Explainable Artificial Intelligence to Predict Neurocognitive Disorder Progression in Multiple Sclerosis Using MRI and Clinical Data.

Multiple sclerosis (MS) is a complex disease characterized by diverse clinical presentations and progression patterns. Accurate classification and prediction of disease severity are crucial for personalized treatment. We applied machine learning (ML) to demographic, clinical and MRI data to distinguish MS patients from healthy controls (HC), classify MS phenotypes and predict disability using the Expanded Disability Status Scale (EDSS) score. We included 1,554 MS patients and 520 HC from the Italian Neuroimaging Network Initiative repository, all with neurological assessment and brain T2-/3D T1-weighted MRI. Derived MRI features included total and regional T2 lesion volumes (LV), and normalized tissue volumes from cortical and subcortical grey matter (GM), white matter, cerebellum and brainstem. ML models, including support vector machines, multi-layer perceptron networks, Random Forest and Gradient Boosting were trained for classification and prediction tasks. SHAP analysis ranked the most influential variables. ML models achieved 89-96% accuracy in distinguishing MS patients from HC, driven mainly by T2 LV and brainstem/cerebellar GM volumes. Relapsing vs progressive MS was classified with 92% accuracy, with EDSS, age, thalamic and cortical GM volumes as key predictors. EDSS prediction achieved an intra-class correlation of 0.56-0.76; most relevant contributors were T2 LV, sex, cortical/cerebellar GM and thalamic volumes. ML models demonstrated high accuracy in detecting MS, differentiating phenotypes, and predicting disability. Integrating demographic, clinical and MRI measures emerges as an effective strategy for patients' classification and disease severity assessment.

Multiple sclerosis (MS) susceptibility is influenced by genetics; however, little is known about genetic determinants of disease expression. We aimed at assessing genetic factors influencing quantitative neuroimaging measures in two cohorts of progressive MS (PMS) and relapsing-remitting MS (RRMS) patients. Ninety-nine PMS and 214 RRMS patients underwent a 3-T brain magnetic resonance imaging (MRI) scan, with the measurement of five MRI metrics including T2 lesion volumes and measures of white matter, grey matter, deep grey matter, and hippocampal volumes. A candidate pathway strategy was adopted; gene set analysis was carried out to estimate cumulative contribution of genes to MRI phenotypes, adjusting for relevant confounders, followed by single nucleotide polymorphism (SNP) regression analysis. Seventeen Kyoto Encyclopedia of Genes and Genomes pathways and 42 Gene Ontology (GO) terms were tested. We additionally included in the analysis genes with enriched expression in brain cells. Gene set analysis revealed a differential pattern of association across the two cohorts, with processes related to sodium homeostasis being associated with grey matter volume in PMS (p=0.002), whereas inflammatory-related GO terms such as adaptive immune response and regulation of inflammatory response appeared to be associated with T2 lesion volume in RRMS (p=0.004 and p=0.008, respectively). As for SNPs, the rs7104613T mapping to SPON1 gene was associated with reduced deep grey matter volume (β=-0.731, p=3.2*10-7 ) in PMS, whereas we found evidence of association between white matter volume and rs740948A mapping to SEMA3A gene (β=22.04, p=5.5*10-6 ) in RRMS. Our data suggest a different pattern of associations between MRI metrics and functional processes across MS disease courses, suggesting different phenomena implicated in MS.

Objective to investigate whether optical coherence tomography (OCT) could demonstrate axonal loss through thinning of the retinal nerve fiber layer (RNFL) in multiple sclerosis (MS) patients. Furthermore, the degree of RNFL loss was compared in the different MS subgroups (with or without optic neuritis, affected or fellow eye, single episode or recurrent optic neuritis, relapsing remitting or progressive MS). RNFL thinning was also determined in MS patients who had serial OCT without any intervening clinical optic neuritis. Design: Retrospective chart review. Setting: Academic tertiary care MS centers. Participants: 177 patients (334 eyes) with multiple sclerosis, with or without optic neuritis, and 159 healthy controls that underwent OCT RNFL measurements. Main Outcome: Retinal nerve fiber layer measurements by OCT. Results: Average RNFL measurements were thinner in MS patients (90 μm) compared to controls (105 μm), p <0.0001. RNFL was significantly reduced in MS patients with optic neuritis (87 μm) versus those without optic neuritis (94 μm), p <0.0001. Among the different quadrants, the degree of RNFL loss was greatest in the temporal quadrant of MS patients (22%). Progressive (primary and secondary progressive) MS patients had thinner RNFL (82 μm) compared to all relapsing remitting MS patients (90 μm), p <0.0001. Greater RNFL loss was seen in SPMS patients (77 μm) versus PPMS (88 μm), p = 0.004. In the 45 MS patients without any intervening clinical optic neuritis, serial OCT (mean of 2 OCT scans per patients, averaging 11 months apart) showed that RNFL decreased by 3.7 μm per year. Conclusion: Retinal nerve fiber layer is significantly reduced in patients with multiple sclerosis. Progressive MS subtypes showed more marked RNFL thinning than relapsing remitting MS. This study reflects the role of OCT in MS patient monitoring and its potential as a surrogate marker in MS therapeutic trials.

