Evaluation of Early Initiation of Disease-Modifying Treatment for Patients with Multiple Sclerosis Within a Real-World Population for Long-Term Outcomes

Abstract

Background: There is varied practice with Disease-Modifying Treatment (DMT) for Multiple Sclerosis worldwide. We evaluated early DMT initiation within a real-world population for long-term outcomes. Method: The Scottish Multiple Sclerosis Register (SMSR) identified participants diagnosed with Relapsing Remitting Multiple Sclerosis (RRMS) in 2010/2011. We compared two groups of propensity-matched participants at diagnosis, who went on to receive either early treatment (<12 months from diagnosis) or late/never treated. Participants underwent detailed clinicoradiological evaluation and patient-reported outcome measures 11–13 years post-diagnosis. The primary outcome was mean Expanded Disability Status Scale (EDSS). Results: The SMSR identified 298 participants. A total of 141 had complete retrospective clinical data and 81 agreed to participate, with 32 successfully matched (16 pairs). Median time on DMT was 10.8 years (range 0.4–12.5) for those treated early and 4.0 (0–11.5) years for the late/never-treated group. A total of 7/16 (44%) never received a DMT of those not treated early. All early-treated participants commenced first-line DMT (5/16 subsequently escalated to second-line DMTs). Of those treated later (9/16), 7/9 participants (78%) commenced first-line and 2/9 s-line DMT. There were no serious adverse events identified with any DMT. There was no significant difference in the primary outcome, with mean EDSS 3.93 in the late/never-treated group vs. 4.53 in the early-treated group at 11–13 years post-diagnosis (p = 0.57). There was no significant difference in median change in EDSS from the time of diagnosis to prospective assessment between early and late/never-treated groups. Patient Reported Outcome Measurement Information System (PROMIS) scores for cognition favoured no early treatment (p = 0.02), whilst satisfaction with treatment choice favoured early treatment (p = 0.03). Conclusions: Our cohort did not show clear benefit with early DMT in RRMS, contrasting with other larger studies, with no significant differences between early and late/never-treated patients on clinicoradiological outcomes. Possible explanations include confounding by variables not included in matching and group allocation based on diagnosis date rather than first clinical symptom. Most participants were treated with injectable DMTs, not in keeping with current practice. A prospective, long-term follow-up deep phenotyping study would help characterise benefits of early DMT use, but this is clearly challenging in practice.

1. Introduction

Multiple Sclerosis (MS) is the most common disease of the central nervous system in young adults and affects between 2 and 3 million people worldwide [1,2], with Scotland having a particularly high incidence [3,4]. Patients are generally diagnosed between 20 and 40 years of age and can accumulate significant disability, resulting in major personal and societal costs [5]. The vast majority of patients have Relapsing Remitting MS (RRMS), whereby neurological disability is interspersed with relative resolution of symptoms between relapses of neurological dysfunction.

Since the introduction of Disease-Modifying Treatments (DMTs) in the 1990s, the landscape of RRMS treatment has changed drastically, but with varied practice in their use: the indications for initiation and switching between therapies and the timing of their use in the course of the disease has been a matter of debate over time [6,7,8]. Relapse reduction and Magnetic Resonance Imaging (MRI) activity have been the primary outcome measures of pivotal clinical trials [9,10], often conducted over a relatively short time period within a lifelong disease. Disability is often a secondary outcome in clinical therapeutic trials in RRMS, as measured by the Expanded Disability Status Scale (EDSS) and, more recently, progression independent of relapse activity (PIRA) [11], but is clearly important to patients and society, along with cognitive function. We therefore wished to capture changes in this on a longitudinal basis in the context of DMT use. Clinical and imaging studies support long-standing pathological data, suggesting that the neurodegenerative component of MS, measured by brain atrophy and cognitive dysfunction, occurs in parallel with clinicoradiological inflammatory disease at the onset of disease [12,13] and is predictive of long-term disability [14]. Observational and long-term extension studies are often flawed by a number of methodological difficulties and cannot fully address the impact of treatment on longer-term impairments. It is therefore less clear what effect early treatment has on long-term disability and subsequent patient quality of life over time.

