Clinical applications of deep learning in neuroinflammatory diseases: A scoping review.

BACKGROUND: Deep learning (DL) is an artificial intelligence technology that has aroused much excitement for predictive medicine due to its ability to process raw data modalities such as images, text, and time series of signals. OBJECTIVES: Here, we intend to give the clinical reader elements to understand this technology, taking neuroinflammatory diseases as an illustrative use case of clinical translation efforts. We reviewed the scope of this rapidly evolving field to get quantitative insights about which clinical applications concentrate the efforts and which data modalities are most commonly used. METHODS: We queried the PubMed database for articles reporting DL algorithms for clinical applications in neuroinflammatory diseases and the radiology.healthairegister.com website for commercial algorithms. RESULTS: The review included 148 articles published between 2018 and 2024 and five commercial algorithms. The clinical applications could be grouped as computer-aided diagnosis, individual prognosis, functional assessment, the segmentation of radiological structures, and the optimization of data acquisition. Our review highlighted important discrepancies in efforts. The segmentation of radiological structures and computer-aided diagnosis currently concentrate most efforts with an overrepresentation of imaging. Various model architectures have addressed different applications, relatively low volume of data, and diverse data modalities. We report the high-level technical characteristics of the algorithms and synthesize narratively the clinical applications. Predictive performances and some common a priori on this topic are finally discussed. CONCLUSION: The currently reported efforts position DL as an information processing technology, enhancing existing modalities of paraclinical investigations and bringing perspectives to make innovative ones actionable for healthcare.

A decade of fingolimod in multiple sclerosis: Insights from a large real-world cohort study.

BACKGROUND AND OBJECTIVES: Ten years after its authorization, data about fingolimod use in real-world setting is still scarce. Here we describe the long-term evolution of fingolimod-treated relapsing-remitting MS (RRMS) patients and determine baseline characteristics associated with risk of relapses or disability. METHODS: We analyzed baseline characteristics and clinical evolution of 1227 patients with RRMS treated with fingolimod from 2010 to 2019 in 4 French MS referral centers. We used Cox models to determine risks factors of relapses and sustained EDSS worsening. RESULTS: Median follow-up duration was 50 months, and 63% of patients remained fingolimod-treated at the end of follow-up. Mean 5-years annualized relapse rate (ARR) decreased from 0.63 (0.60-0.67) to 0.26 (0.24-0.29, P<0.001), while the mean EDSS rose from 2.5 (2.4-2.6) to 3.0 (2.8-3.1, P<0.001). Female sex, lower age, higher EDSS and use of natalizumab were associated with relapse risk. Female sex was associated with sustained EDSS increase risk. CONCLUSIONS: Based on a large real-world cohort, our results confirm the durable reduction of the ARR described in pivot studies. Switching from moderate-efficacy DMT to fingolimod decreased the relapse risk. Switching patients from high-efficacy DMT increased risk of relapse, but the overall five-years ARR remained stable.

Impact of methodological choices in comparative effectiveness studies: application in natalizumab versus fingolimod comparison among patients with multiple sclerosis.

BACKGROUND: Natalizumab and fingolimod are used as high-efficacy treatments in relapsing-remitting multiple sclerosis. Several observational studies comparing these two drugs have shown variable results, using different methods to control treatment indication bias and manage censoring. The objective of this empirical study was to elucidate the impact of methods of causal inference on the results of comparative effectiveness studies. METHODS: Data from three observational multiple sclerosis registries (MSBase, the Danish MS Registry and French OFSEP registry) were combined. Four clinical outcomes were studied. Propensity scores were used to match or weigh the compared groups, allowing for estimating average treatment effect for treated or average treatment effect for the entire population. Analyses were conducted both in intention-to-treat and per-protocol frameworks. The impact of the positivity assumption was also assessed. RESULTS: Overall, 5,148 relapsing-remitting multiple sclerosis patients were included. In this well-powered sample, the 95% confidence intervals of the estimates overlapped widely. Propensity scores weighting and propensity scores matching procedures led to consistent results. Some differences were observed between average treatment effect for the entire population and average treatment effect for treated estimates. Intention-to-treat analyses were more conservative than per-protocol analyses. The most pronounced irregularities in outcomes and propensity scores were introduced by violation of the positivity assumption. CONCLUSIONS: This applied study elucidates the influence of methodological decisions on the results of comparative effectiveness studies of treatments for multiple sclerosis. According to our results, there are no material differences between conclusions obtained with propensity scores matching or propensity scores weighting given that a study is sufficiently powered, models are correctly specified and positivity assumption is fulfilled.

