Marzena Pedrini

Introduction: Timely identification of the transition from relapsing-remitting multiple sclerosis (RRMS) to secondary- progressive MS (SPMS) is of great weight for effective treatment planning. However, this transition is typically diagnosed with an average delay of three years, leading to missed opportunities for early intervention. Building on our previous artificial intelligence (AI) work with Swedish electronic health record data, we extend our approach to the global MSBase cohort. Objectives/Aims: To develop AI models that predict disease progression from RRMS to SPMS proactively in a globally heterogeneous MS population while enabling user-defined confidence levels and interpretable predictions. Methods: We utilized two large-scale MS registries: MSBase (110,000 patients; 1.3 million visits across 45 countries) and the Swedish MS registry (22,000 patients; 200,000 visits). We trained random forest classifiers to predict disease states at each clinical encounter, integrating conformal prediction to quantify predictive uncertainty and explainable AI to enhance interpretability and transparency. Results: The global model achieved an F1 score of 0.83 and outperformed country-specific models in several regions. However, in certain countries, local models were better fitted. Calibration curves revealed marked differences in RRMS and SPMS diagnoses across countries. We identified groups with aligned predictions by clustering countries based on calibration similarity. While the global model generalized well, clustered models improved local accuracy. Conclusion: We developed AI models that provide accurate and interpretable predictions of MS progression trained on registry data from 18 countries. The global model offers scalability, while localized approaches better capture regional diagnostic practices. This framework supports harmonizing diagnostic standards and can enhance clinical trial design and international data interpretation.

Introduction: The study includes the latest findings from the MSBase Registry on the real-world Australian experience with Ofatumumab (OFA) in the treatment of relapsing multiple sclerosis (RMS). Here, we report the latest patient demographics, disease characteristics, and prior therapy history, providing valuable insights into the clinical efficacy and treatment persistence of OFA in both treatment-naive patients and those who have switched from prior therapies. Objectives/Aims: To continue to characterize the use of OFA in Australia through an evaluation of patient demographics, baseline characteristics and prior therapy of patients initiating OFA as recorded in MSBase. In addition, evaluations of clinical efficacy included annual relapse rate and estimates of time to first clinical event. Methods: This is the second analysis of a retrospective, secondary use of data study from the MSBase Registry, describing the baseline characteristics of RMS patients in Australia initiated on OFA treatment. Analyses included demographics, expanded disability status scale (EDSS) and treatment history with diseasemodifying therapies (DMTs). Kaplan Meier estimates were calculated for persistence, time to EDSS milestone and confirmed disability worsening. Results: As of 1st March 2025, MSBase has included 605 Australian patients who received at least one dose of ofatumumab, with a median age of 42.25 (35.86, 50.73) years and 76.7% female. The median disease duration was 8.75 (3.21, 16.43) years, and 22.3% of the participants were treatmentnaive. The most common previous DMTs included ocrelizumab (26.9%) and natalizumab (16.4%). The annualized relapse rate (ARR) during the follow-up period was significantly reduced to 0.05 (0.04, 0.07) compared to 0.34 (0.29, 0.40) in the year prior to starting ofatumumab (p<0.0001). Additionally, the change in EDSS from baseline indicated stable disability scores with median change 0 (-0.5, 0), 0 (-0.5, 0.5) and 0 (-0.5, 0.5) at 12, 18 and 24 months respectively. Kaplan Meier estimates calculated treatment persistence at 96.6% [0.9441, 0.9795] at one year and 93.3% [0.8999, 0.9553] at two years follow-up. Conclusion: The second analysis of the real-world data from the MSBase registry provides further insight into the Australian experience of relapsing MS patients initiated on OFA, with a high treatment persistence and stable disability over 24 months on treatment.

Introduction: The major histocompatibility complex (MHC) locus carries a significant genetic risk burden for MS, though within the MHC the structurally diverse complement component 4 (C4) alleles remain largely understudied. Objectives/Aims: To investigate C4 genetic structural variations in MS risk and better understand how the MHC region shapes disease susceptibility. Methods: We used an established protocol to impute and analyse C4 alleles based on available genotyping data from two case-control cohorts (N1= 3252 cases and 5725 controls; N2= 8978 cases and 6976 controls), a clinical MS cohort (N3= 2387 cases) and a cohort with immune cell expression data (N4= 33 cases and 33 controls). We also performed gene-level analysis to examine the shared genetic landscape between MS onset (NGWAS= 14802 cases and 26703 controls) and plasma C4 protein (NGWAS= 68716). Results: Our data showed that C4 genetic structural variants were associated with significant changes in the risk of development of MS and the progression of MS. For instance, higher C4AL copy number burden was associated with lower risk of MS onset (fixed effect meta-analysis odds ratio= 0.89, P= 5.65×10-6) and reduced hazard of reaching MS disability milestones such as Expanded Disability Status Scale 3 (hazard ratio= 0.79, P= 9.0x10-15). In downstream gene expression analysis, we found C4 alleles may also modulate C4 expression in diseaserelevant immune cell types such as CD8+ T cells. Further, we identified that candidate genes shared between MS onset risk and plasma C4 protein level were enriched in biological pathways of immune regulation, Epstein-Barr virus infection and other autoimmune diseases such as lupus. Conclusion: These findings support future investigations of the C4 genetic structural variants as potential mechanistic and therapeutic targets in MS pathogenesis and disease progression.