This study aimed to evaluate whether early DMT initiation would influence long-term clinicoradiological outcomes and patient-reported outcome measures, which represent what really matters to people with MS, using real-world longitudinal data. Our primary outcome compared mean EDSS at prospective assessment between our propensity-matched groups. Secondary outcomes explored safety of DMTs and evaluated detailed objective measures of physical disability and cognitive dysfunction, patient reported functional, and quality of life measures between propensity-matched patients.

2. Materials and Methods

2.1. Study Population

This study was designed to examine data using participants identified from the Scottish MS Register (SMSR). The SMSR is a national register within the Scottish Healthcare Audits programme at the Information Services Division of the National Health Services (NHS) National Services Scotland (NSS). The SMSR includes all patients newly diagnosed with MS in Scotland since 1 January 2010, as defined by Poser (1983) [15] and the McDonald Criteria (2005 [16] and subsequently 2017 [17]). Patients with “possible MS” or clinically isolated syndrome are not included. Data is held in accordance with NSS data protection guidelines and included variables such as demographic information, date of first symptoms, family history of MS, diagnostic categorisation, investigations undertaken, and MS nurse involvement after diagnosis. Case ascertainment in the SMSR is thought to be very high, with around 90% of the true number of patients diagnosed with MS registered. All patients who make contact with the MS services are included, minimising referral bias. Whilst the register itself does not include all clinical data potentially relevant to the patient or proposed research questions, all NHS patients in Scotland have a unique identifier, the Community Health Index (CHI) number. The CHI number, which is used in all community and hospital health records in Scotland, provided the potential for a much broader scope of relevant data access and collection, both clinical and radiological. NHS provides free healthcare, so all patients are offered DMTs, in line with national guidelines and free of charge, ensuring equal access. This study included individuals diagnosed with RRMS from two of the four tertiary MS centres in Scotland, resourced for this study.

2.2. Ethics

This study received ethical approval from the West of Scotland Research Ethics Service (reference 16/WS/0017) and approval from the Integrated Research Applications System (Project ID 195025). Written informed consent was obtained from enrolled participants. The study was registered with on https://clinicaltrials.gov/ (reference NCT05446285) which is a publicly accessible database of clinical trials.

2.3. Study Design

Inclusion criteria were age > 16 years, diagnosed with RRMS in the Glasgow and Grampian regions between 2010 and 2011, and registered on the SMSR. Exclusion criteria included primary progressive MS (PPMS) and contraindications for Magnetic Resonance Imaging (MRI). Only living patients with complete datasets were eligible for invitation to enrol in the study. Retrospective data were gathered from community and hospital records using their CHI number and stored anonymously using a unique patient study code under password-protected worksheets, saved on NHS encrypted devices.

The study was split into two phases—retrospective data collection and prospective clinical assessment. Retrospective data were gathered for propensity matching based on ten co-variates known to influence clinician recommendations to start treatment, with a census date of 8 October 2020: age at diagnosis, sex, ethnicity, duration of disease prior to diagnosis, type of initial relapse, recovery from initial relapse, time from first to second relapse, number of relapses prior to diagnosis, number of T2 lesions on MRI brain at diagnosis, and EDSS at diagnosis. A relapse was defined as the onset of new neurological symptoms or worsening of pre-existing symptoms attributable to demyelinating disease lasting for more than 24 h and preceded by improving or stable neurological status for at least 30 days from the onset of any previous relapse, in the absence of infection, fever or significant metabolic disturbance. Relapse types were divided into motor, brainstem, sensory with associated functional impairment, optic neuritis or intrusive pain. Recovery from relapse was categorised as complete or incomplete. EDSS at diagnosis was estimated from the documented clinical examination at initial consultation with a neurologist when not explicitly stated in clinical notes.

We created two cohorts based on time to DMT initiation. We defined early DMT as commencement within 12 months from diagnosis, and late/never as greater than 12 months from diagnosis. Cohorts were propensity matched using a calliper distance of 0.3 (Figure 1). Prospective evaluation of matched participants occurred 11–13 years after diagnosis.

Prospective individual participant observation was undertaken by a blinded physician at 11–13 years follow up from diagnosis. This included examination and assessment of mobility/disability (EDSS), upper limb function (9-Hole Peg Test, 9-HPT), and cognition (Brief International Cognitive Assessment for Multiple Sclerosis, BICAMS). The BICAMS test comprises three domains: Information Processing Speed (Symbol Digit Modalities Test, SDMT); Verbal Memory (Immediate Recall) (California Verbal Learning Test II, CVLT-II), and Visual Memory (Immediate Recall) (Brief Visuospatial Memory Test Revised, BVMT-R).

