. Atrophy and correlations with ataxia severity are most pronounced in the cerebellar WM and pons. Structural covariance analysis showed reduced correlations between cerebellar and cerebral WM volume in SCA1 participants. Interpretation: Cross-sectional stratification of a large SCA1 cohort by ataxia severity indicates a pattern of atrophy spread across the brainstem, cerebellum, and subcortical grey and white matter. Ongoing volume loss throughout the disease course is most evident in a core set of infra-tentorial brain regions. Atrophy of cerebellum spans both motor and cognitive functional zones. Cerebellar degeneration is not directly mirrored by downstream effects in the cerebrum.

ObjectiveTo describe the design, feasibility, and baseline characteristics of the Migraine Impact on Neurocognitive Dynamics (MIND) study, a 30-day smartphone-based cohort for high-frequency assessment of cognition and symptoms in adults with migraine. BackgroundCognitive symptoms are an important component of migraine burden, but they are difficult to measure using single-visit testing or retrospective questionnaires. Repeated smartphone-based assessment may better capture real-world variability in cognition and symptoms. MethodsAdults meeting International Classification of Headache Disorders, 3rd edition, criteria for migraine were enrolled remotely and completed 30 days of once-daily ecological momentary assessments and mobile cognitive tasks delivered through the Mobile Monitoring of Cognitive Change platform. Baseline measures assessed demographics, migraine characteristics, disability, mood, stress, and treatment patterns. Feasibility was evaluated using enrollment, completion, and retention metrics. ResultsA total of 177 participants enrolled (mean age 38.8 {+/-} 11.9 years; 79.7% female), including 80/177 (45.2%) with chronic migraine. Across the 30-day protocol, 3688 daily assessments were completed, representing 70.8% of all possible study days, and 70.6% of participants completed at least 20 days of monitoring. Completion remained above 60% across study days. At baseline, chronic migraine was associated with greater burden than low-frequency and high-frequency episodic migraine, including higher MIDAS scores (98.6 vs. 38.7 and 70.3), more days with concentration difficulty (16.0 vs. 7.9 and 11.5), and more days with functional interference (18.5 vs. 7.6 and 13.0). ConclusionsThe MIND study demonstrates the feasibility of high-frequency smartphone-based assessment of cognition and symptoms in migraine and provides a methodological foundation for future analyses of within-person cognitive and symptom dynamics across the migraine cycle.

ImportanceMigraine attacks often occur unpredictably, limiting the ability of individuals to initiate timely preventive or preemptive treatment. Short-term probabilistic forecasting of migraine risk could enable more targeted management strategies. ObjectiveTo externally validate the previously developed Headache Prediction Model (HAPRED-I), evaluate an updated continuously learning model (HAPRED-II), and assess the feasibility and short-term safety of delivering individualized probabilistic migraine forecasts directly to patients. Design, Setting, and ParticipantsProspective 8-week cohort study conducted remotely at two academic medical centers in the United States (Massachusetts General Hospital and Wake Forest Health Sciences) between 2015 and 2019. Adults with recurrent migraine or tension-type headache completed twice-daily electronic diaries. A total of 230 participants contributed 23,335 diary entries across 11,862 participant-days of observation. Main Outcomes and MeasuresOccurrence of a headache attack within 24 hours following each evening diary entry. Model performance was evaluated using discrimination (area under the receiver operating characteristic curve [AUC]) and calibration. ResultsExternal validation of HAPRED-I demonstrated modest discrimination (AUC, 0.59; 95% CI, 0.57-0.61) and poor calibration, with predicted probabilities consistently exceeding observed headache risk. In contrast, the continuously updating HAPRED-II model demonstrated progressive improvement in predictive performance as participant-specific data accumulated. Discrimination increased from an AUC of 0.59 (95% CI, 0.57-0.61) during the first 14 days to 0.66 (95% CI, 0.63-0.70) after the first month, accompanied by improved calibration across predicted risk levels. Over the study period, 6999 individualized forecasts were delivered directly to participants. No evidence suggested that receipt of forecasts was associated with increasing headache frequency or worsening predicted headache risk trajectories. Conclusions and RelevanceA static migraine forecasting model demonstrated limited transportability to new individuals. In contrast, models that continuously update within individuals may improve predictive accuracy over time and enable real-time delivery of personalized migraine risk forecasts. Further work incorporating richer physiologic and contextual predictors will likely be necessary before such systems can reliably guide clinical treatment decisions.

