Systematic literature review and network meta-analysis of cladribine tablets versus alternative disease-modifying treatments for relapsing–remitting multiple sclerosis

Abstract

Objective: To assess the comparative efficacy and safety of cladribine tablets versus alternative disease modifying treatments (DMTs) in patients with active relapsing–remitting multiple sclerosis (RRMS), and in a subgroup with high disease activity (HRA + DAT), using systematic literature review (SLR) and network meta-analysis (NMA).

Methods: MEDLINE, Embase, MEDLINE In-Process and CENTRAL databases were systematically searched to identify English-language publications of relevant studies of approved DMTs for RRMS. Searches were conducted from database inception to January 2017. Conference websites and trial registries were also searched. NMA considered the effects of DMTs on annualized relapse rate (ARR), confirmed disease progression (CDP), no evidence of disease activity (NEDA) and safety.

Results: Of 10,825 articles retrieved and screened, 44 studies assessing 12 DMTs contributed to the NMA. In patients with active RRMS, cladribine tablets were associated with a significant 58% reduction in ARR versus placebo (p < .05); cladribine tablets were similar or significantly better than other DMT regimens and ranked fourth among DMTs, behind alemtuzumab, natalizumab and ocrelizumab. For CDP for 6 months and NEDA, improvements with cladribine tablets were significantly greater than those of placebo (p < .05), with no comparator DMT demonstrating significantly better results. Similar findings were reported in the HRA + DAT population. Overall adverse event risk for cladribine tablets did not differ significantly from that of placebo and most alternative DMTs.

Conclusion: In this first NMA to consider cladribine tablets, ocrelizumab and daclizumab for treatment of RRMS, cladribine tablets are a comparatively effective and safe alternative to other DMTs in both active RRMS and HRA + DAT populations.

Keywords:

• Cladribine tablets
• disease modifying treatments
• relapsing–remitting multiple sclerosis
• high disease activity
• network meta-analysis
• efficacy
• safety

Introduction

Multiple sclerosis (MS) is a chronic, inflammatory, neuro-degenerative, autoimmune disorder of the central nervous system, in which multifocal lymphocytic infiltration causes demyelination and axonal damage. Approximately 85% of patients with MS present with the relapsing–remitting form of the disease (RRMS), which is characterized by periods of remission attenuated by sporadic relapses^1,2. Relapses are unpredictable events characterized by acute symptoms including vision changes, loss of co-ordination, and changes in bowel and bladder function that may stabilize over days or weeks before partially or completely resolving¹.

RRMS can be treated with disease-modifying treatments (DMTs) to suppress autoimmune response^1–3. Treatment options include intramuscular (interferon beta 1a [Avonex]) subcutaneous (interferon betas, glatiramer acetate, and daclizumab), infused (ocrelizumab, alemtuzumab [two 5 and 3 day cycles one year apart] and natalizumab), and oral (e.g. fingolimod, dimethyl fumarate and teriflunomide) treatments. The benefit and harms of DMTs have been studied across a large number of randomized controlled trials (RCTs) that included placebo or interferon beta as a comparator. Whilst these studies provide valuable insight into the use of DMTs in RRMS, alone they fail to address the question of the relative efficacy and safety of different DMTs. With the lack of head-to-head trials between DMTs, indirect comparisons and network meta-analyses (NMAs) have been performed to bridge this evidence gap and help inform clinical decision making^4–7. Over time, these analyses require updating as new evidence emerges, new treatment options gain approval and new outcome measures gain relevance.

In August 2017, the European Medicines Agency (EMA) approved cladribine tablets, an oral, monotherapy, fixed-course regimen comprising two short courses given over two years, as a treatment for adult patients with RMS with highly active disease as defined by clinical or imaging features^8,9. Cladribine tablets were granted approval based on the results of two RCTs that compared cladribine tablets versus placebo (CLARITY)¹⁰ and cladribine tablets plus interferon versus placebo plus interferon (ONWARD)¹¹. In subgroup analyses of the CLARITY data, cladribine tablets showed consistent efficacy in patients with high disease activity (HRA + DAT), defined as patients with ≥2 relapses in the previous year, whether on Disease modifying drug (DMD) treatment or not (high relapse activity; HRA) plus patients with ≥1 relapse and ≥1 T1 Gd + lesion or ≥9 T2 lesions in the previous year while on therapy with other DMDs (disease activity on treatment, DAT)⁹. As with other DMTs, the benefits and harms associated with use of cladribine tablets as an alternative to other treatments for RRMS requires evaluation.