Introduction: Serum neurofilament light chain (sNfL) has become a promising biomarker for acute and chronic neuroaxonal damage in relapsing and progressive multiple sclerosis (MS). Evidence is accumulating that sNfL levels have a predictive value for the risk of disability progression in MS. Objectives/Aims: To assess serum neurofilament light chain (sNfL) levels in patients with progressive MS. Methods: We analysed sNfL levels in a cross-sectional cohort study involving 181 patients with progressive MS (PMS). Additionally, we investigated the correlation between disease activity and sNfL levels in 341 patients with relapsing MS (RMS) using single molecule array technology. Results: Serum NfL levels (pg/mL) were significantly elevated in patients with RMS and PMS compared to HC (29.4 vs 12.3, p<0.05; 38.4 vs 12.3, p<0.001, respectively). Additionally, higher sNfL levels were correlated with disability. A statistically significant difference in sNfL levels was observed between relapsing and progressive disease in individuals not receiving treatment (35.2 vs 42.0, p<0.01, adjusted for age). Furthermore, treatment significantly reduced sNfL levels in RMS compared to those not receiving treatment (24.4 vs 36.9, p<0.01, adjusted for age). Notably, a statistically significant difference was observed only with natalizumab and fingolimod, but not with interferons or glatiramer acetate when compared to individuals not receiving treatment. Conclusion: The findings support the utility of sNfL measurement to monitor MS disease activity and progression alongside magnetic resonance imaging.

Introduction: Vascular comorbidities (VCM) are common in patients with multiple sclerosis (MS). VCM are associated with increased disease activity and disability progression, but long-term clinical follow-up data in larger cohorts is limited.

Objectives/Aims: To investigate whether VCM are associated with relapse rates or disability progression in relapsing-remitting MS (RRMS).

Methods: We analysed longitudinal data from MSBase, an international prospective registry of patients with MS. Adults with RRMS, no relapses over the preceding year and no commencement or switching of disease-modifying therapy (DMT) were identified from 2015–2017. Included participants were followed up for 5 years. Baseline VCM (hypertension, hyperlipidaemia, diabetes, ischaemic heart disease) were analysed as individual VCM and as cumulative count of VCM (0/1/2+). Poisson and Cox regression were used to investigate the association between VCM, relapse rates and disability progression over 5 years of follow-up, respectively, with models adjusted for age, sex, disease duration and baseline DMT. As a sensitivity analysis, we analysed a subgroup of 18 sites participating in the MSBase Safety substudy with more complete registry data.

Results: 42185 participants were followed-up over 5 years. The mean age was 44.4 years (SD 12.7), and 72.3% were female. Recorded VCM were hypertension (n=912), diabetes (n=451), hyperlipidaemia (n=325) and ischaemic heart disease (n=186). Median Expanded Disability Status Scale (EDSS) was 2.0 at baseline (IQR 1.0–4.0) and 2.5 at 5 years (IQR 1.5–5.5).

7454 participants (17.7%) had 1 or more relapses over 5 years. Hypertension was associated with increased relapse rate (IRR [95% CI] 1.75 [1.55–1.99]; p<0.001), as were hyperlipidaemia (IRR 1.93 [1.58–2.35]; p<0.001) and diabetes (IRR 1.89 [1.61–2.22]; p<0.001), but not ischaemic heart disease (IRR 1.28 [0.96–1.70]; p=0.09). The relapse rate was increased in those with 1 VCM (IRR 1.81 [1.64–2.00]; p<0.001) and 2 VCM+ (IRR 1.74 [1.36–2.23]; p<0.001), compared to those with no VCM. In the sensitivity analysis, hypertension (IRR 1.20 [1.01–1.42]; p=0.04), diabetes (IRR 1.74 [1.44–2.09]; p<0.001) and 1 VCM (IRR 1.37 [1.20–1.57]; p<0.001) were also associated with increased relapse rates. VCM were not associated with EDSS worsening.