Atypical myelitis in patients with multiple sclerosis: Characterization and comparison with typical multiple sclerosis and neuromyelitis optica spectrum disorders.

BACKGROUND: Atypical myelitis in multiple sclerosis (MS) is characterized by extensive myelitis in the longitudinal (longitudinally extensive transverse myelitis) or axial plane (transverse myelitis). OBJECTIVE: To characterize a cohort of MS patients with atypical myelitis. METHODS: Atypical myelitis was extracted from the French and Luxembourg MS databases and compared to two cohorts of MS patients with typical myelitis and neuromyelitis optica spectrum disorders (NMOSDs) patients with myelitis. RESULTS: We enrolled 28 MS patients with atypical myelitis, 68 MS patients with typical myelitis and 119 NMOSD patients with a first episode of myelitis. MS patients with atypical myelitis were characterized by a mean age of 34.0 (±10.7) years and 64.3% were women. In 82.1% of the patients, atypical myelitis was the first episode of MS. Mean Expanded Disability Status Scale (EDSS) scores at nadir and 3-6 months after onset were 4.1 ± 2.1 and 3.3 ± 2, respectively. Differences between groups revealed a predominance of cervicothoracic myelitis and a higher level of disability in NMOSD patients. Disability in MS patients with atypical myelitis was more severe than in the MS patients with typical myelitis; 28% had already converted to progressive MS within our mean follow-up of 39.6 (±30.4) months. CONCLUSION: Atypical myelitis may be the first presentation of MS and is associated with poorer prognosis.

Impact of disease-modifying treatments on humoral response after COVID-19 vaccination: A mirror of the response after SARS-CoV-2 infection.

OBJECTIVE: To analyze the humoral response after COVID-19 vaccination in patients with multiple sclerosis (MS) according to disease-modifying treatments (DMTs) and in comparison with the humoral response after SARS-CoV-2 infection. METHODS: We included 28 MS patients with serological results after COVID-19 vaccination (Pfizer-BioNTech or Moderna ARNm) and 61 MS patients with serological results after COVID-19 (COVID-19 group) among patients followed up at the MS Center of Strasbourg, France, between January and April 2021. The primary endpoint was the IgG index according to DMTs (anti-CD20 mAb, sphingosine 1-phosphate receptor [S1PR] modulator and other treatments) and COVID-19 vaccine or COVID-19 groups. RESULTS: In the vaccinated MS patients, the median IgG index was lower in patients treated with anti-CD20 mAb and in patients treated with S1PR modulator compared to patients receiving other or no DMTs (4.80 [1.58-28.6], 16.5 [16.3-48.5], 1116 [434-1747] and 1272 [658-1886], respectively, P<0.001). Similar results were found for MS patients after COVID-19. CONCLUSIONS: Patients with MS and treated with S1PR modulators or anti-CD20 mAb had a reduced humoral response after COVID-19 vaccine.

Identification and measurement of cervical spinal cord atrophy in neuromyelitis optica spectrum disorders (NMOSD) and correlation with clinical characteristics and cervical spinal cord MRI data.

INTRODUCTION: The spinal cord is one of the two main targets of neuromyelitis optica (NMO). The aim of this study was to highlight cervical spinal cord atrophy in NMO patients as compared to controls and to assess correlations between atrophy and clinical characteristics and cervical spinal cord MRI data. METHODS: This prospective study investigated 15 patients with a diagnosis of NMOSD and 15 healthy controls. The whole cervical spinal cord was explored by MRI. The cross-sectional area (CSA) was estimate at every level of cup. This measurement was then averaged on the whole cervical spinal cord, providing a single measurement for every subject, denoted as mean CSA. RESULTS: Mean CSA was 68.5 mm2 in the population of NMO patients and 72.8 mm2 in the population of healthy subjects. NMO patients had significantly smaller cervical spinal cord area than healthy controls (T test=0.009). Cervical spinal cord atrophy was associated with clinical signs of medullary involvement (T test=0.0006). There was a tendency toward a relation between cervical spinal cord atrophy and the Expanded Disability Status Scale (EDSS) (T test=0.07). This correlation seems statistically significant (T test<0.05) at the level of the upper cervical spinal cord (C2-C3) CONCLUSION: This study provides the first evidence of cervical spinal cord atrophy in NMOSD by studying the entire cervical spinal cord. Upper cervical spinal cord atrophy was substantially correlated to clinical disability and seems more involved in the development of clinical disability in NMOSD patients in comparison to the lower cervical spinal cord.