Motor neuron (MN) loss is a hallmark of neurodegenerative disorders, yet its assessment remains variable, confounding mechanistic and therapeutic interpretation. To address this, we conducted a systematic review and meta-analysis of spinal muscular atrophy (SMA) mouse studies, revealing 60% variability in reported MN loss, largely attributable to nonspecific spinal cord sampling. Using a whole-segment approach with tissue clearing, MN tracing, and multimodal imaging, we confirmed segment-dependent differences in MN counts. Common MN markers (SMI-32, Nissl) lacked specificity, whereas choline acetyltransferase (ChAT) provided robust labeling in murine and human spinal cords. Deep learning-based whole-mount segmentation enabled unbiased MN quantification and validated manual counts. Integrating analysis with computational modeling established segment sampling as a key driver of variability and revealed degeneration patterns: widespread MN loss in amyotrophic lateral sclerosis (ALS), selective MN loss in severe SMA, and preservation in mild SMA models. These findings establish a framework for reproducible MN quantification. HighlightsO_LISpinal cord segment-specific analysis reduces variability and allows accurate MN quantification C_LIO_LIChAT is the most reliable MN marker in murine and human spinal cords C_LIO_LIDeep learning-based segmentation enables unbiased MN quantification in intact spinal cords C_LIO_LIMN degeneration is widespread in ALS but restricted to pools innervating proximal muscles in severe SMA C_LI

ObjectiveIn individuals with drug-resistant epilepsy, accurately identifying the brain regions where seizures originate is a critical prerequisite to guide surgical treatment and achieve seizure freedom. To accomplish this, intracranial EEG is considered the gold standard, providing the spatiotemporal high-resolution data necessary to pinpoint epileptogenic activity. However, this precision is achieved through an invasive procedure with significant patient burden, which is fundamentally limited by the electrode placement and spatial coverage. MethodsIn this study, we investigated the potential utility of preoperative resting-state fMRI to non-invasively map alterations in brain dynamics at the whole brain level. Region-wise brain dynamics were quantified with complementary measures of local autocorrelation decay rates. We assessed the capacity of these derived features to effectively identify intracranial EEG confirmed seizure onset zones in 18 individuals with drug-resistant medial temporal lobe epilepsy. Overall, the study cohort contained 3867 implanted electrodes of which 159 classified as seizure onset zones by two independent board-certified epileptologists. ResultsOverall, our findings reveal more constrained temporal dynamics for brain regions associated with seizure onsets compared to non-seizure onset zones. Individual-level prediction showed a performance better than chance in 15 of the 18 patients. The overall predictive performance across all patients yielded a median AUC of 0.81, a median true positive rate of 0.75, and a median true negative rate of 0.83. Furthermore, in a subset of 13 patients, those with negative seizure outcomes showed higher probabilities of seizure onset zone predictions outside the resection area compared to those with good outcomes. SignificanceOverall, our findings suggest that altered temporal dynamics derived from preoperative resting-state fMRI represent a promising non-invasive approach for delineating epileptogenic tissue, potentially informing intervention strategies and guiding electrode placement.

BackgroundAccelerated long-term forgetting (ALF), defined as an increased rate of memory loss over extended intervals, has so far been detected in a pilot study of patients with mild multiple sclerosis (MS). This study aimed to (I) confirm the presence of ALF in a larger, heterogeneous MS sample, (II) explore associations with patient-reported outcomes, and (III) assess the diagnostic performance of ALF tests for subjective memory impairment. MethodsThis study compared 62 MS patients and 65 age-, sex-, and education-matched healthy controls using standardized memory tests (RAVLT, WMS-IV Logical Memory subtest). Recall was assessed immediately, after 30 minutes, and after 7 days. Seven-day/30-minute recall ratios (QRAVLT, QWMS) served as primary outcomes. Self-report measures included memory complaints, fatigue, depression, and sleep disturbances. Linear regression and Receiver operating characteristic (ROC) analyses assessed predictors and diagnostic accuracy. ResultsALF was observed in multiple sclerosis since QRAVLT was lower in patients than in controls (0.64 [95% CI 0.59-0.69] vs. 0.78 [0.73-0.82], p < 0.001), as was QWMS (0.79 [95% CI 0.74-0.84] vs. 0.95 [0.90-1.00], p < 0.001), despite comparable initial learning. Greater fatigue, higher memory complaints, longer disease duration, older age, and greater disability were associated with lower ALF scores. The combined ALF score moderately discriminated subjective memory impairment (AUC 0.74; sensitivity 0.73; specificity 0.73). ConclusionMS patients showed ALF despite normal initial learning, indicating a specific memory deficit undetected by standard tests. Long-delay recall using RAVLT and WMS-IV Logical Memory subtest may improve cognitive impairment detection in MS.