In a RCT setting, the most common primary efficacy endpoint is annualized relapse rate (ARR)². Reduction in relapses alone is, however, not sufficient in daily practice, and in both RCTs and clinical settings, other measures of effect are considered, including brain lesion load from imaging data, and disease progression on the Expanded Disability Status Scale (EDSS). A combined endpoint representing freedom from relapse, progression and lesion load, termed “no evidence of disease activity” (NEDA), has also emerged providing a measure of the efficacy of DMT in stopping all forms of disease activity.

The aim of this analysis was to assess the comparative efficacy and safety of cladribine tablets versus alternative DMTs in patients with RRMS. A systematic literature review (SLR) and NMA were performed using methods similar to those of previous studies. The NMA focuses on the effects of DMTs on ARR and disability progression (confirmed disease progression [CDP]), and also includes the newer endpoint, NEDA, and safety.

Methods

Literature search

A SLR was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (PRISMA), and the requirements of health technology appraisal organizations in the United States, United Kingdom, Germany and France. The search strategy and review protocol are provided in Supplementary Tables 1 and 2, respectively.

MEDLINE, Embase, MEDLINE In-Process and CENTRAL databases were systematically searched using a combination of medical subject headings (MeSH) and free-text terms to identify English-language publications of relevant studies of approved DMTs for RRMS. Searches were conducted from database inception to January 2017 (date of search). Conference abstracts were included and identified via Embase or hand-searches of relevant conference websites. Trial registries (e.g. ClinicalTrials.gov and European Clinical Trial Registry [EUCTR]), the United States Food and Drug Administration (FDA) and EMA websites were also searched. Data for cladribine tablets was extracted from available clinical trial reports, and additional post-hoc analyses of the PRISMS study were performed to fill evidence gaps.

Study selection and data extraction

Citations retrieved through the literature search were screened for inclusion in the review based on the following prospectively defined eligibility criteria: RCTs of DMTs approved by United States or European regulatory agencies in adult patients with RRMS, or a patient population with subgroup of ≥80% RRMS patients. Screening was performed by two independent reviewers in two stages, including an initial abstract and title review followed by full text screening. Two independent reviewers extracted relevant data from the included studies. Any discrepancies between reviewers in screening or data extraction were reconciled by a third independent reviewer.

Assessment of risk of bias and heterogeneity of included trials

The risk of bias was assessed using National Institute for Health and Care Excellence (NICE)¹², German Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen (IQWiG)¹³, and French Haute Autorité de Santé (HAS)¹⁴ checklists, and by study grade (for adequacy of concealment of allocation) and Jadad score (for study quality and study reporting)¹⁵. Sources of clinical heterogeneity were summarized and assessed.

Network meta-analysis

The comparative efficacy and safety of cladribine tablets versus alternative DMTs was assessed via a series of Bayesian network meta-analyses (NMA) based on the methods of Dias et al.¹⁶, and performed in WinBUGS version 1.4.3.

The analysis included both random and fixed-effect models, with results reported for the best fitting model according to residual deviance (closest total residual deviance to the number of data points) and deviance information criterion (DIC) (lowest being best)¹⁶. Each analysis used three independent Monte Carlo Markov chains, each containing a sample of 100,000 simulations after a burn-in of 20,000. Vague prior distributions were assigned to study-level treatment effects, baseline outcomes and between-study heterogeneity parameters. The posterior of the between-trial standard deviation was used to assess between-study heterogeneity. The level of consistency between direct and indirect evidence was tested using the loop-specific approach reported by Veroniki et al.¹⁷.