A comparative evaluation of different neuromyelitis optica spectrum disorder sets of criteria.

BACKGROUND AND PURPOSE: Three different sets of criteria have been proposed for the diagnosis of neuromyelitis optica spectrum disorder (NMOSD). The objective was to compare the specificity, sensitivity and diagnostic accuracy of the three different sets of NMOSD criteria in patients presenting with inflammatory disorders of the central nervous system suggestive of NMOSD. METHODS: For 236 suspected NMOSD patients referred for serum aquaporin-4 immunoglobulin G antibody (AQP4-IgG) testing between 2012 and 2014, the three sets of NMOSD criteria [1999, 2006 NMO criteria and 2015 International Panel for NMO Diagnosis (IPND) criteria] were applied and compared to the final diagnosis. RESULTS: Seventy-six patients fulfilled at least one set of criteria and 28 patients fulfilled all NMOSD sets of criteria. The final diagnosis was NMOSD in 66 cases, multiple sclerosis according to the MacDonald 2010 criteria in 85 cases and another diagnosis in 85 cases. The 2006 NMO criteria have the highest specificity (99%) and the 2015 IPND NMOSD criteria the highest sensitivity (97%). For the 1999, 2006 and 2015 IPND NMOSD criteria, the accuracy was respectively 82%, 87% and 97%. CONCLUSIONS: Our study highlights the limitations of the first set of criteria that include the optico-spinal form of multiple sclerosis. The accuracy of NMOSD diagnostic criteria improved from 1999 to 2015. It confirms the increased performance of the last set of criteria which covers a larger spectrum of clinical presentation. This study raises some concerns for classifying patients with seronegative transverse myelitis or optic neuritis, and myelin oligodendrocyte glycoprotein (MOG) antibody- associated disease.

MSCopilot, a new multiple sclerosis self-assessment digital solution: results of a comparative study versus standard tests.

BACKGROUND AND PURPOSE: Assessing patients' disability in multiple sclerosis (MS) requires time-consuming batteries of hospital tests. MSCopilot is a software medical device for the self-assessment of patients with MS (PwMS), combining four tests: walking, dexterity, cognition and low contrast vision. The objective was to validate MSCopilot versus the Multiple Sclerosis Functional Composite (MSFC). METHODS: This multicentre, open-label, randomized, controlled, crossover study enrolled 141 PwMS and 76 healthy controls (HCs). All participants performed MSCopilot and MSFC tests at day 0. To assess reproducibility, 46 PwMS performed the same tests at day 30 ± 3. The primary end-point was the validation of MSCopilot versus MSFC for the identification of PwMS against HCs, quantified using the area under the curve (AUC). The main secondary end-point was the correlation of MSCopilot z-scores with MSFC z-scores. RESULTS: In all, 116 PwMS and 69 HCs were analysed. The primary end-point was achieved: MSCopilot performance was non-inferior to that of MSFC (AUC 0.92 and 0.89 respectively; P = 0.3). MSCopilot and MSFC discriminated PwMS and HCs with 81% and 76% sensitivity and 82% and 88% specificity respectively. Digital and standard test scores were highly correlated (r = 0.81; P < 0.001). The test-retest study demonstrated the good reproducibility of MSCopilot. CONCLUSION: This study confirms the reliability of MSCopilot and its usability in clinical practice for the monitoring of MS-related disability.

Do disease-modifying drugs (DMD) have a positive impact on the occurrence of secondary progressive multiple sclerosis? Yes.