Focal cortical dysplasias (FCDs) are one of the most common structural causes of drug-resistant epilepsy in children but are frequently subtle and difficult to detect on conventional MRI. Many automated lesion detection methods have therefore been proposed to support neuroradiological assessment. In this study, we externally validated two recently developed deep-learning approaches for FCD detection, MELD Graph and 3D-nnUNet, in a pediatric cohort. In this retrospective single-center study, brain MRI scans of 71 children evaluated for epilepsy were analyzed, including 35 MRI-positive patients with suspected FCD and 36 MRI-negative cases based on the primary radiology reports. Both models were applied to standard 3D T1-weighted and 3D FLAIR images. Detected lesions were reviewed by an experienced pediatric neuroradiologist and classified as true positive, false positive, or false negative. Clinical semiology and EEG findings were additionally evaluated for cases with false-positive detections. At the lesion level, MELD Graph achieved a precision of 0.85 and recall of 0.52, while 3D-nnUNet achieved a precision of 0.91 and recall of 0.48. In the MRI-negative patients, MELD Graph produced more false-positive detections than 3D-nnUNet (0.53 vs. 0.14 false-positive lesions per patient). At the patient level, MELD Graph showed slightly higher sensitivity than 3D-nnUNet (0.63 vs. 0.54), whereas 3D-nnUNet demonstrated markedly higher specificity (0.86 vs. 0.56). Improved FLAIR image quality was associated with trends toward improved model performance. Both models demonstrated high precision but moderate sensitivity, indicating that they are valuable decision-support tools but cannot replace expert neuroradiological evaluation. Optimized MRI acquisition protocols are needed to further improve automated lesion detection in pediatric epilepsy.

Background: Sleep wake and circadian disturbances are increasingly recognised in people living with amyotrophic lateral sclerosis (plwALS), but endogenous circadian phase timing and its prognostic significance in early disease remain unclear. We assessed whether salivary dim-light melatonin onset (DLMO), an objective marker of central circadian phase, is altered in early plwALS and whether it provides prognostic information. Methods: In this prospective longitudinal observational study, plwALS within 18 months of symptom onset underwent home-based salivary melatonin sampling under dim light conditions at six predefined time points around habitual sleep onset (HSO). Melatonin profiles were modeled using cubic smoothing splines, and DLMO was defined as the first time the fitted curve reached 3 pg/mL. Clinical, respiratory, and sleep assessments were collected at baseline (T0) and after 6 months (T6); a subgroup repeated saliva sampling at T6. Age and sex matched controls underwent melatonin profiling. Associations with disease progression, incident respiratory symptoms, and survival/tracheostomy were examined using regressions and survival analyses. Results: Fifty plwALS were enrolled. Compared with controls, plwALS showed an earlier DLMO (20:24 vs 20:58; p=0.028) despite similar HSO and chronotype. Within ALS cohort, a later baseline DLMO correlated with worse functional/motor status, faster progression of disease, incident dyspnea/orthopnea by T6 (adjusted OR 3.02; p=0.017), and poorer survival/tracheostomy-free outcome. In re-sampled subgroup (n=28), DLMO and other melatonin-derived metrics did not change over 6 months. Conclusions: Circadian phase alterations are detectable in early ALS. Baseline DLMO may represent a non-invasive prognostic biomarker for progression, respiratory symptom emergence and survival, warranting validation in larger multicentre cohorts.