BackgroundThe hippocampus is a clinically relevant region where neurogenesis and neuroplasticity occur throughout the whole lifespan. Neuroinflammation and cardiorespiratory fitness (CRF) may influence hippocampal integrity by modulating the processes promoting...

Although MR elastography allows for quantitative evaluation of liver stiffness to assess chronic liver diseases, it has associated drawbacks related to additional scanning time, patient discomfort, and added costs. To develop a machine learning model that can categorically classify the severity of liver stiffness using both anatomical T2-weighted MRI and clinical data for children and young adults with known or suspected pediatric chronic liver diseases. We included 273 subjects with known or suspected chronic liver disease. We extracted data including axial T2-weighted fast spin-echo fat-suppressed images, clinical data (e.g., demographic/anthropomorphic data, particular medical diagnoses, laboratory values) and MR elastography liver stiffness measurements. We propose DeepLiverNet (a deep transfer learning model) to classify patients into one of two groups: no/mild liver stiffening (<3kPa) or moderate/severe liver stiffening (≥3kPa). We conducted internal cross-validation using 178 subjects, and external validation using an independent cohort of 95 subjects. We assessed diagnostic performance using accuracy, sensitivity, specificity and area under the receiver operating characteristic curve (AuROC). In the internal cross-validation experiment, the combination of clinical and imaging data produced the best performance (AuROC=0.86) compared to clinical (AuROC=0.83) or imaging (AuROC=0.80) data alone. Using both clinical and imaging data, the DeepLiverNet correctly classified patients with accuracy of 88.0%, sensitivity of 74.3% and specificity of 94.6%. In our external validation experiment, this same deep learning model achieved an accuracy of 80.0%, sensitivity of 61.1%, specificity of 91.5% and AuROC of 0.79. A deep learning model that incorporates clinical data and anatomical T2-weighted MR images might provide a means of risk-stratifying liver stiffness and directing the use of MR elastography.

Characteristics of COVID-19 disease in multiple sclerosis patients

Lymphocyte calcium influx characteristics and its modulation by Kv1.3 and IKCa1 potassium channel inhibitors in multiple sclerosis

Introduction: Dynamic functional connectivity (dFC) allows capturing recurring patterns (states) of interaction among functional networks. In this study, we investigated resting state (RS) dFC abnormalities across the different clinical phenotypes of multiple sclerosis (MS) and assessed their correlation with motor and cognitive performances. Methods: RS functional magnetic resonance imaging (fMRI) and 3D T1-weighted MRI data were acquired from 128 MS patients (69 relapsing-remitting [RR] MS, 34 secondary progressive [SP] MS, and 25 primary progressive [PP] MS) and 40 healthy controls (HC). RS fMRI data underwent independent component analysis and sliding-window correlations, to identify recurring dFC states and between-group dFC differences in the main networks. Results: dFC identified three recurring connectivity states: State 1 (frequency of appearance during fMRI acquisition = 57%, low dFC strength), State 2 (frequency = 19%, middle-high dFC strength), and State 3 (frequency = 24%, high sensorimotor and visual dFC strength). Compared to HC, MS (as a whole), RRMS, and PPMS patients exhibited lower State1/State 3 dFC (p = 0.0001, corrected) between sensorimotor, cerebellar, and cognitive networks, and some dFC increments (p = 0.001-0.05, uncorrected) in sensorimotor, visual, default-mode, and frontal/attention networks in States 2 and 3. Similar results were observed in SPMS versus RRMS patients. In MS, dFC decrease in sensorimotor, default-mode, and frontal/attention networks was correlated with worse motor and cognitive performances. Conclusions: MS patients exhibited overall lower dFC, and marginally higher dFC in sensorimotor/cognitive networks in the less-frequent middle/high-connected States. dFC abnormalities became more severe in progressive MS and correlated with motor and cognitive impairment, suggesting the presence of maladaptive mechanisms concomitant with accumulation of damage. Impact statement This is the first study exploring reorganization of dynamic functional connectivity in patients with multiple sclerosis (MS) across the main clinical phenotypes of the disease. Here, we demonstrated abnormalities of connectivity dynamism, which were present at all disease stages, but became more severe in progressive MS and correlated with worse motor and cognitive performances. These findings suggested that progressive MS patients might experience a maladaptive neuronal response to transient loss of dynamic coordination and flexibility among sensory and cognitive brain regions, leading to the progression of clinical impairment.