During the 20 past years, the management of multiple sclerosis (MS) has largely changed especially concerning therapeutical approach. Before 1996, treatments were restricted to corticosteroids for relapses, several symptomatic treatments and unselective immunosuppressive drugs (azathioprine, cyclophosphamide, methotrexate) with a low evidence of any efficacy. In the present review, we analyze the principal real-life cohorts of MS during several periods (before therapeutical modern area, first-generation treatment area and most recent period). Despite many methodological problems, we observe globally a delay of around 3-5 years between untreated cohorts and first-generation treatments for going to EDSS 6 which is probably the most robust score. This delay is clearly increase to at least 15 years with the most recent cohort treated first and second-line treatments confirming that early and more intensive treatment are necessary to have a long-term efficacy on disability progression and especially on severe disability represent by EDSS 6. Larger cohorts with longer follow-up is necessary to confirm these tendencies and OFSEP observatory or MS base will probably provide us the possibility to conclude in a couple of years.

MOG encephalomyelitis: international recommendations on diagnosis and antibody testing.

Over the past few years, new-generation cell-based assays have demonstrated a robust association of autoantibodies to full-length human myelin oligodendrocyte glycoprotein (MOG-IgG) with (mostly recurrent) optic neuritis, myelitis and brainstem encephalitis, as well as with acute disseminated encephalomyelitis (ADEM)-like presentations. Most experts now consider MOG-IgG-associated encephalomyelitis (MOG-EM) a disease entity in its own right, immunopathogenetically distinct from both classic multiple sclerosis (MS) and aquaporin-4 (AQP4)-IgG-positive neuromyelitis optica spectrum disorders (NMOSD). Owing to a substantial overlap in clinicoradiological presentation, MOG-EM was often unwittingly misdiagnosed as MS in the past. Accordingly, increasing numbers of patients with suspected or established MS are currently being tested for MOG-IgG. However, screening of large unselected cohorts for rare biomarkers can significantly reduce the positive predictive value of a test. To lessen the hazard of overdiagnosing MOG-EM, which may lead to inappropriate treatment, more selective criteria for MOG-IgG testing are urgently needed. In this paper, we propose indications for MOG-IgG testing based on expert consensus. In addition, we give a list of conditions atypical for MOG-EM ("red flags") that should prompt physicians to challenge a positive MOG-IgG test result. Finally, we provide recommendations regarding assay methodology, specimen sampling and data interpretation.

Myelin oligodendrocyte glycoprotein antibodies in neuromyelitis optica spectrum disorder.

Neuromyelitis optica spectrum disorder (NMOSD) is a severe inflammatory disease of the central nervous system characterized, in particular, by disabling episodes of optic neuritis and longitudinal extensive transverse myelitis. Its main pathogenic characteristic is the presence of anti-aquaporin-4 antibodies (AQP4-Abs) in the serum of affected patients. However, a proportion of patients with the typical NMOSD phenotype are, in fact, negative (seronegative) for AQP4-Abs and, within this category of patients, a proportion of them instead express antibodies to myelin oligodendrocyte glycoprotein (MOG-Abs). The presence of MOG-Abs in the sera of seronegative NMOSD patients is more frequently associated with monophasic disease and moderate symptom severity, and also appears to predict a better outcome. The present report is a review of the clinical and immunological features of MOG-Ab-positive NMOSD.

Is the Choosing Wisely® campaign model applicable to the management of multiple sclerosis in France? A GRESEP pilot study.