ObjectiveVagus nerve stimulation (VNS) is an established neuromodulation therapy used in the management of drug-resistant epilepsy (DRE), or when other intracranial surgical modalities have not reduced seizure burden. We evaluated whether prior intracranial surgery for epilepsy influences safety and effectiveness outcomes with adjunctive VNS, using real-world data from the CORE-VNS study. MethodsCORE-VNS (NCT03529045), a prospective, multicenter, international observational study, was designed to collect data on seizure and non-seizure outcomes in patients with DRE treated with VNS. Participants were identified as having or not having undergone prior intracranial brain surgery for epilepsy (ICSE) and received an initial VNS implant. Baseline seizure frequency data and patient-reported outcome measures were collected at 3, 6, 12, 24, and 36 months. This analysis compared the baseline data for VNS therapy and safety outcomes at 36 months. ResultsAmong 531 participants implanted with VNS, prior ICSE was performed in 84. Median percentage seizure reductions at 36 months for all seizures (76.6% and 76.3%), all focal seizures (83.3% and 71.8%), and all generalized seizures (77.8% and 76.2%) were found to be similar between those without and with a history of ICSE, respectively. The 50% responder rate for all seizures reported at baseline was similar, 64.8% and 61.8%, in both groups and complete seizure freedom was reported by 17.9% and 8.8%, respectively. Implant-related adverse events (AE) and serious AE rates were similar between groups. ConclusionVNS was associated with clinically meaningful seizure reductions and showed a consistent safety profile irrespective of the history of ICSE. Prior ICSE should not be a contraindication to the consideration of VNS.

1.PurposeSYNGAP1-related developmental and epileptic encephalopathy (SYNGAP1-DEE) is characterized by high rates of both epilepsy and autism spectrum disorder (ASD). While the clinical spectrum is well-documented, the link between specific seizure semiologies and caregiver-reported autistic behaviors is not well understood. This study analyzed the correlation between ten distinct seizure types, their frequencies, and a caregiver-reported autistic behavior score. MethodClinical data were extracted from the PATRE (PATient-based phenotyping and evaluation of therapy for Rare Epilepsies) Registry for SYNGAP1, in the framework of the EURAS project (Grant No. 101080580, Horizon Europe). This study employed a retrospective cross-sectional analysis of caregiver-reported registry data. Analysis was restricted to an analytic cohort of N=337 participants with complete data for both the epilepsy questionnaire (including epilepsy status, seizure semiology, and peak seizure frequency items) and the behavior questionnaire (from a total N=522 registry participants). Caregiver-reported autistic behaviors were quantified using a standardized caregiver-reported scale (Likert 1-5). Statistical associations were evaluated using the Wilcoxon rank-sum test to compare caregiver-reported autistic behavior scores across different seizure semiologies and Spearmans rank correlation to assess the impact of seizure frequency (9-point scale). ResultsWithin the analytic cohort (N=337), epilepsy was reported in 259 patients. Eyelid myoclonia was the most prevalent semiology, affecting 64.9% (n=168) of the epilepsy-positive group. Atypical absences (n=77) demonstrated the most profound and statistically robust association with higher caregiver-reported autistic behavior scores (FDR-adjusted p = 0.001). Significant associations were also observed for typical absences (n=70, FDR-adjusted p = 0.018), eyelid myoclonia (FDR-adjusted p = 0.018), myoclonic-atonic seizures (n=40, FDR-adjusted p = 0.019), and atonic seizures (n=72, FDR-adjusted p = 0.025). Focal and tonic-clonic seizures showed weaker associations (FDR-adjusted p = 0.026 and p = 0.047, respectively). Crucially, quantitative analysis revealed no significant correlation between ordinal caregiver-reported peak seizure frequency ratings and caregiver-reported autistic behavior scores across all semiologies (e.g., Eyelid Myoclonia: p=0.096; Atypical Absences: p=0.744), indicating no detectable association between peak-frequency ratings and caregiver-reported autistic behavior scores. ConclusionHigher caregiver-reported autistic behavior scores in SYNGAP1-DEE were most strongly associated with the presence of atypical absences, representing a generalized, thalamocortical seizure network dysfunction. In contrast, no detectable association was observed between caregiver-reported autistic behavior scores and the ordinal caregiver-reported peak seizure frequency metric. Atypical absences and related semiologies may serve as clinical "red flags" for increased neurodevelopmental comorbidity severity, regardless of reported peak seizure frequency. Abstract SummaryThis study investigates the relationship between ten seizure semiologies, seizure frequency, and severity of caregiver-reported autistic behaviors in a large-scale international cohort of N=337 patients with SYNGAP1-DEE. We identify a robust association between elevated caregiverreported autistic behavior scores and specific thalamocortical seizure patterns, most prominently atypical absences. Notably, our analysis reveals that this association is independent of seizure frequency, demonstrating no detectable association between this ordinal, caregiver-reported seizure frequency metric and caregiver-reported autistic behaviors.