Conclusion: Vascular risk factors are associated with increased disease activity over 5 years in RRMS.

Introduction: Family planning is an integral aspect in the care of women with multiple sclerosis (MS). As disease-modifying therapy (DMT) discontinuation may increase relapse activity, treatment benefits need to be carefully balanced against the risk to the unborn. Ocrelizumab may be a suitable option for women with MS who are planning to conceive, but real-world data on disease activity during pregnancy and postpartum versus other DMTs are scarce.

Objectives/Aims: To describe relapse activity in women with MS before, during and after pregnancy treated with ocrelizumab versus other DMTs.
Methods: Pregnancies in women with MS recorded in MSBase since 2011 up to 1st March 2023 treated with ocrelizumab (OCR), natalizumab (NAT), dimethyl fumarate (DMF) or low-efficacy DMT (interferon-beta, glatiramer acetate) before pregnancy were included. Annualised relapse rates (ARR) were calculated per three-month period for the one-year preconception, during and after pregnancy.

Results: A total of 1654 women with 1901 pregnancies were included (OCR, n=58; NAT, n=398; DMF, n=156; low-efficacy DMT, n=1289). For NAT, use was further subset into continuation into pregnancy third trimester with early (⩽1 month) postpartum re-initiation (NAT-E, n=77)), and cessation prior to four-weeks gestation with late postpartum re-initiation (NAT-L, n=79). One-year preconception ARR [95% CI] was 0.15 [0.06-0.30] for OCR; 0.11 [0.08-0.15] for NAT; 0.12 [0.07-0.19] for DMF; and 0.22 [0.19-0.25] for low-efficacy DMT. During pregnancy, ARR [95% CI] was 0 [0-0.09] for OCR; 0.04 [0-0.13] for NAT-E; 0.34 [0.20-0.52] for NAT-L; 0.12 [0.07-0.21] for DMF; and 0.12 [0.10-0.14] for low-efficacy DMT. Postpartum ARR [95% CI] was 0.11 [0.02-0.33] for OCR; 0.11 [0.03-0.28] for NAT-E; 0.77 [0.52-1.11] for NAT-L; 0.41 [0.28-0.59] for DMF; and 0.43 [0.38-0.49] for low-efficacy DMT. Three women treated with ocrelizumab preconception experienced a postpartum relapse, two of which occurred prior to ocrelizumab re-initiation; none of these women had relapses pre-conception or during pregnancy.

Conclusion: This real-world analysis suggests that women with MS receiving ocrelizumab pre-conception are likely at low risk for within-pregnancy relapse and not at increased risk of postpartum relapses. However, treatment pauses after birth may increase relapse risk. Further data including predictors of postpartum relapses are warranted to ensure optimal care to women with MS and their infants.

Introduction: DNA methylation at CpG sites is an epigenetic mechanism that can influence gene expression. Epigenome-wide association studies (EWASs) of immune cells have implicated differential methylation in Multiple Sclerosis MS. Furthermore, differences in global DNA methylation profiles have been demonstrated after treatment with various disease modifying therapies such as interferon beta and dimethyl fumarate. Whether or not these changes can be used to predict response to treatment at baseline is unknown.
Objectives/Aims: The objective of this study was to assess if there are changes in global DNA methylation that can predict response to cladribine tablets in people with MS (pwMS).
Methods: Whole blood DNA was collected from a subset of the prospectively followed cohort from the CLOBAS study which is a phase IV, multicentre, open label, six-year study of cladribine tablets for pwMS. Participants were considered non-responders if they had a clinical relapse or new MRI activity between three- and 30-months post first dose of cladribine tablets. Bisulfite converted DNA was hybridised to Illumina Infinium EPIC v2 arrays and analysed using the ChAMP R package. GLMNet was used to select meaningful features of baseline DNA methylation that could discriminate between responders and non-responders.
Results: Fifteen of 102 participants were considered non-responders after 30 months of cladribine tablet use. The single most significant feature was cg15041948, which maps to the transcription start site of the GPBP1 gene, which codes for the Vasculin protein. This CpG has a good ability to discriminate between responders and non-responders (Area Under the Curve (AUC) = 0.78). Using the top 28 features identified by GLMNet, the ability to discriminate between responders and non-responders was perfect (AUC=1).
Conclusion: We are able to use global DNA methylation profiles to discriminate between responders and non-responders to cladribine treatment. Low numbers mean results should be interpreted with caution, however, the CLOBAS study is a six-year prospective cohort, therefore, longer term follow studies will follow.