Separate NMAs were performed on the efficacy endpoints of ARR, CDP sustained for 6 months at 24 months (CDP6M), CDP sustained for 3 months at 24 months (CDP3M), proportion relapse-free at 24 months and NEDA at 24 months. For brevity, results for relapse-free and CDP3M analyses are reported in Supplementary file 1. The safety analysis was performed on the proportion of patients with any adverse events (AEs) as reported in each RCT. Treatment effect estimates comprised relapse rate ratio (RR) for ARR, hazard ratios (HR) for CDP6M and CDP3M, and odds ratios (OR) for NEDA and safety endpoints. The number needed to treat (NNT) and number needed to harm (NNH) comparing DMT versus placebo were also reported. Summaries included posterior effect size, 95% credible intervals, Bayesian p-values, and median rank in line with ISPOR guidelines¹⁸. A Bayesian p-value of <.05 was used to determine statistical significance.

Where feasible, the efficacy analyses were performed on the intent-to-treat (ITT) populations of included studies, and on the subgroup with HRA + DAT. The ITT population was commonly defined as all patients with ≥1 relapse in the previous year or ≥2 relapses in the previous two years (active RRMS). Where data from patients with HRA + DAT was not reported, available subgroup data for patients with HRA or patients with DAT were used and generalized to the full HRA + DAT population. Safety analyses were performed on ITT populations only, as safety data specific to subgroups are rarely reported.

A comprehensive set of sensitivity analyses were performed to assess the impact on results of excluding outlier studies, studies that utilized Poser diagnostic criteria or had an unclear diagnostic criteria, studies published prior to 2000, open-label studies or studies for which blinding status was unclear, and phase II studies (Supplementary file 2). These analyses were supplemented by a meta-regression on baseline risk to explore the impact of between-study heterogeneity on treatment effect. Differences in patient characteristics among trials are likely to reflect heterogeneity in baseline risk, and hence act as a surrogate for underlying factors. Due to data demands, meta-regression analyses were restricted to the active RRMS efficacy analysis.

Results

Systematic literature review

A total of 10,825 articles were retrieved and screened following a search run on 4 January 2017. Forty-nine trials, from 778 publications and two clinical trial reports, met the inclusion criteria and were included in the review. Of these, 44 studies assessing 12 DMTs contributed to the NMA (Figure 1)^19–61. Design details of these studies are summarized in Supplementary Tables 3 and 4.

Figure 1. PRISMA diagram showing the flow of studies through the systematic review process. ¹Five of the 49 studies were excluded because they evaluated unlicensed treatments (n = 2), did not include patients with active RRMS (n = 2) or did not report any outcomes of interest (n = 1). Abbreviations. CSR, clinical study report; HRA + DAT, high disease activity, defined as patients with ≥2 relapses in the previous year, whether on DMD treatment or not (high relapse activity; HRA) plus patients with ≥1 relapse and ≥1 T1 Gd + lesion or ≥9 T2 lesions in the previous year while on therapy with other DMDs (disease activity on treatment, DAT); ITT, intent-to-treat; DMD: disease modifying drug; SGA: Sub-group analysis.

Study and patient characteristics

The 44 studies included in the review were published between 1987 and 2017 and showed variability in diagnostic criteria, blinding, study phase, sample size (31 to 2244), year of publication (Supplementary Table 3; Supplementary Figure 1(a)), and the definitions of “relapse” and disability progression. The majority of studies defined relapse as new or worsening symptoms that lasted at least 24 hours and occurred in the absence of fever or infection. Disability progression was commonly defined as an increase in EDSS of ≥1 point, or ≥0.5 point for baseline EDSS ≥5.5, confirmed during two subsequent neurological examinations separated by an interval of at least 3–6 months free of relapses. The duration of the randomized phase of each study ranged from 12 to 260 weeks; 24 of the 44 studies reported a randomized phase of ≥100 weeks’ duration (Supplementary Table 3). Treatment regimens evaluated are summarized in Table 1, which also reports the regimen abbreviations used in this paper.

Table 1. Treatment regimens evaluated in the network meta-analysis.

