1 Introduction

Multiple sclerosis (MS) is a chronic and progressive neuroinflammatory condition affecting the central nervous system, characterised by heterogeneous clinical manifestations and disease course [1]. MS is the most common cause of serious physical disability in adults of working age and can have a profound impact on quality of life and employment [2, 3].

There have been more than fifteen disease modifying therapies (DMTs) approved for the management of relapsing–remitting MS (RRMS), since the approval of Interferons in 1996 [4]. DMTs can be divided into non-high-efficacy therapies (non-HET), with a much better-defined safety profile, and high-efficacy therapies (HET), with potentially higher risk of serious side effects [5].

The choice of the first DMT in treatment-naïve MS patients is a topic of great interest for both patients and clinicians, and whether HET should be preferred to a non-HET remains a topic for debate in the MS community. The recent ABN guidelines have recommended that HET should be considered as the first option in eligible patients [6]. Previous international guidelines recommended HET for highly active MS (ECTRIMS [European Committee for Treatment and Research in Multiple Sclerosis]/EAN [European Academy of Neurology] and American Academy of Neurology [AAN] guidelines) [7, 8].

Real-world cohorts have shown that DMTs significantly reduced the risk of reaching disability milestones in the long-term [9, 10]. When real-world cohorts have been used to compare the efficacy of DMTs, they have rarely included statistical matching techniques, which are necessary to account for systematic differences in baseline characteristics between patients initiating different DMTs when estimating the effectiveness of DMTs [11, 12]. However, when they did use them, there were other shortcomings: (1) They studied only clinical relapses and disability, and not MRI activity, (2) They combined treatment-naïve patients together with patients previously treated with other DMTs and (3) They performed single comparisons of two DMTs [13-16].

Head-to-head randomised clinical trials (RCTs) offer the gold standard design for estimating the effects of treatments. However, they have made only one-to-one comparisons, and included patients switching therapies together with treatment naïve patients; additionally, the trial conditions are not often generalisable and, for example, they do not include patients with multiple co-morbidities. There is evidence from single-arm, open-label studies that treatment-naïve patients maintain low disease activity when they initiate HET, but they have been not compared to other therapies, and are affected by ‘selection bias’ [17].

No evidence of disease activity (NEDA) has been used as a tool for measuring the efficacy of DMTs. NEDA (also known as NEDA3) is achieved if there are no new clinical relapses, new activity on magnetic resonance imaging (MRI) and signs of clinical disease progression, measured by expanded disability status scale (EDSS) over a defined period of time [18]. Although NEDA has its own limitations as a tool to measure the heterogeneity of MS worsening, poor treatment response in the early stages of MS has significant prognostic value [19, 20].

We aimed to compare treatment outcomes (numbers of relapses, disability worsening, MRI activity and NEDA status at 1 and 2 years) between treatment-naïve patients commencing HET versus non-HET, whereas accounting for systematic differences in baseline characteristics of the two groups with propensity score matching.

2 Methods

3 Results

A flow chart of patients is shown in Figure 1. We identified 1986 patients in our whole cohort: 433 on HET and 1553 on non-HET. Table S1 shows the demographics, clinical and radiological characteristics of the whole cohort. We found 853 treatment-naïve RRMS patients who commenced their first DMT: 164 started HET and 689 non-HET. Table S2 shows the demographics, clinical and radiological characteristics of the treatment-naïve cohort. Out of 853 patients, 735 (164 HET and 649 non-HET) were eligible for NEDA assessment at year 1 and 2. Table S3 shows the clinical outcomes of the treatment-naïve cohort at year 1 and 2. This table demonstrates that the HET group has better outcomes than the non-HET group at year 1 and 2. Although ocrelizumab has the larger number of patients, both natalizumab and ocrelizumab have shown similarly better outcomes compared to the non-HET group. The number of patients losing NEDA in the natalizumab and ocrelizumab group was 8.0% and 5.1%, respectively, whereas the number of patients losing NEDA in the glatiramer acetate, dimethyl fumarate and fingolimod group was 24.4%, 15.0% and 41.7%, respectively.

After propensity score matching, we included 448 treatment-naïve patients (110 on HET and 338 on non-HET). Demographics, clinical and radiological characteristics at treatment initiation for this cohort are shown in Table 1. Clinical and radiological outcomes at year 1 and 2 are reported in Table 2.

Table S4 shows the reasons patients stopped treatment early and therefore were not eligible for NEDA assessment. Only patients who initiated non-HET stopped early, with the majority of patients stopping dimethyl fumarate and glatiramer acetate due to side effects.

3.1 Differences in Outcomes Between HET and Non-HET at 1 and 2 Years

When compared with non-HET, the probability of losing NEDA in the HET group was 57% lower at year 1 (HR = 0.43; 95% CI = 0.35, 0.52; p < 0.01), 39% lower at year 2 versus baseline (HR = 0.61; 95% CI = 0.45, 0.84; p < 0.01) and 28% lower between year 1 and year 2 (HR = 0.72; 95% CI = 0.53, 0.97; p = 0.03) (Figure 2).