Patient-reported outcomes collected at prospective assessment included the Multiple Sclerosis Impact Scale 29 (MSIS-29), Neuro-Quality Of Life (Neuro-QOL) item banks for upper and lower limb function, and Patient-Reported Outcome Measurement Information System (PROMIS) for cognition.

MRI brain sequences were standardised, examining T2 lesion/whole brain volume, undertaken on 3T scanners in both centres, with 1 mm cuts. Local neuroradiologists examined images for incidental findings, but anonymised quantitative analysis was undertaken via IcoBrain software, version 5.10.4 (Icometrix^®, Leuven, Belgium).

Patients were purposely blinded to the aim of the study to minimise potential bias in their responses in the Patient Reported Outcome Measures (PROMs). Once assessments were completed, the assessing physician was unblinded to DMT status; a questionnaire-led clinical history was recorded, and clinical notes subsequently examined to capture any relapses or treatment changes over the 11–13-year study period. This included factors that led to changes such as adverse events or side effects.

2.4. Outcomes

The primary study endpoint was Mean EDSS score at prospective assessment, 11–13 years after diagnosis, comparing those treated with DMT early (<12 months) or late/never. Secondary end points included proportion experiencing (serious) adverse events after diagnosis, proportion attaining EDSS 3.0, 4.0, and 6.0, mean change in EDSS between treatment groups (using estimated EDSS at diagnosis and actual EDSS at the prospective assessment), mean whole brain volume on MRI, mean T2 lesion volume on MRI, mean scores on patient reported outcome measures MSIS-29, Neuro-QOL and PROMIS, mean 9-HPT, and mean BICAMS score.

2.5. Statistical Analysis

Statistical analysis was undertaken by an independent statistician consultant who calculated propensity scores and generated matched pairs. A calliper distance of 0.3 was selected to balance a reduction in bias whilst retaining matched numbers of participants [18].

Non-parametric analysis was performed. Results are shown as mean, median, and interquartile range for continuous variables and as percentages for categorical variables. Significance testing was carried out using the non-parametric Wilcoxon matched pairs test. Effect size was calculated using the z to r transform.

Logistic regression model was used to estimate the likelihood of starting treatment within the first year (propensity score) using the following co-variates listed: age at diagnosis, sex, ethnicity, duration of disease prior to diagnosis, type of initial relapse, recovery from initial relapse, time from first to second relapse, number of relapses prior to diagnosis, number of T2 lesions on MRI brain at diagnosis, and EDSS at diagnosis. These formed the co-variates for calculating each patient’s propensity score with a calliper distance of 0.3. A post hoc power analysis was carried out.

3. Results

The SMSR identified 298 participants, with 141 having complete retrospective clinical data. A total of 81 agreed to participate, with 32 successfully matched (16 pairs) as demonstrated in Figure 1. The baseline clinical and demographic characteristics of our cohort are reported in

Table 1. Baseline characteristics of patients after matching.

	No DMT in First Year Mean (SD) Median [IQR]	DMT in First Year Mean (SD) Median [IQR]	p † (Effect Size r)
Patient numbers	16	16
Age at diagnosis	34.3 (7.07) 33 [11.5]	34.3 (7.07) 33 [11.5]	0.19 (0.23)
Duration of disease *	2.01 (2.05) 1.44 [2.08]	2.20 (2.42) 1.48 [2.13]	0.84 (0.04)
Years between relapse *	1.92 (1.87) 1.13 [3.15]	1.75 (2.05) 1.07 [2.05]	0.57 (0.10)
EDSS score	2.22 (1.59) 1.75 [2.50]	1.69 (1.48) 1.5 [1.0]	0.38 (0.16)
Number of relapses	2.0 (0.63) 2.0 [0.0]	2.19 (0.75) 2.0 [0.75]	0.45 (0.13)
	N (%)	N (%)
SEX (% female)	9 (56%)	11 (69%)	0.69 ‡ (0.17)
Initial relapse symptoms Sensory only Motor Sensorimotor Polysymptoms	9 (56%) 5 (31%) 2 (13%)0	9 (56%) 3 (19%) 3 (19%) 1 (6%)	0.62 (0.09)
Number T2 lesions zero Less than 9 9 or more	2 (13%) 4 (25%) 10 (63%)	2 (13%) 4 (25%) 10 (63%)	1.0
Recovery from 1st relapse Complete Partial None	9 (57%) 1 (6%) 6 (38%)	8 (50%) 3 (19%) 5 (31%)	0.90 (0.02)

DMT, Disease-Modifying Treatment; EDSS, expanded disability severity score. Data are given as mean (SD) and median (IQR, interquartile range) and n (%) * Log transform of variable used in matching; p values and effect size based on log transform. ^† p based on Wilcoxon signed rank test; effect size: Z to r transform. ^‡ McNemar test; effect size is Cohen’s g.