BACKGROUND: Launched in the US in 2012, Choosing Wisely® is a campaign promoted by the American Board of Internal Medicine (ABIM) Foundation with the goal of improving healthcare effectiveness by avoiding wasteful or unnecessary medical tests, treatments and procedures. It uses concise recommendations produced by national medical societies to start discussions between physicians and patients on the relevance of these services as part of a shared decision-making process. The Multiple Sclerosis Focus Group (Groupe de Reflexion Autour de la Sclérose en Plaques; GRESEP) undertook a pilot study to assess the relevance and feasibility of this approach in the management of multiple sclerosis (MS) in France. METHODS: Recommendations were developed using the formal consensus method from the guidelines of the French National Health Authority (HAS). A steering committee selected the themes and drafted concise evidence reviews. An independent rating group then assessed these recommendations for clarity, relevance and feasibility. RESULTS: Seven recommendations were accepted: (1) avoid systematic ordering of multimodal evoked potential studies for diagnosing MS; (2) do not treat MS relapses with low-dose oral corticosteroids; (3) when treating MS relapse with high-dose corticosteroids, the systematic use of the intravenous route is unnecessary if the oral route can be used; (4) systematic hospitalization is not necessary for treating MS relapse with high-dose corticosteroid therapy, particularly if the oral route is used, except for the first treated relapse and the presence of exclusion or non-eligibility criteria; (5) in the absence of clinical signs or symptoms of urinary infection, avoid systematic screening with urine microscopy and culture before the administration of corticosteroid therapy for MS relapse in patients using intermittent self-catheterization; (6) avoid antibiotic treatment of clinically asymptomatic MS patients using intermittent self-catheterization, even if urine microscopy and culture reveal the presence of microorganisms; and (7) avoid introducing symptomatic drug treatment for MS-related fatigue. CONCLUSION: This pilot study, the first of its kind in France, has demonstrated the relevance and feasibility of adapting the Choosing Wisely® model to MS by practitioners specializing in the disorder. However, the acceptability of these recommendations by other practitioners in other specialist fields as well as their impact on everyday clinical practices now need to be studied.

[MOG encephalomyelitis: international recommendations on diagnosis and antibody testing]. / MOG-Enzephalomyelitis: Internationale Empfehlungen zu Diagnose und Antikörpertestung.

Over the past few years, new-generation cell-based assays have demonstrated a robust association of autoantibodies to full-length human myelin oligodendrocyte glycoprotein (MOG-IgG) with (mostly recurrent) optic neuritis, myelitis and brainstem encephalitis, as well as with acute disseminated encephalomyelitis (ADEM)-like presentations. Most experts now consider MOG-IgG-associated encephalomyelitis (MOG-EM) a disease entity in its own right, immunopathogenetically distinct from both classic multiple sclerosis (MS) and aquaporin-4 (AQP4)-IgG-positive neuromyelitis optica spectrum disorders (NMOSD). Owing to a substantial overlap in clinicoradiological presentation, MOG-EM was often unwittingly misdiagnosed as MS in the past. Accordingly, increasing numbers of patients with suspected or established MS are currently being tested for MOG-IgG. However, screening of large unselected cohorts for rare biomarkers can significantly reduce the positive predictive value of a test. To lessen the hazard of overdiagnosing MOG-EM, which may lead to inappropriate treatment, more selective criteria for MOG-IgG testing are urgently needed. In this paper, we propose indications for MOG-IgG testing based on expert consensus. In addition, we give a list of conditions atypical for MOG-EM ("red flags") that should prompt physicians to challenge a positive MOG-IgG test result. Finally, we provide recommendations regarding assay methodology, specimen sampling and data interpretation, and propose for the first time diagnostic criteria for MOG-EM.

Efficacy of rituximab in refractory neuromyelitis optica.

BACKGROUND: Despite a growing use of rituximab (RTX) in neuromyelitis optica (NMO), data are lacking in patients with refractory NMO (RNMO), defined as cases with at least one relapse during immunosuppressive therapy. OBJECTIVE: The purpose of this study was to assess RTX as a maintenance therapy in RNMO. METHODS: Out of a total of 305 NMO cases from a population-based cohort, 21 RNMO patients received RTX during a mean follow-up period of 31 months. RESULTS: After RTX, 11 patients (52.3%) were relapse free, meaning that 47.7% were refractory to RTX. The mean annualized relapse rate decreased from 1.3 to 0.4 (p<0.001) and median EDSS from 5 to 3 (p=0.02). Body mass index (BMI) was predictive of EDSS worsening. CONCLUSIONS: RTX is an effective and well-tolerated treatment in RNMO. BMI could be a predictive factor for efficacy.

Neuromyelitis optica spectrum disorder (NMOSD): A new concept.