BackgroundTBX4 variants are a recognised cause of paediatric pulmonary hypertension (PH), often associated with interstitial lung disease (ILD). Evidence for ILD-directed therapy in this group is lacking. MethodsWe conducted a retrospective study of children ([&le;]18 years) with TBX4-associated PH at a national centre (2001-2025). ILD was defined using ChILD-EU criteria. Patients treated with pulsed intravenous methylprednisolone were assessed for response using ChILD-EU categories. Secondary outcomes included respiratory severity score (RSS), functional class (FC), echocardiographic measures, and NT-proBNP. ResultsOf 21 children, 11 (52%) had ILD; 9 received corticosteroids. Median age at treatment was 0.8 years. A clear or best response occurred in 7/9 (78%). RSS improved in 6/9 (p=0.02), with all children on respiratory support showing partial or complete weaning. Functional class improved in all with FC III/IV at baseline (p=0.02). Right ventricular function improved (TAPSE z-score +1.65, p=0.04), and elevated NT-proBNP normalised. Key clinical milestones included ECMO weaning, transplant delisting, and discontinuation of prostacyclin therapy. No significant adverse effects were observed. Untreated children showed no early improvement. ConclusionsCorticosteroids were associated with meaningful improvements in respiratory and PH outcomes in TBX4-associated PH with ILD. Prospective evaluation is warranted.

Disability worsening is the critical long-term outcome in multiple sclerosis, yet the Expanded Disability Status Scale incompletely captures neurological deterioration and has limited sensitivity in the short time windows of clinical trials. Composite endpoints incorporating functional measures have been proposed to address these limitations, but whether they reliably improve detection of treatment effects has not been established across trials. We conducted a post-hoc analysis of individual patient data from ten phase III randomised controlled trials (ASCEND, BRAVO, CONFIRM, DEFINE, EXPAND, INFORMS, OLYMPUS, OPERA I/II, and ORATORIO; n = 9,369), spanning relapsing-remitting and progressive multiple sclerosis. Confirmed disability worsening was defined using harmonised criteria with the msprog package and confirmed at 24 weeks. Treatment effects were estimated using Cox proportional hazards models and combined across trials in a one-stage individual patient data framework. Composite endpoints were constructed from the Expanded Disability Status Scale, the timed 25-foot walk test, and the nine-hole peg test using logical unions (OR-type), intersections (AND-type), and majority-vote structures. Sensitivity to treatment effect was quantified using Z-scores (the ratio of the pooled log-hazard ratio to its standard error) and compared to the Expanded Disability Status Scale reference using interaction tests. Event rates varied across components: the timed walk test generated the highest rates (up to 46.8%) while the nine-hole peg test generated the lowest (as low as 2.1%). OR-type composite endpoints showed weaker treatment effects than the Expanded Disability Status Scale alone, with the largest reductions in sensitivity observed for endpoints incorporating the timed walk test ({Delta}Z up to +2.26; interaction p = 0.004). These findings were confirmed across disease subtypes and were pronounced in relapsing-remitting trials, where no composite endpoint outperformed the Expanded Disability Status Scale. In progressive multiple sclerosis, the combination of the Expanded Disability Status Scale and the nine-hole peg test showed numerically stronger treatment effects ({Delta}Z = -1.65), though interaction tests did not reach statistical significance (p = 0.051). Composite endpoints do not systematically improve treatment effect detection in multiple sclerosis trials. Increased event capture driven by the timed walk test introduces noise that dilutes the treatment signal rather than amplifying it, highlighting that event rate and endpoint quality are not interchangeable. Upper limb function assessed by the nine-hole peg test provides complementary and specific information, particularly in progressive disease. The combination of global disability and upper limb measures represents a promising direction for future endpoint development in progressive multiple sclerosis trials, warranting validation.