Treatment regimen	Route of administration	Abbreviation used
Alemtuzumab 12 mg per day	Intravenous infusion	Alemtuzumab
Cladribine as a cumulative dose of 0.875 mg/kg over a course of 4 or 5 consecutive days of a 28 day period at weeks 1, 5, 48 and 52 resulting in total cladribine dose of 3.5 mg/kg during a treatment period of 96 weeks	Oral	Cladribine tablets
Daclizumab high yield process 150 mg every 4 weeks	Subcutaneous	Daclizumab HYP
Dimethyl fumarate 240 mg twice daily	Oral	DMF
Fingolimod 0.5 mg per day	Oral	Fingolimod
Glatiramer acetate 20 mg per day	Subcutaneous	GA 20 mg
Glatiramer acetate 40 mg three times weekly	Subcutaneous	GA 40 mg
Interferon beta-1a 22 mcg three times weekly	Subcutaneous	IFN beta-1a 22 mcg
Interferon beta-1a 30 mcg once weekly	Intramuscular	IFN beta-1a 30 mcg
Interferon beta-1a 44 mcg three times weekly	Subcutaneous	IFN beta-1a 44 mcg
Interferon beta-1b 250 mcg eod	Subcutaneous	IFN beta-1b
PEG interferon beta-1a 125 mcg q2w	Subcutaneous	PEG IFN beta-1a
Natalizumab 300 mg every 4 weeks	Intravenous infusion	Natalizumab
Ocrelizumab 600 mg once every 6 months	Intravenous infusion	Ocrelizumab
Placebo		Placebo
Teriflunomide 14 mg once daily	Oral	Teriflunomide 14 mg
Teriflunomide 7 mg once daily	Oral	Teriflunomide 7 mg

The majority of included RCTs enrolled adult patients with RRMS who had a baseline EDSS score of ≤6 and had experienced at least one relapse in the preceding year or two relapses in the previous 2 years (Supplementary Table 3). Baseline characteristics were generally comparable across included studies with the notable exception of baseline disease duration, which ranged from 1.2 to 9.1 years (Supplementary Figures 1(b) and (c)).

Risk of bias in the included studies

Across included studies, the risk of bias was generally low according to NICE, IQWiG and HAS checklists (Supplementary File 3). Exceptions were the open-label study of IFN beta-1a 44 mcg versus IFN beta-1b⁴⁹ and studies evaluating alemtuzumab^20–22, which were all single-, assessor-blinded and considered at higher risk of bias. Included studies were generally of good quality in terms of reporting, despite a notable absence of reporting on the clinical significance of study findings.

Network meta-analysis – efficacy

The evidence networks showing the DMT comparisons contributing to the efficacy analyses in patients with active RRMS are presented in Supplementary Figure 2. In total, 41 studies reported data for patients with active RRMS^{19–21,23–29,31–52,54–61} and 11 studies reported data for patients with HRA + DAT^{20–22,28,29,38,39,43,56,58,59}. Of the 11 HRA + DAT studies, only two reported data for the full HRA + DAT population^43,59. Eight of the nine remaining studies reported data from patients with HRA only and one study reported data from patients with DAT only. Overall results of the NMA for clinical endpoints are summarized in Table 2.

Table 2. Summary of efficacy results between cladribine tablets and comparators.

Cladribine tablets versus	ARR	CDP3M 24M	CDP6M 24M	RF 24M	NEDA 24M
Placebo	↑↑	↑↑	↑↑	↑↑	↑↑
Teriflunomide 14 mg od	↑↑	↑	↑	↑↑	↑↑
Teriflunomide 7 mg od	↑↑	↑	↑	↑↑	↑↑
IFN beta-1a 30 mcg q1w	↑↑	↑	↑	↑↑	–
IFN beta-1b 250 mcg eod	↑↑	↑	↓	↑	–
Glatiramer acetate 20 mg qd	↑↑	↑	↑	↑	↑↑
Glatiramer acetate 40 mg tiw	↑↑	–	–	–	–
Fingolimod 0.5 mg qd	↔	↑	↑	↔	–
Alemtuzumab 12 mg qd	↓	↓	↓	↓↓	–
PEG IFN beta-1a 125 mcg q2w	↑↑	–	–	–	–
IFN beta-1a 22 mcg tiw	↑↑	↔	–	↑	–
IFN beta-1a 44 mcg tiw	↑↑	↔	↑	↑	–
DMF 240 mg bid	↑	↔	↑	↑	↑↑
Natalizumab 300 mg q4w	↓	↓	↓	↔	↓
Daclizumab HYP 150 mg q4w	↔	↔	↔	↑	–
Ocrelizumab 600 mg once every 6 months	↓	↓	↓	↓	–