When looking at the individual NEDA components, the findings were similar. When compared with non-HET, the probability of relapse in the HET group was 94% lower at year 1 (OR = 0.06; 95% CI = 0.01, 0.28; p < 0.01), and 71% lower at year 2 (OR = 0.29; 95% CI = 0.10, 0.84; p = 0.02) (Figure 3). EDSS at year 1 was 30% lower in the HET group (2.43 ± 1.88) (Coeff = −0.30; 95% CI = -0.42, −0.18; p < 0.01) compared with non-HET (2.80 ± 1.87), whereas no difference was detected at year 2 (Coeff = −0.16; 95% CI = -0.34, 0.02; p = 0.09) (Figure 4). The probability of active MRI was 82% lower at year 1 in the HET group (OR = 0.18; 95% CI = 0.04, 0.86; p = 0.03) compared with non-HET. A statistical analysis was not possible at year 2 versus year 1 due to the lack of active MRI in the HET group; in the non-HET group the percentage of patients with MRI activity was 21.7% (Figure 5).

4 Discussion

In this single centre, real-world prospective observational study, we evaluated the short and medium-term treatment response in treatment-naïve RRMS patients who commenced their first therapy with HET or non-HET. Using a propensity score-based matching, we compared RRMS patients with similar demographics and baseline disease characteristics. We found that the probability of relapse, EDSS progression and MRI activity was lower in the HET group compared with the non-HET group. Collectively, the probability of losing NEDA was lower in the HET group compared with the non-HET group. This is due to a more potent effect on disease activity of the high-efficacy DMTs [20].

The main novelties of our study are that we adjusted for possible differences in treatment allocation driven by different clinical characteristics, included several medications used in the real world, and analyzed MRI treatment outcomes. The strength of our study is the well defined study population that reflects routine clinical practice in a Universal Health Care system: MS patients are followed by MS neurologists at a public neuroscience center and have access to all licenced DMTs. We aimed to minimise selection bias by including all consecutive patients and describing reasons for exclusion.

Subgroup analyses and open-label extensions of RCTs on ocrelizumab, cladribine and natalizumab have confirmed their high efficacy in controlling disease activity when initiated early [17, 21-23]. However, single-arm, open-label studies are affected by selection bias, as they recruit participants who have responded (and continue to respond) to the medication. Five real-world retrospective studies, which have investigated non-high-efficacy therapy versus high-efficacy therapy strategies, suggested an association of high-efficacy therapy with a lower risk of clinical relapses, a lower mean change in EDSS score at 5–10 years, a higher median time to sustained accumulation of disability and a lower risk of conversion to secondary progressive MS, compared with starting non-high-efficacy agents [24-29]. However, these studies have lacked MRI data, and considering the baseline MRI features are a crucial prognostic factor to guide the first treatment choice, we included MRI data at baseline as a covariate for the patient propensity score-matching procedure and as a measure of outcome that could trigger treatment escalation.

Some clinicians are advocating the initiation of a non-HET in treatment-naïve patients, with a timely switch to a more potent agent which can control breakthrough disease in the case of an inadequate response to the first-line DMTs, on the basis of the available evidence and international consensus guidelines: (1) The European (ECTRIMS/EAN) guidelines of 2018 suggest deciding MS therapy depends on patient characteristics, disease severity, safety profile and drug accessibility and advise escalating treatment in the case of disease activity while on non-high-efficacy DMT [7]. (2) The AAN guidelines recommend that patients with a highly active disease should be treated with high-efficacy DMT [8]. Neither the AAN nor the European guidelines recommend a specific treatment strategy. (3) The most recent guidelines from the ABN recommend that high-efficacy therapy should be considered as the first option in eligible patients [6].

There are some limitations in our study. No patients on interferons were included in the non-HET group and no patients on cladribine or alemtuzumab were included in the HET group. We aim to include all therapies as we collect data in due course, but therapies such as interferons and alemtuzumab are currently rarely offered. Also, until recently, cladribine could only be prescribed to patients with highly active and rapidly evolving severe MS, and would require a longer follow-up than 2 years. Since fingolimod is a second-line DMT in England, only a small number of patients were included in this study who started fingolimod as a first-line therapy for a variety of reasons including needle phobia and gastrointestinal conditions preventing them from commencing other first-line therapies. Considering studies have shown that fingolimod has a similar efficacy profile to dimethyl fumarate, we have categorised patients starting fingolimod in the non-HET group in our cohort [30, 31]. We have included several therapies in the non-HET group which may have varying degrees of efficacy; however, real-life observational studies have not shown significant differences between dimethyl fumarate and fingolimod and between interferon and glatiramer acetate [16, 32, 33]. The number of patients on each disease modifying drug was too small to allow a more granular comparison between different therapies.