Background: Current medical large language model (LLM) evaluations largely rely on small collections of cases, whereas rigorous safety testing requires large-scale, diverse, and complex cases with verifiable ground truth. Multiple Sclerosis (MS) provides an ideal evaluation model, with validated diagnostic criteria and numerous paraclinical tests informing differential diagnosis, investigation, and management. Methods: We generated synthetic MS cases with ground-truth labels for diagnosis, localisation, and management. Four frontier LLMs (Gemini 3 Pro/Flash, GPT 5.2/5 mini) were instructed to analyse cases to provide anatomical localisation, differential diagnoses, investigations, and management plans. An automated evaluator compared these outputs to the ground-truth labels. Blinded subspecialty experts validated 70 cases for realism and automated evaluator accuracy. We then evaluated LLM decision-making across 1,000 cases and scaled to 10,000 to characterise rare, catastrophic failures. Results: Subspecialist expert review confirmed 100% synthetic case realism and 99.8% (95% CI 95.5 to 100) automated evaluation accuracy. Across 1,000 generated MS cases, all LLMs successfully included MS in the differential diagnoses for more than 91% cases. However, diagnostic competence did not associate with treatment safety. Gemini 3 models had low rates of clinically appropriate steroid recommendations (Flash: 7.2% 95% CI 5.6 to 8.8; Pro: 15.8% 95% CI 13.6 to 18.1) compared to GPT 5 mini (23.5% 95% CI 20.8 to 26.1), frequently overlooking contraindications like active infection. OpenAI models inappropriately recommended acute intravenous thrombolysis for MS cases (9.6% GPT 5.2; 6.4% GPT 5 mini) compared to below 1% for Gemini models. Expanded evaluation (to 10,000 cases) probed these errors in detail. Thrombolysis was recommended in 10.1% of cases lacking symptom timing information and paradoxically persisted (2.9%) even when symptoms were explicitly documented as more than 14 days old. Conclusion: Automated expert-level evaluation across 10,000 cases characterised artificial intelligence clinical blind spots hitherto invisible to small-scale testing. Massive-scale simulation and automated interrogation should become standard for uncovering serious failures and implementing safety guardrails before clinical deployment exposes patients to risk.

Background: Rett Syndrome (RTT) is a severe neurodevelopmental disorder affecting approximately 1 in 10,000 live female births worldwide. The Rett Syndrome Behaviour Questionnaire (RSBQ), remains one of the most widely used standardized behavioral assessment tools for RTT. However, the RSBQ was originally validated only in British English, limiting its applicability for Spanish-speaking caregivers and clinical centers across Latin America and Spain. Objective: The primary aim of this study was to develop and validate the comprehension of the Spanish translation of the RSBQ to ensure cultural and linguistic equivalence, enhance data reliability, and facilitate earlier, more accurate clinical assessments among Spanish-speaking RTT populations. Methods: Surveys were administered in two phases to Spanish-speaking caregivers between November 2023 and September 2025. Phase I consisted of 12 guided survey administrations with participants being able to ask clarifying questions and offer linguistic modifications of RSBQ questions. Phase II consisted of independent online administration of the refined Spanish RSBQ and a retest at least 7 days later. Participants were recruited through direct outreach and supported virtually during questionnaire completion. Results: Following data cleaning and quality control, a total of 51 caregivers successfully completed both surveys. The Spanish RSBQ demonstrated high caregiver comprehension and strong engagement across multiple Latin American countries, including Argentina, Mexico, and Peru. Responses were highly correlated between test and retest timepoints, and no question showed biased response distributions. A slight effect of response interval on test-retest correlation was observed, potentially indicating the impact of natural disease progression confounding retest evaluation for long (>80 day) intervals; however this effect did not impact the overall linguistic validation results as analysis of only <21 day test-retest responders confirmed the findings. Conclusions: This linguistic validation study represents the first formal step toward the clinical validation of the Spanish RSBQ, enabling broader inclusion of Spanish-speaking populations in RTT research. The collaborative, bilingual data collection strategy proved both feasible and effective, paving the way for multinational trials and expanding therapeutic accessibility through localized, patient-centered innovation.