Depression in Multiple Sclerosis

Conventional brain magnetic resonance imaging (MRI) in systemic diseases with central nervous system involvement (SDCNS) may imitate MRI findings of multiple sclerosis (MS). In order to better describe the MRI characteristics of these conditions, in our study we assessed brain volume parameters in MS (n = 58) and SDCNS (n = 41) patients using two-dimensional linear measurements (2DLMs): bicaudate ratio (BCR), corpus callosum index (CCI) and width of third ventricle (W3V). In SDCNS patients, all 2DLMs were affected by age (CCI p = 0.005, BCR p < 0.001, W3V p < 0.001, respectively), whereas in MS patients only BCR and W3V were (p = 0.001 and p = 0.015, respectively). Contrary to SDCNS, in the MS cohort BCR and W3V were associated with T1 lesion volume (T1LV) (p = 0.020, p = 0.009, respectively) and T2 lesion volume (T2LV) (p = 0.015, p = 0.009, respectively). CCI was associated with T1LV in the MS cohort only (p = 0.015). Moreover, BCR was significantly higher in the SDCNS group (p = 0.01) and CCI was significantly lower in MS patients (p = 0.01). The best predictive model to distinguish MS and SDCNS encompassed gender, BCR and T2LV as the explanatory variables (sensitivity 0.91; specificity 0.68; AUC 0.86). Implementation of 2DLMs in the brain MRI analysis of MS and SDCNS patients allowed for the identification of diverse patterns of local brain atrophy in these clinical conditions.

Neurocognitive disorders (NCDs) have a variable decline in cognitive function, while loneliness was associated with cognitive impairment and increased dementia risk. In the present study, we examined the associations of loneliness with functional and cognitive status in patients with minor (mild cognitive impairment) and major NCDs (dementia). We diagnosed mild NCD (n = 42) and major NCD (n = 164) through DSM-5 criteria on 206 participants aged > 65 years using the UCLA 3-Item Loneliness Scale (UCLA-3) to evaluate loneliness, the activities of daily living (ADL) and the instrumental activities of daily living (IADL) scales to measure functional status, and Mini-Mental State Examination (MMSE) to assess cognitive functions. In a multivariate regression model, the effect of loneliness on cognitive functions was negative in major (β = -1.05, p < 0.0001) and minor NCD (β = -0.06, p < 0.01). In the fully adjusted multivariate regression model (sex-age-education-multimorbidity-depressive symptoms-antidementia drug treatment), the effect of loneliness remained negative for major NCD and became positive for minor NCD (β = 0.09, p < 0.001). The effect of loneliness on IADL (β = -0.26, p < 0.0001) and ADL (β = -0.24, p < 0.001) showed a negative effect for major NCD across the different models, while for minor NCD, the effect was positive (IADL: β = 0.26, p < 0.0001; ADL: β = 0.05, p = 0.01). Minor NCD displayed different levels of MMSE (β = 6.68, p < 0.001) but not ADL or IADL, compared to major NCD for the same levels of loneliness. MANOVA pill test suggested a statistically significant and different interactive effect of loneliness on functional and cognitive variables between minor and major NCDs. We confirmed the relationships between loneliness and cognitive and functional status in major NCD, observing a novel trend in minor NCD.