The single-centre design of this study may limit generalisability. However, our cohort is derived from a tertiary referral hospital, with a large catchment area, including a large multi-national and multi-ethnical population. Another limitation is that we evaluated patients over a period of 2 years, which is the common duration of clinical trials, but the long-term effects of treatments will require longer follow-ups; future studies will assess the outcomes of these cohorts in the long-term. This is especially the case for EDSS; we showed EDSS reduction at year 1 in HET (possibly reflecting rapid anti-inflammatory activity) but failed to find any further changes at year 2, with the activity on neurodegenerative aspects of progression requiring longer observation time. EDSS is the most used parameter for the assessment of neurological impairment and disease progression, as it is easily performed and intuitively interpreted by any clinician familiar with the standard neurological examination. However, it is important to note that one of the main limitations of EDSS is that it is regarded as heavily weighted on mobility and not very sensitive to cognitive impairment or upper limb dysfunction [34].

Different DMTs have been offered to patients at different time periods over the last two decades, depending on treatment availability and, therefore, a possible confounder leading to patients' inhomogeneity is the different treatment allocation over time (i.e., ocrelizumab became available in the UK in 2019), which, however, may have affected both groups of patients. Finally, we only included MRI brain analysis and future work will aim to include spinal cord MRI scans, as well as data of other therapies with longer-term outcomes.

The propensity score matching cannot control for unknown confounders (and for any covariate that was not measured and/or not included in propensity score calculation). Therefore, results of two ongoing prospective, multicentre, randomised, pragmatic trials comparing early high-efficacy DMT with escalation approach may provide useful information (the TRaditional versus Early Aggressive Therapy for MS (TREAT-MS, ClinicalTrials.gov no. NCT03500328) and the Determining the Effectiveness of Early Intensive versus Escalation Approaches for the Treatment of RRMS (DELIVER-MS, ClinicalTrials.gov no. NCT03535298)). However, the advantage of real-world cohorts is that they may include patients older than 55 years and with multiple co-morbidities, who are often excluded from trials.

In summary, the analysis of a real-world cohort using propensity score matching showed that the use of highly effective DMTs increases the probability of achieving a status of no disease activity at 1 and 2 years, with no concerns related to safety, suggesting that high-efficacy DMTs should be considered as the first option for treatment-naïve patients.

Author Contributions

Al-Araji had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Al-Araji, Ciccarelli. Acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: Al-Araji, Moccia, Ciccarelli. Critical revision of the manuscript for important intellectual content: Al-Araji, Moccia, Thompson, Barkhof, Ciccarelli. Statistical analysis: Al-Araji, Moccia. Obtained funding: Ciccarelli. Administrative, technical, or material support: Al-Araji, Moccia, Ciccarelli. Supervision: Ciccarelli.

Conflicts of Interest

Sarmad Al-Araji: received sponsorship from Sanofi to attend ECTRIMS and speaker honoraria from Roche. Marcello Moccia: funded by the MUR PNRR Extended Partnership (MNESYS no. PE00000006). In the last 3 years, he has received honoraria from Biogen, BMS Celgene, Ipsen, Merck, Novartis, Roche and Sanofi-Genzyme. Alessia Bianchi has received a research grant from the Italian Society of Neurology and is an ECTRIMS-MAGNIMS Fellow. Charmaine Yam is a Cleveland Clinic London PhD Neuroscience Fellow. Weaam Hamed: Nothing to report. Suraya Mohamud: Nothing to report. Alan J. Thompson: receives an honorarium from SAGE Publishers as Editor-in-Chief of Multiple Sclerosis Journal; has received support from UCL/UCLH NIHR Biomedical Research Centre; acted as co-chair at UCL-Eisai Steering Committee drug discovery collaboration (fees paid to the institution). Ahmed T. Toosy: speaker honoraria from Merck, Biomedia, Sereno Symposia International Foundation, Bayer and At the Limits and meeting expenses from Merck, Biogen Idec and Novartis. He was the UK PI for two clinical trials sponsored by MEDDAY pharmaceutical company (MD1003 in optic neuropathy [MS-ON-NCT02220244] and progressive MS [MS-SPI2—NCT02936037]). He has been supported by recent grants from the MRC (MR/S026088/1), NIHR BRC (541/CAP/OC/818837) and RoseTrees Trust (A1332 and PGL21/10079). He is an associate editor for Frontiers in Neurology—Neuro-ophthalmology section and on the editorial board for Neurology and Multiple Sclerosis Journal. Frederik Barkhof: acts as a member of the steering committee or Data Safety Monitoring Board for Biogen, Merck, ATRI/ACTC and Prothena. Consultant for Roche, Celltrion, Rewind Therapeutics, Merck, IXICO, Jansen and Combinostics. Research agreements with Merck, Biogen, GE Healthcare and Roche. Co-founder and shareholder of Queen Square Analytics LTD. Olga Ciccarelli: NIHR Research Professor (RP-2017-08-ST2-004); over the last 2 years, member of independent DSMB for Novartis; she gave a teaching talk in a Merck local symposium, and contributed to an Advisory Board for Biogen; she is Deputy Editor of Neurology, for which she receives an honorarium; she has received research grant support from the MS Society of Great Britain and Northern Ireland, the NIHR UCLH Biomedical Research Centre, the Rosetree Trust, the National MS Society and the NIHR-HTA.