: there was no significant difference between the groups in the measures included.

3.1. Measures of Physical Disability, Cognitive Dysfunction, Patient-Reported Function, and Quality of Life

Table 2. EDSS, 9-HPT, BICAMS, NeuroQol, PROMIS, MSIS-29, and relapses by DMT in first year by group assessed 11–13 years from diagnosis.

	No DMT in First Year (n = 16) Mean (SD) Median [IQR] {Range}	DMT in First Year (n = 16) Mean (SD) Median [IQR] {Range}	p † (Effect Size r)
EDSS at diagnosis	2.22 (1.59) 1.75 [2.5] {0–6}	1.69(1.48) 1.50 [1.00] {0–6}	0.38 (0.16)
EDSS at visit	3.93 (2.41) 3.25 [3.9] {0–7.5}	4.53 (2.00) 4.50 [2.50] {1–9}	0.57 (0.10)
Change in EDSS	1.72 (1.58) 1.50 [2.25] {−0.5–5.0}	2.84 (1.88) 3.0 [3.4] {0–7}	0.09 (0.30)
% EDSS >= 3.0	10 (62.5%)	13 (81.3%)	0.45 ‡ (0.21)
Average 9-HPT *	38.35 (28.14) 23.75 [25.75] {18.75–123}	36.09 (27.19) 28.25 [7.56] {20.75–132.75}	1.0 (0)
Information Processing (SDMT) *	47.00 (18.02) 53.0 [8.0] {0–67}	42.10 (11.52) 42.0 [15.0] {20–61}	0.14 (0.27)
Visual Memory (BVMT-R) *	21.93 (10.07) 26.0 [14.0] {0–31}	24.13 (5.57) 25.0 [8.0] {14–32}	0.86 (0.03)
Verbal Memory (CVLT-II)	14.69 (3.40) 16.0 [0] {6–16}	15.00 (2.07) 16.0 [1] {9–16}	0.89 (0.02)
Lower limb function (NeuroQol)	32.13 (9.99) 37.0 [13.0] {9–40}	29.44 (8.79) 31.5 [12.0] {8–40}	0.41 (0.15)
Upper limb function (NeuroQol)	36.06 (5.36) 39.0 [7.75] {22–40}	33.81 (8.83) 35.0 [6.75] {8–40}	0.55 (0.11)
Cognition (PROMIS)	124.0 (27.56) 132.5 [35.25] {70–164}	93.19 (31.74) 94.0 [40.25] {47–165}	0.02 (0.40)
Quality of life (MSIS-29)	68.0 (26.25) 68.5 [44.5] {31–122}	87.25 (26.33) 89.5 [39.0] {33–139}	0.11 (0.28)
Number of relapses	1.88 (1.50) 1.50 [1.75] {0–5}	2.0 (1.10) 2.50 [1.50] {0–3}	0.75 (0.06)
Annualised relapse rate	0.15 (0.12) 0.12 [0.13] {0–0.41}	0.17 (0.09) 0.20 [0.17] {0–0.27}	0.57 (0.10)
	N (%)	N (%)
Any severe relapses	5 (41%)	7 (44%)	0.73 ‡ (0.13)
Any disabling relapses	8 (50%)	10 (62%)	0.69 ‡ (0.17)
Any incomplete recovery	6 (38%)	10 (62%)	0.29 ‡ (0.25)

DMT, Disease Modifying Treatment; EDSS, expanded disability severity score; 9-HPT, 9-hole peg test; SDMT, Symbol Digit Modalities Test; BVMT-R, Brief Visuospatial Memory Test Revised, BVMT-R; CVLT-II, California Verbal Learning Test II; PROMIS, Patient-Reported Outcome Measurement Information System; MSIS-29, Multiple Sclerosis Impact Scale 29). Data are given as mean (SD) and median (IQR, interquartile range) and n (%). * One person did not complete these tests, so results are based on 15 pairs. ^† p based on Wilcoxon signed rank test; effect size: Z to r transform. ^‡ McNemar test; effect size is Cohen’s g.