Background: Hypoperfusion, vascular pathology, and cardiovascular risk factors are associated with disease severity in multiple sclerosis (MS).1,2 In particular, the total cerebral arterial blood flow (CABF), measured as a sum of all arterial flow in the neck, was associated with the cognitive performance of MS patients.3&#x0D; Objective: To assess relationships between CABF and serum neurofilament light chain (sNfL), as neuronal damage biomarker with good prognostic value and treatment responsiveness.4 If the cerebrovascular changes are an independent pathophysiological factor in MS, a relationship should remain significant after controlling for common MS-based disease measures (i.e., T2 lesion volume and brain volume).&#x0D; Materials and methods: Total CABF was measured in 137 patients (86 clinically isolated syndrome (CIS)/relapsing-remitting (RR) and 51 progressive MS (PMS)) and 48 healthy controls (HCs) using Doppler ultrasound. sNfL was quantitated using a single molecule assay (Simoa). Three point zero T magnetic resonance imaging (MRI) examination allowed quantification of T2 lesion and whole-brain volume (WBV). Multiple linear regression models determined the sNfL associated with CABF after correction for demographic and MRI-derived variables.&#x0D; Results: After adjustment for age, sex and body mass index (BMI), total CABF remained statistically significant and model comparisons showed that CABF explained additional 2.6% of the sNfL variance (β=-0.167, p=0.044). (Table 1) CABF also remained significant in a step-wise regression model (β=0.18, p=0.034) upon the inclusion of T2 lesion burden and WBV effects. The explained sNfL variance improved from 17.4%, 22.7% with the presence of at least 2 CVD variable and 25.8% with both CVD and CABF predictors. Lastly, the disease-modifying therapy was not kept in the final model as an independent predictor of sNfL. Patients in the lowest CABF quartile (CABF≤761 mL/min) had significantly higher sNfL (34.6 pg/mL versus 23.9 pg/mL, adjusted-p=0.042) when compared to the highest quartile (CABF≥1130 mL/min).&#x0D; Conclusions: Lower CABF is associated with increased sNfL in MS patients, highlighting direct and independent relationship between cerebral hypoperfusion and axonal pathology. This relationship remained significant in the CIS/RRMS after adjusting for age, sex, and BMI effects.

The hippocampus has a critical role in episodic memory and visuospatial learning and consolidation. We assessed the patterns of whole and regional hippocampal atrophy in a large group of multiple sclerosis (MS) patients, and their correlations with neuropsychological impairment. From 103 MS patients and 28 healthy controls (HC), brain dual-echo and high-resolution 3D T1-weighted images were acquired using a 3.0-Tesla scanner. All patients underwent a neuropsychological assessment of hippocampal-related cognitive functions, including Paired Associate Word Learning, Short Story, delayed recall of Rey-Osterrieth Complex Figure and Paced Auditory Serial Attention tests. The hippocampi were manually segmented and volumes derived. Regional atrophy distribution was assessed using a radial mapping analysis. Correlations between hippocampal atrophy and clinical, neuropsychological and MRI metrics were also evaluated. Hippocampal volume was reduced in MS patients vs HC (p < 0.001 for both right and hippocampus). In MS patients, radial atrophy affected CA1 subfield and subiculum of posterior hippocampus, bilaterally. The dentate hilus (DG:H) of the right hippocampal head was also affected. Regional hippocampal atrophy correlated with brain T2 and T1 lesion volumes, while no correlation was found with disability. Damage to the CA1 and subiculum was significantly correlated to the performances at hippocampal-targeted neuropsychological tests. These results show that hippocampal subregions have a different vulnerability to MS-related damage, with a relative sparing of the head of the left hippocampus. The assessment of regional hippocampal atrophy may help explain deficits of specific cognitive functions in MS patients, including memory and visuospatial abilities.

Should Spinal MRI Be Routinely Performed in Patients With Clinically Isolated Optic Neuritis?

It has been suggested that multiple sclerosis (MS) patients with positive anticardiolipin antibodies (ACLA) have some atypical features, including absent oligoclonal bands (OCB) in the cerebrospinal fluid (CSF). Our aim was to compare the frequencies of ACLA and related laboratory and clinical features in OCB negative (OCB-) and positive (OCB+) MS patients. We compared 41 OCB- patients attending a MS Clinic in a tertiary referral center, with 206 OCB+ patients. ACLA, anti-beta2-glycoprotein and other autoantibodies, lupus anticoagulant and coagulation markers were measured. We found a higher frequency of ACLA in OCB- patients, 18/41 versus 33/206 in OCB+ patients (P<0.0001). OCB- patients had more progressive MS than OCB+ subjects. There were no differences in age, sex, Expanded Disability Status Scale (EDSS) score, antiphospholipid syndrome symptoms between the groups. ACLA+ MS patients were more frequently in the OCB- group. Although this may suggest that they represent a special subgroup of MS, no other clinical or laboratory findings distinguish the groups. Although OCB- MS patients may be thought to be less active immunologically, this study shows they have more frequently ACLA than OCB+ patients. OCB- MS patients in our cohort do not appear to have a more benign form of MS, as has previously been suggested.