The relationship between neuromyelitis optica (NMO) and multiple sclerosis (MS) has long been controversial. NMO was previously considered a form of MS involving predominantly the spinal cord and optic nerve. However, since the discovery of NMO-IgG/aquaporin-4 (AQP4) antibody, an NMO-specific autoantibody to AQP4, some unique clinical features, and magnetic resonance imaging (MRI) and other laboratory findings in NMO, have been further clarified. AQP4 antibody is now the most important laboratory finding for the diagnosis of NMO. Besides typical NMO, some patients with recurrent optic neuritis or recurrent longitudinally extensive transverse myelitis alone are also often positive for AQP4 antibody. Moreover, studies of AQP4 antibody-positive patients have revealed that brain and brainstem lesions are not uncommon in NMO, and some patterns appear to be unique to NMO. All these findings have expanded the NMO concept into 'NMO spectrum disorder' (NMOSD), and new criteria have recently been published. A new antigenic target, myelin oligodendrocyte glycoprotein (MOG), has also been discovered recently. This new antibody seems to correspond to around 20% of seronegative patients, but its specificity needs to be evaluated more precisely, especially in pediatric populations. These recent findings may also have therapeutic impact, as it has been demonstrated that many MS drugs can exacerbate NMO. This report provides an overview of the clinical and neuroimaging features of NMOSD, followed by its treatment.

Chronic inflammatory demyelinating polyradiculoneuropathy: A new animal model for new therapeutic targets.

Animal models are fundamental to advance knowledge of disease pathogenesis and to test/develop new therapeutic strategies. Most of the current knowledge about the pathogenic mechanisms underpinning autoimmune demyelination processes implicating autoantigens has been obtained using the Experimental Autoimmune Neuritis (EAN) animal model. The most widely used EAN model is obtained by active immunization of Lewis rats using a peptide, P0 (180-199), issuing from the major peripheral nervous system myelin protein. But this model mimics only the classical monophasic acute form of demyelinating polyradiculoneuropathy, i.e. Guillain-Barré syndrome (GBS). We developed a new model by immunizing Lewis rats using the same immunodominant neuritogenic peptide P0 (180-199) but this time with its S-palmitoyl derivative, S-palm P0 (180-199). All of the animals immunized with the S-palm P0 (180-199) peptide developed a chronic relapsing-remitting form of the disease corresponding to the electrophysiological criteria of demyelination (slow sensory nerve conduction velocity, prolonged motor nerve latency, partial motor nerve conduction blocks) with axon degeneration. These findings were confirmed by immunohistopathology study and thus, appear to mimic human chronic inflammatory demyelinating polyradiculopathy (CIDP). This new model opens up new avenues of research for testing new anti-inflammatory and neuroprotective therapeutic strategies.

Neuromyelitis optica and multiple sclerosis: Seeing differences through optical coherence tomography.

Neuromyelitis optica (NMO) is an inflammatory autoimmune disease of the central nervous system that preferentially targets the optic nerves and spinal cord. The clinical presentation may suggest multiple sclerosis (MS), but a highly specific serum autoantibody against the astrocytic water channel aquaporin-4 present in up to 80% of NMO patients enables distinction from MS. Optic neuritis may occur in either condition resulting in neuro-anatomical retinal changes. Optical coherence tomography (OCT) has become a useful tool for analyzing retinal damage both in MS and NMO. Numerous studies showed that optic neuritis in NMO typically results in more severe retinal nerve fiber layer (RNFL) and ganglion cell layer thinning and more frequent development of microcystic macular edema than in MS. Furthermore, while patients' RNFL thinning also occurs in the absence of optic neuritis in MS, subclinical damage seems to be rare in NMO. Thus, OCT might be useful in differentiating NMO from MS and serve as an outcome parameter in clinical studies.

Tear analysis as a tool to detect oligoclonal bands in radiologically isolated syndrome.

BACKGROUND: Although radiologically isolated syndrome (RIS) is a newly defined entity, incidental findings of T2 hypersignals on brain MRI can lead to misdiagnosis or useless investigations. The detection of oligoclonal bands (OCBs) in cerebrospinal fluid (CSF) is a major indicator that helps in diagnosis of subclinical inflammatory disease of the central nervous system, but lumbar puncture still remains an invasive option. METHODS: We have prospectively included patients with RIS, have compared the results of CSF and tear OCB detection by isoelectric focusing (IEF) and assessed concordance between OCB detection in tears and in CSF. Tears were collected using a Schirmer strip. RESULTS: In 45 recruited RIS patients, OCBs were detected in CSF for 55% (25/45) and in tears for 50% (21/42) of samples. CONCLUSIONS: We suggest that tear OCB detection may replace CSF OCB detection as a diagnostic tool in patients with RIS and be useful in follow-up.