shows the mean (SD) and median [IQR] for each group for the outcomes assessed: higher scores reflect worse disability for EDSS, 9-HPT, and MSIS-29; for BICAMS, NeuroQol, and PROMIS, lower scores reflect worse symptoms. For most outcomes, those treated with DMT in the first year had numerically worse scores at follow-up, but there was only a significant difference between groups for cognition in patient-reported outcome (PROMIS, p = 0.02) favouring the late/never treated group, with a moderate effect size (r = 0.4).

There was no significant difference in progression to EDSS 3 between early and late groups (81% in early DMT initiation and 63% in later initiation, p = 0.45).

3.2. Relapses

There was no significant difference between groups for any relapse measure (Table 2).

All but 3 of the 32 cases had a relapse during follow up (2 with no DMT in first year, 1 with DMT in first year). Table 2 compares groups on number and type of relapse. The number of relapses ranged from zero to five, with a median of 1.5 for the No DMT group and 0 to 3, with a median of 2.5, for the DMT group. The annualised relapse rate ranged from 0 to 0.41, with a median of 0.12, for the No DMT group and zero to 0.27 with a median of 0.20 for the DMT group. A total of 5/16 (31%) had any severe relapses in the No DMT group compared to 7/16 (44%) in the DMT group. A total of 8/16 (50%) had any disabling relapses in the No DMT group compared to 10/16 (62%) in the DMT group. Of the 16 pairs, 10 were at the same level for both persons (either both disabling or both not disabling); for 4 pairs, the person with DMT had a disabling relapse, and in 2 pairs, the person without DMT had a disabling relapse. A total of 6/16 (38%) had any incomplete recoveries in the No DMT group compared to 10/16 (44%) in the DMT group. Of the 16 pairs, 8 were at the same level for both persons (either both had an incomplete recovery or both did not have an incomplete recovery); for 6 pairs, the person with DMT had an incomplete recovery, and in 2 pairs, the person without DMT had an incomplete recovery.

3.3. Treatments

First-line treatments during the period of this study included interferons (IFN), glatiramer acetate (GA), dimethyl fumarate (DMF), and teriflunomide. Second-line treatments included natalizumab, fingolimod, ocrelizumab, ofatumumab, and cladribine.

Figure 2 shows the treatment journey for each cohort. All early-treated participants commenced first-line DMT (IFN or GA), with 5/16 (31%) subsequently escalated to second-line DMTs.

Median time on DMT was 10.8 years (range 0.4–12.5) for those treated early and 4.0 (0–11.5) years for the late/never-treated group as shown in

Table 3. Treatments at follow-up by group.

Treatment	No DMT in First yr (n = 16) Mean (SD) Median [IQR] {Range}	DMT in First Year (n = 16) Mean (SD) Median [IQR] {Range}	p † (Effect Size r)
No. of DMTs	0.75 (0.78) 1.0 [1.0] {0–2}	2.81 (0.83) 3.0 [0] {1–4}	0.001 (0.61)
Time on DMT (years)	4.2 (4.4) 4.0 [9.1] {0–11.5}	10.4 (2.9) 10.8 [1.5] {0.4–12.5}	0.001 (0.59)
No. of 1st-line DMTs	0.56 (0.73) 0 [1.0] {0–2}	2.38 (0.96) 3.0 [1.75] {1–4}	0.001 (0.61)
Time on 1st-line DMT (years)	3.1 (3.9) 0 [8.5] {0–10.2}	8.5 (4.0) 10.2 [6.2] {0.4–12.5}	0.002 (0.55)
No. of 2nd-line DMTs	0.19 (0.40) 0 [0] {0–1}	0.44 (0.73) 0 [1] {0–2}	0.27 (0.19)
Time on 2nd-line DMT (days)	402 (966) 0 [0] {0–3428}	684 (1179) 0 [1572] {0–3596}	0.48 (0.12)
Switching 1st- to 2nd-line [n(%)]	3 (18.8%)	5 (31.3%)	0.73 ‡ (0.13)
Happy with treatment [n(%)]	8 (50%)	9 (100%)	0.008 ‡ (0.50)

DMT, Disease-Modifying Treatment. Data are given as mean (SD) and median (IQR, interquartile range) and n (%). p based on Wilcoxon signed rank test; ^† effect size: Z to r transform. ^‡ p based on McNemar test; effect size is Cohen’s g.