Abbreviations. ↑, better efficacy for cladribine 3.5 mg/kg; ↑↑, statistically significant (p < .05) results in favor of cladribine 3.5 mg/kg; ↓, lower efficacy for cladribine 3.5 mg/kg; ↓↓, statistically significant (p < .05) results in favor of comparator; ↔, equivalent efficacy of cladribine 3.5 mg/kg and comparator; –, analysis was not feasible for these comparisons considering limited evidence; ARR, annualized relapse rate; bid, twice daily; CDP3M 24M; confirmed disease progression sustained for 3 months at 24 months; CDP6M 24M, confirmed disease progression sustained for 6 months at 24 months; DMF, dimethyl fumarate; eod, every other day; HYP, high yield process; IFN, interferon; NEDA 24M, no evidence of disease activity at 24 months; od, once daily; RF 24M, relapse free at 24 months; qd, per day; q1w, once weekly; q2w, every 2 weeks; q4w, every 4 weeks; tiw, three times weekly.

Annualized relapse rate

Forty-one studies contributed to the active RRMS NMA for ARR (the random-effects model was best fit), and 11 studies contributed to the HRA + DAT NMA for ARR (fixed-effects model).

In the active RRMS population, cladribine tablets were associated with a statistically significant reduction (p < .05) in ARR of 58% versus placebo, which translated to a significant reduction of 36–48% when compared with both regimens of teriflunomide, all IFN beta regimens and both GA regimens (Table 3; Figure 2(a)). There were no statistically significant difference in ARR between cladribine tablets and DMF, fingolimod, daclizumab HYP, alemtuzumab, natalizumab or ocrelizumab. The top four DMTs based on rank were alemtuzumab, natalizumab, ocrelizumab and cladribine tablets (Table 3). Results of the NNT analysis were consistent with those of the RR analysis (Table 3). The median NNT to avoid one relapse event (versus placebo) over a 2 year period was 2 for alemtuzumab and natalizumab, 3 for cladribine tablets, ocrelizumab, daclizumab HYP, fingolimod and DMF, and ranged from 5 to 8 for all other DMTs (Table 3).

Figure 2. Summary plot of efficacy risk ratios for cladribine tablets versus comparators. (a) Annualized relapse rate. (b) Confirmed disease progression for 6 months at 24 months.

Table 3. Summary of annualized relapse rate results in the ITT population.

Drug	RR^a	95% CrI	NNT	95% CrI	Rank^b
Alemtuzumab 12 mg qd	1.30	0.93, 1.83	2	2, 3	1
Natalizumab 300 mg q4w	1.22	0.89, 1.68	2	2, 3	2
Ocrelizumab 600 mg once every 6 months	1.14	0.81, 1.6	3	2, 3	3
Cladribine tablets	NA	NA	3	2, 3	4
Daclizumab HYP 150 mg q4w	0.92	0.66, 1.25	3	3, 4	5
Fingolimod 0.5 mg qd	0.91	0.67, 1.22	3	3, 3	6
DMF 240 mg bid	0.78	0.57, 1.07	3	3, 4	7
Glatiramer acetate 20 mg qd	0.64	0.48, 0.85	5	4, 6	8
PEG IFN beta-1a 125 mcg q2w	0.64	0.44, 0.92	5	3, 10	9
IFN beta-1a 44 mcg tiw	0.63	0.47, 0.84	5	4, 6	10
Teriflunomide 14 mg od	0.62	0.46, 0.84	5	4, 7	11
Glatiramer acetate 40 mg tiw	0.62	0.44, 0.87	5	3, 9	12
IFN beta-1b 250 mcg eod	0.62	0.47, 0.83	5	4, 7	13
IFN beta-1a 22 mcg tiw	0.58	0.42, 0.81	5	4, 10	14
Teriflunomide 7 mg od	0.54	0.4, 0.72	7	5, 14	15
IFN beta-1a 30 mcg q1w	0.52	0.39, 0.68	8	6, 14	16
Placebo	0.42	0.32, 0.54	NA	NA	17

Bolded values indicate statistically significant difference from cladribine tablets, as indicated by 95% CrI.