Multiple sclerosis (MS) is a debilitating disease affecting more than 1 million Americans, and today is assessed primarily through magnetic resonance imaging (MRI) and observational clinical symptoms. Given the autoimmune nature of MS, we hypothesized that high-dimensional gene expression data from peripheral blood mononuclear cells (PBMCs), when analyzed with the assistance of AI, may collectively serve as valuable biomarkers for the real-time risk and progression of MS. Here, we present PBMC RNA sequencing (RNAseq) results from N=997 samples, including 540 MS, 221 neuromyelitis optica (NMO), and 149 healthy controls. We constructed and optimized ensemble models for three clinical outcomes: (1) discrimination of early MS (EDSS [&le;] 2.0) from healthy individuals with 74% AUC at 100% coverage, (2) differential diagnosis of MS from NMO with 91% AUC at 80% coverage, and (3) subtyping RRMS from progressive MS with 79% AUC at 80% coverage. To our knowledge, no prior molecular test has been reported for any of these three MS clinical tasks, and these results may have immediate impact on clinical management of MS patients. Two innovations that improved the stratification accuracy of our models: selection of gene sets based on expression variance in disease states, and use of non-linear rank sort and conviction weighting in the ensemble score calculation.

Abstract/SummaryO_ST_ABSBackgroundC_ST_ABSCerebrospinal fluid (CSF) -synuclein seed amplification assay (SAA) has emerged as a diagnostic biomarker for Parkinsons disease (PD) and has been linked to differences in disease severity and progression. However, whether SAA status predicts responsiveness to levodopa remains unknown. We investigated the longitudinal association between SAA status, levodopa responsiveness, dopaminergic denervation, and motor complications in sporadic PD. MethodsIn this longitudinal analysis, PD participants from the Parkinsons Progression Markers Initiative (PPMI) cohort with CSF SAA testing who initiated levodopa treatment were included. SAA- and SAA+ patients were matched on sex, age, and disease duration at treatment initiation. Motor severity was assessed using MDS-UPDRS Part III, with proportional and absolute responsiveness derived from ON and OFF medication states. Motor complications were assessed using MDS-UPDRS Part IV, and dopaminergic dysfunction was quantified using caudate DAT-SPECT. Linear mixed-effects models examined longitudinal differences as a function of SAA status. FindingsIn this analysis, 40 SAA- patients were compared to 183 matched SAA+ patients. SAA+ patients showed a slower rate of ON-state motor progression than SAA- patients (0.87 vs 3.47 points/year; p = 0.01). Consistently, proportional levodopa responsiveness increased over time in SAA+ patients while declining in SAA- patients (p = 0.036). These differences were accompanied by lower caudate DAT binding at treatment initiation in SAA- patients (p = 0.002) and faster dopaminergic decline over time (p = 0.008). Although SAA+ patients had fewer motor complications at treatment initiation, their progression was similar. InterpretationCSF -synuclein SAA status is associated with divergent levodopa response in PD, with SAA+ patients showing sustained and progressively greater motor benefit, while SAA- patients show declining responsiveness. Faster dopaminergic denervation in SAA- patients may underlie this difference. SAA status captures clinically relevant heterogeneity that may inform patient stratification and therapeutic decision-making.

BackgroundIn patients requiring respiratory support, clinicians rely on physical exam, radiologic, laboratory, and ventilator-derived measures for the provision of sufficient support while minimizing ventilator and "work of breathing" induced lung injury. Point of care lung ultrasound (LUS) is a widely available tool in hospital and clinic environments. To date, LUS has not been used to evaluate lung strain. MethodsWe collected LUS images in four anesthetized, neuromuscularly blocked, and mechanically ventilated pigs being used for another experiment. A feature tracking tool was developed which tracked echo-bright lung structures in ten second clips obtained in triplicate of the right and left, upper and lower lung fields using tidal volumes of 4, 6, 8, 10, and 12 mL/kg. Pleural lines were manually drawn and a program for quantifying lung strain developed with assistance from Anthropic Claude Artificial Intelligence tool. Structures were identified in inspiratory and expiratory frames and tracked bidirectionally with median strain per frame used for calculations. ResultsTriplicate measures of lung ultrasound images in four pigs had a median coefficients of variation of 35% (23-47% IQR) and linear modeling of strain with tidal volumes of 4-12 mL/kg showed positive correlation with R2 value ranging from 0.89 to 0.97. Strain measurements were similar after bronchial administration of 1.5M hydrochloric acid. ConclusionsRegional lung strain quantification using LUS is a viable and potentially useful tool for respiratory support management.