. A total of 7/16 (44%) never received a DMT. Of those treated later (9/16), 7/9 participants (78%) commenced first-line (IFN, GA or DMF) and 2/9 (22%) initiated a second-line DMT as their first therapeutic agent. Of those initiated on a first-line agent, one (14%) required escalation to second-line therapy. At time of assessment, 2/16 (13%) of the early DMTs were no longer on a DMT, with 4/9 (44%) of the late-treated group having stopped DMT.

There were no serious adverse events identified with any DMT.

3.4. Radiological Outcomes

MRI assessment calculated FLAIR lesion volume (mL), T1 lesion volume (mL), whole-brain volume (mL), and grey matter volume (mL). Whole-brain and grey matter volume were also expressed as a population percentile based on each patient’s volumes in comparison to age/sex-matched population controls so that the lower the percentile, the worse the patient was in comparison to the normal population.

Table 4. MRI scans at follow-up.

	No DMT in First Year (n = 12) Mean (SD) Median [IQR] {Range}	DMT in First Year (n = 12) Mean (SD) Median [IQR] {Range}	p † (Effect Size r)
FLAIR lesion volume (mL)	7.3 (6.3) 5.55 [11.35] {0.7–18.5}	9.8 (9.5) 6.60 [15.1] {1.7–30.8}	0.53 (0.13)
T1 lesion volume (mL)	5.43 (4.77) 4.0 [9.66] {0.1 –13.1}	7.5 (7.6) 4.7 [11.6] {1.3–24.9}	0.58 (0.11)
Whole-brain volume (mL)	1515 (86.6) 1512 [140] {1346–1647}	1463 (50.0) 1453 [88] {1390–1542}	0.14 (0.30)
Whole-brain volume population percentile	30.2 (26.6) 26.3 [57.7] {1–68}	13.6 (21.3) 2.45 [17.0] {1–68}	0.12 (0.32)
Grey matter volume (mL)	913 (60) 910 [61] {802–1041}	819 (233) 864 [73] {90–959}	0.15 (0.30)
Grey matter volume population percentile	48.7 (28.6) 49.8 [47.3] {1–98.8}	32.4 (35.7) 20.6 [70.7] {1–89.1}	0.39 (0.18)

DMT, Disease Modifying Treatment. Data are given as mean (SD) and median (IQR, interquartile range) and n (%). p based on Wilcoxon signed rank test; ^† effect size: Z to r transform.

shows the median scores by group—four patients did not undergo scans, two from each group, with different pairings providing 12 pairs for comparisons.

There were no significant differences between groups on any MRI outcome criteria. Despite sizable variability in all criteria, as shown by the large IQR, those with no DMT in the first year had better numerical outcomes for all criteria, with a lower median number of lesions (FLAIR and T1), larger median of brain and grey matter volumes, and were at higher median percentiles compared to the population; again, there was no statistically significant difference between the groups.

4. Discussion

This observational study has examined the effectiveness of early DMT initiation using propensity-matched participants identified from the Scottish MS Register in 2020. Detailed prospective clinicoradiological assessment of participants were undertaken, as well as patient-reported outcome measures. Overall, there were no statistically significant differences between the groups aside from patient-reported cognitive function, which favoured no/late treatment, and satisfaction with treatment overall, which favoured early treatment. However, numerically, the no/late treatment group had better outcomes across almost all measures aside from satisfaction with treatment choice. This contradicts larger published studies which demonstrate a beneficial effect from early treatment, as well as worldwide published treatment guidance, including those published from the American Academy of Neurology [19] and the European Committee for Treatment and Research in Multiple Sclerosis and the European Academy of Neurology [20], which advise that patients with MS should be offered treatment as early as possible. Changes in diagnostic criteria for MS have allowed earlier diagnosis with a growing acceptance of treatment benefits and increased earlier treatment in clinical practice. Potential reasons for the unexpected lack of benefit from early DMT in RRMS in our study are considered below.