^a Rate ratio versus cladribine tablets; random-effects model.

^b Based on RR.

Abbreviations. bid, twice daily; CrI, credible interval; DMF, dimethyl fumarate; eod, every other day; HYP, high yield process; IFN, interferon; NNT, number needed to treat with active therapy instead of placebo; NA, not applicable; od, once daily; qd, per day; q1w, once weekly; q2w, every 2 weeks; q4w, every 4 weeks; RR; rate ratio (random-effects model); tiw, three times weekly.

In the HRA + DAT population, cladribine tablets were associated with a significant (p < .05) reduction in ARR of 65% versus placebo, which translated to significant reductions in ARR of 49–56% versus teriflunomide 7 mg, IFN beta-1a 30 mcg, IFN beta-1a 44 mcg and GA 20 mg. There was no significant difference in ARR between cladribine tablets and alemtuzumab, and fingolimod, and non-significant but numerical reductions of 34% and 37% versus DMF and teriflunomide 14 mg, respectively. No comparisons were possible versus ocrelizumab and daclizumab HYP due to lack of published subgroup data. Overall, cladribine tablets were ranked second behind natalizumab. The NNT to avoid one relapse event in patients with HRA + DAT were consistent with results in the active RRMS across all DMTs.

Confirmed disease progression sustained for 6 months at 24 months

Twenty studies contributed to the active RRMS NMA for CDP6M at 24 months (random-effects model), and four studies contributed to the HRA + DAT NMA (fixed-effects model).

In patients with active RRMS, cladribine tablets demonstrated improvements in CDP6M that were significantly greater than placebo (HR = 0.54, p < .05) and statistically comparable to other DMTs (Table 4; Figure 2(b)). Cladribine tablets was ranked highest of all other oral DMTs including DMF, fingolimod and both teriflunomide regimens and was ranked amongst the top 50% of all licensed MS therapies alongside alemtuzumab, ocrelizumab, natalizumab and daclizumab HYP. It should be noted that IFN beta-1b was also ranked amongst the top 50% of all licensed MS DMTs; however, data informing this analysis is derived from the INCOMIN study and this result is discussed later. The NNT to avoid one progression event versus placebo over a 2 year period was 11 for cladribine tablets, and ranged from 8 to 22 for other DMTs (Table 4).

Table 4. Summary of confirmed disease progression sustained for 6 months at 24 months results in the ITT population.

Drug	HR^a	95% CrI	NNT	95% CrI	Rank^b
IFN beta-1b 250 mcg eod	1.79	0.65, 4.73	8	6, 17	1
Alemtuzumab 12 mg qd	1.37	0.58, 3.32	9	6, 19	2
Ocrelizumab 600 mg once every 6 months	1.26	0.51, 2.98	9	7, 26	3
Natalizumab 300 mg q4w	1.21	0.52, 2.77	10	7, 24	4
Daclizumab HYP, 150 mg, q4w	1.07	0.42, 2.65	11	NNTH 11 to ∞ to NNTB 46	5
Cladribine tablets	NA	NA	11	6, 49	6
DMF 240 mg bid	0.85	0.41, 1.81	15	8, 60	7
Glatiramer acetate 20 mg qd	0.81	0.37, 1.73	16	NNTH 48 to ∞ to NNTB 99	8
Fingolimod 0.5 mg qd	0.79	0.37, 1.64	17	NNTH 36 to ∞ to NNTB 89	9
IFN beta-1a 30 mcg q1w	0.79	0.37, 1.64	17	NNTH 50 to ∞ to NNTB 103	10
IFN beta-1a 44 mcg tiw	0.76	0.35, 1.61	18	NNTH 80 to ∞ to NNTB 122	11
Teriflunomide 14 mg qd	0.66	0.31, 1.38	23	NNTH 233 to ∞ to NNTB 272	12
Teriflunomide 7 mg qd	0.57	0.27, 1.18	22	NNTH 465 to ∞ to NNTB 490	13
Placebo	0.54	0.29, 0.99	NA	NA	14

Bolded values indicate statistically significant difference from cladribine tablets, as indicated by 95% CrI.

