Multiple Sclerosis Journal

We conducted a systematic review to examine the use of spinal cord (SC) magnetic resonance imaging (MRI) in clinical trials for multiple sclerosis (MS) and discussed the significant challenges that impede its widespread implementation.

We identified 291 phase II and phase III clinical trials. Of these, 228 (78.4%) included at least one MRI outcome. Eight clinical trials (2.7%) were planned to include at least one SC MRI outcome, but only three (1.0%) reported related methodologies or results. The MRI metrics considered in these studies included cervical SC cross-sectional area (1.0%) and SC T₂-lesion number (0.3%).

A very small proportion of clinical trials have used SC MRI metrics as outcome measures. Given the substantial impact of SC disease on disability, it is essential to standardize SC imaging, post-processing, and analytical methods across centres to enhance the effectiveness and applicability of SC MRI in future clinical trials.

Disability accrual in multiple sclerosis (MS) is highly variable and challenging to predict, complicating personalised care. Integrating machine learning (ML) with patient-reported outcomes (PROs) and clinician-assessed outcomes (CAOs) may support tailored interventions.

To develop and validate an interpretable ML model for predicting disability accrual trajectories in MS.

A multicentre data set of 1,176 MS patients with up to 8 years of follow-up was used. A random forest model was trained to predict disability accrual at 2, 3, 4, and 5 years, using baseline clinical variables, PROs and initial risk class as predictors. Model performance was assessed using accuracy, area under the curve, and survival analysis.

437 patients composed final cohort. The model predicted disability changes with an accuracy of 0.82 (95% CI: 0.77–0.86) at 2 years and 0.73 (95% CI: 0.66–0.80) at 5 years. Initial risk class and baseline Expanded Disability Status Scale (EDSS) were the most influential predictors. Survival analysis confirmed model’s ability to effectively capture the time-dependent patterns of disability accrual events at the population level (log rank p > .05).

The model offers robust and interpretable predictions that may support clinical decision-making using routine clinical data.

Identifying long-term outcome substrates in pediatric multiple sclerosis (Ped-MS) can inform treatment selection. We investigated clinical/magnetic resonance imaging (MRI) correlates over a median 12.3-year follow-up.

Fifty-two Ped-MS patients and 23 healthy controls underwent baseline 3.0 T MRI. Neurological evaluations were performed every 6 months. MRI metrics included regional lesion burden, brain and choroid plexus (CP) volumes, diffusion tensor-derived fractional anisotropy (FA), averaged in normal-appearing white matter (NAWM). Multivariable Cox, Andersen–Gill, linear regression models identified long-term outcome correlates.

At follow-up, 69% relapsed, 21% had 6-month confirmed disability worsening (6m-CDW), and 35% showed EDSS worsening. Infratentorial lesion number (hazard ratio (HR) = 1.09, 95% confidence interval (CI) = 1.02 to 1.18), spinal cord lesion presence (HR = 2.45, 95% CI = 1.18 to 5.10), and thalamic volume (HR = 0.77, 95% CI = 0.60 to 0.99) were associated with shorter time to first relapse; high-efficacy treatment (HET) exposure with longer (HR = 0.20, 95% CI = 0.06 to 0.68), and CP volume (HR = 1.58, 95% CI = 0.93 to 2.68) was associated marginally. WM lesion volume (HR = 1.04, 95% CI = 1.01 to 1.06) and HET exposure (HR = 0.21, 95% CI = 0.11 to 0.40) were associated with higher/lower overall relapse risk, respectively; baseline EDSS (HR = 1.29, 95% CI = 0.96 to 1.73), spinal cord lesion presence (HR = 1.52, 95% CI = 0.95 to 2.44), and CP volume (HR = 1.35, 95% CI = 1.00 to 1.83) contributed marginally. Younger age (HR = 0.82, 95% CI = 0.68 to 0.98) and NAWM FA (HR = 0.67, 95% CI = 0.51 to 0.88) were associated with earlier 6m-CDW; NAWM FA with EDSS worsening (β = −0.26, 95% CI = −0.50 to −0.03).

Advanced MRI markers and HET exposure associate with long-term outcomes in Ped-MS.

Characterizing spinal cord multiple sclerosis (MS) lesions in MRI is critical for diagnosis, monitoring, and treatment evaluation. However, current automated approaches for lesion detection and segmentation are typically designed for specific MRI contrasts or acquisition sites, limiting their generalizability in real-world clinical settings where imaging protocols vary widely. This work proposes a robust multi-site, multi-contrast segmentation framework for spinal cord lesions.

The segmentation model was trained and evaluated on a large-scale dataset comprising 4428 annotated images from 1849 persons with MS across 23 imaging centers, encompassing six MRI contrasts (T1w, T2w, T2*w, PSIR, STIR, and UNIT1) acquired at 1.5 tesla (T), 3 T, and 7 T.

Likert-type assessment performed by neuroradiologist ratings demonstrated superior generalization of the model compared to existing contrast-specific pipelines (p < 0.01). Additional experiments evaluated robustness across spinal levels, acquisition resolutions, binarization thresholds, and quantitative evaluation on external labeled datasets.

The proposed model can achieve accurate and reliable spinal cord MS lesion segmentation across heterogeneous MRI data, addressing a key barrier to clinical translation. The model is available in the Spinal Cord Toolbox v7.2 and higher.