The definition of early treatment is very heterogeneous. Most define early initiation as within 12 months from first symptom; however, our study examined time to initiation from date of diagnosis. Iaffaldano et al. [21] demonstrated the time interval between disease onset and first DMT as a predictor of disability accumulation independent of relapses over the long term, with risk increasing for those who start treatment after 1.2 yrs from disease onset. A longer prodrome may imply minimal disease burden and therefore influence clinician and patient decision to initiate treatment. Other randomised control trials (RCTs) examining injectable therapies for clinically isolated syndrome (CIS) have shown a significant effect on delaying CIS to clinically definite MS, but long-term benefits on disability accumulation are inconsistent [22,23,24].

Observational prospective studies using electronic databases are more representative of the MS population; however, they are often limited by poor statistical power due to insufficient numbers. They also can have confounders and bias when compared to RCTs, which is particularly challenging to adjust for. However, they can better reflect real-world populations not restricted by inclusion and exclusion criteria. Propensity matching attempts to adjust for variables inherent in the lack of randomisation but is limited in only addressing identifiable factors. Chalmers et al. [25] propensity matched registry patients from their database with those treated in the pivotal RCT. This demonstrated a significantly reduced risk of mortality in the RCT, suggesting further unmeasured cofounders. It is possible that other variables not included in our propensity matching may have led to confounders, resulting in a poorer prognosis for the early treated group, which could have been further compounded by small sample size.

Retrospective analysis is limited by clinical information available in historic clinical notes. Prospectively, we were able to analyse many parameters which were not included in retrospective data and are therefore unable to adjust for differences in baseline.

Our cohort was restricted, with only 32 of 298 eligible participants included. Using a lower calliper can improve robustness when sample size is restricted, so we used a calliper of 0.3 to compensate for this. Small sample sizes are susceptible to skew from outlying participants. A large proportion of eligible participants declined invitation to participate. Participants were not required to provide reasons for declining invitation. It is hypothesised that potential explanations could include working commitments or significant disability preventing travel for prospective assessment, which could skew results. There was a higher proportion of those treated late/never who declined the invitation to participate. Of the deaths, three of the four were in the late/never group, which would have a significant effect on the mean EDSS scores.

With the landscape of therapy advancement during the follow-up period, those treated later were more likely to initiate a moderate to highly effective second-line therapy. Due to the small sample size within the cohort, we were not able to stratify patients based on initial DMT commenced. These newer drugs have been shown to be more effective than injectables with growing evidence and acceptance towards them [26,27,28], initiating highly effective first-line therapy rather than an escalation approach. Brown et al. [29] demonstrated that the more effective DMTs were associated with lower risk of conversion to progressive MS compared to first-line therapies. However, this risk was reduced regardless of if commenced early (<5 years from disease onset) or later [30]. The greater exposure to highly effective treatments may have led to improved outcomes in the late/never-treated cohort rather than the timing of treatment. Our treated cohort were more likely to have relapses detected through increased follow-up encounters and more detailed clinical notes for retrospective examination. This could lead to bias in relapse outcomes in the late/never group. It is possible that unmeasured changes in care over time, with better symptomatic management, lower thresholds for escalating therapies, and more specialist nurses, have affected patient-reported outcome measures.

Whilst there are limitations to this study, there are also strengths. Registry-based studies are thought to be most representative of the true population. Our cohort was assessed prospectively with a considerable follow-up period of 11–13 years of disease duration at time of assessment. We performed extensive clinical and radiological phenotyping, assessing multiple domains considered to be clinically relevant, alongside a focus on patient-centred outcomes, reflecting the real-life experiences of patients with RRMS.

5. Conclusions

We did not find statistically significant differences between two contemporary cohorts of patients with RRMS treated early (within 1 year of diagnosis) or later/never on a wide range of clinicoradiological outcomes assessed 11–13 years after diagnosis. Our cohort did not favour early treatment, contrasting with other larger studies. Possible explanations include confounding by variables not included in propensity matching at diagnosis and group allocation based on diagnosis date rather than first clinical symptom. Small sample size and calliper distance used may have resulted in an unbalanced cohort. Most (early-treated) participants were treated with less-effective injectable DMTs, which is not in keeping with current practice; therefore, results may be different now where higher efficacy DMTs are used more readily. A prospective, deep phenotyping long-term follow-up study would help characterise benefits of early DMT use, but this is clearly challenging in practice.