Code repository: https://github.com/ivadomed/seg-sc-ms-lesion-multicontrast

The central vein sign (CVS) and the paramagnetic rim lesion (PRL) are neuroimaging biomarkers of multiple sclerosis (MS).

To determine the diagnostic performance of CVS and PRL integration in people presenting for initial diagnostic evaluation of MS.

Adults with clinical/radiological suspicion of MS, aged 18–65, with CVS and PRL assessment from the CentrAl Vein Sign in MS (CAVS-MS) pilot were included. Diagnostic performance of CVS and PRL combinations was evaluated, with 2017 McDonald criteria as the reference standard.

Seventy-eight participants were included (71% female, 86% white, 37 with MS). The combination of ⩾1 CVS and ⩾1 PRL demonstrated sensitivity/specificity of 0.76 (95% CI, 0.59, 0.88) and 0.93 (95% CI, 0.80, 0.98). The combination of ⩾6 CVS and ⩾1 PRL demonstrated sensitivity/specificity of 0.57 (95% CI, 0.39, 0.73) and 1.00 (95% CI, 0.91, 1.00). In those with cerebrospinal fluid testing (n = 46, 24 with MS), ⩾6 CVS and ⩾1 PRL diagnostic specificity was 1.00 (95% CI, 0.85, 1.00), similar to that of oligoclonal bands and dissemination in space by magnetic resonance imaging (0.82 [95% CI, 0.60, 0.95]).

Integrating CVS and PRL represents potentially advantageous MS diagnostic biomarkers, with increased specificity and without substantial reduction in sensitivity.

Robust automated classification of paramagnetic rim lesions (PRLs) and remyelinated lesions based on iron and myelin content in people with multiple sclerosis (pwMS).

In this prospective study (2018–2022), three-dimensional (3D) brain magnetic resonance imaging (MRI) from 180 pwMS (mean age: 47 ± 14 years, 108 females) were acquired at 3T including fluid-attenuated inversion recovery, magnetization-prepared 2 rapid acquisition gradient echoes (MP2RAGE), and T2*-weighted segmented echo-planar imaging. Post-processed quantitative susceptibility mapping (QSM) and filtered phase unwrapped (PU) images were generated. Ground truth was established through expert manual rating. Performance was assessed using nested cross-validation (CV) including outer test sets. Three neural network configurations were evaluated: (1) PRL classification with QSM and MP2RAGE, (2) PRL classification with PU and MP2RAGE, and (3) multiple lesion phenotype (MLP) classification with QSM and MP2RAGE.

For PRL classification, our network achieves a top mean validation F1 score of 0.737 ± 0.027 in the MP2RAGE-QSM configuration trained on PRLs, and a best test performance of 0.709 ± 0.040 when trained on MLP. The MP2RAGE-PU configuration yielded validation and test F1 scores of 0.733 ± 0.021 and 0.662 ± 0.037, respectively. The QSM-based MLP classification achieved a macro F1 test score of 0.728 ± 0.012.

Deep learning can automate the classification of PRLs and QSM lesion phenotypes with high accuracy, potentially facilitating the detection of PRLs (included in the new diagnostic criteria) and remyelinated lesions, guiding personalized treatment decisions in pwMS.

Paramagnetic rim lesions (PRLs) detected on susceptibility-sensitive magnetic resonance imaging (MRI) represent chronic active lesions and correlate with disease progression in multiple sclerosis (MS). However, optimal detection methods remain unclear.

To compare PRL detection using filtered-phase imaging versus susceptibility-weighted imaging (SWI) and assess inter-rater reliability among clinical neuroimmunologists.

We retrospectively analyzed 3T MRI from 44 relapsing-remitting MS patients. Four neurologists with neuroimmunology training received 30-minute training sessions and evaluated PRLs on both filtered-phase and SWI sequences. Consensus adjudication resolved discordant ratings and served as the reference standard for calculating diagnostic accuracy.

PRLs were identified in 21/44 patients (48%). Relative to consensus as the reference standard, SWI demonstrated higher sensitivity than filtered-phase imaging (66.0% vs. 60.4%), while filtered phase showed superior positive predictive value (0.903 vs. 0.714). Both sequences achieved very high specificity (>97%). Inter-rater reliability was good for SWI (Intraclass Correlation Coefficient [ICC] = 0.85) and moderate for filtered phase (ICC = 0.71) for per-subject PRL counts. Vascular complexity was the most common cause of false-positive identification (65% of cases).

Both techniques reliably detect PRLs with substantial inter-rater agreement. Filtered-phase imaging offers superior positive predictive value, while SWI provides better sensitivity and inter-rater reliability. Further development of a standardized training program for assessment of PRL and studies directly comparing susceptibility-sensitive sequences on different vendor platforms are warranted.

Compared to the potent efficacy of autologous hematopoietic cell transplantation (AutoHCT) on multiple sclerosis (MS) lesion activity, benefit on disability progression is less clear. Paramagnetic rim lesions (PRLs) are imaging biomarkers associated with progression.

Evaluate effect of AutoHCT on PRLs using 7T magnetic resonance imaging (MRI).

A single rater evaluated PRLs before/after AutoHCT on 7T MRI at an MS referral center.

Seven participants were included with mean age 38.9 years and median disease duration 9 years. Median PRL count pre-transplantation was 2. Post-AutoHCT (median 15 months in 5 participants) none of the PRLs resolved.

These results provide low-level evidence that PRLs do not resolve following AutoHCT.