^a Hazard ratio versus cladribine tablets; random-effects model.

^b Based on HR.

Abbreviations. bid, twice daily; CrI, credible interval; DMF, dimethyl fumarate; eod, every other day; HR, hazard ratio (random-effects model); HYP, high yield process; IFN, interferon; NNT, number needed to treat with active therapy instead of placebo; NNTB, NNT to benefit; NNTH, NNT to harm; NA, not applicable; od, once daily; qd, per day; q1w, once weekly; q2w, every 2 weeks; q4w, every 4 weeks; tiw, three times weekly.

In patients with HRA + DAT, cladribine tablets were associated with a statistically significant 82% reduction in the rate of CDP6M versus placebo, translating to a 61–68% reduction in CDP6M versus alemtuzumab and IFN-beta 1a 44 mcg (p < .05). Cladribine tablets were ranked first ahead of alemtuzumab and IFN-beta 1a 44 mcg. No other DMT comparisons were possible in patients with HRA + DAT due to a lack of published subgroup data.

No evidence of disease activity at 24 months

Five studies contributed to the active RRMS NMA of NEDA at 24 months (fixed-effects model). No NMA was performed for patients with HRA + DAT as only the CLARITY study reported NEDA in this population (cladribine tablets versus placebo).

In patients with active RRMS, cladribine tablets were associated with significantly increased odds of NEDA versus placebo (OR = 4.69) that translated to significant benefits for cladribine tablets versus teriflunomide 14 mg/7 mg (OR = 2.00/2.84), DMF (OR = 2.72) and GA 20 mg (3.39) (Table 5). There was a non-significant but numerically increased odds of NEDA with natalizumab versus cladribine tablets (OR = 0.64). No other DMT comparisons were possible due to lack of published data. Cladribine tablets were ranked second behind natalizumab. The NNT for an additional NEDA versus placebo was 2 for natalizumab, 3 for cladribine tablets, and ranged from 6 to 19 for other DMTs (Table 5).

Table 5. Summary of no evidence of disease activity at 24 months results in the ITT population.

Drug	OR^a	95% CrI	NNT	95% CrI	Rank^b
Natalizumab 300 mg q4w	0.64	0.35, 1.12	2	2, 3	1
Cladribine tablets	NA	NA	3	2, 4	2
Teriflunomide 14 mg od	2.00	1.2, 3.32	6	4, 12	3
DMF 240 mg bid	2.72	1.77, 4.21	11	7, 22	4
Teriflunomide 7 mg od	2.84	1.7, 4.76	12	6, 49	5
Glatiramer acetate 20 mg qd	3.39	2.17, 5.35	19	9, 106	6
Placebo	4.69	3.35, 6.65	NA	NA	7

Bolded values indicate statistically significant difference from cladribine tablets, as indicated by 95% CrI.

^a Odds ratio versus cladribine tablets.

^b Based on OR.

Abbreviations. bid, twice daily; CrI, credible interval; DMF, dimethyl fumarate; NNT, number needed to treat with active therapy instead of placebo; NA, not applicable; od, once daily; OR, odds ratio (fixed-effects model); qd, per day; q4w, every 4 weeks.

Sensitivity analyses

Results of the sensitivity analyses are presented in Supplementary file 2. There was no change in the direction of relative treatment differences between cladribine tablets and comparators, although in some instances the significance level of findings changed. The results of the meta-regression on baseline risk were in line with the NMA suggesting that variations in baseline risk had a limited overall effect on results. Sensitivity analyses were not conducted for NEDA at 24 months because insufficient studies were included in the network.

Consistency between direct and indirect evidence was assessed for ARR and CDP6M, with results suggesting that evidence were generally consistent within each network. The exception being the ARR comparisons of placebo versus IFN beta-1a 44 mcg versus teriflunomide 7 mg. Consistency testing was not feasible for NEDA at 24 months due to a three arm trial contribution in the closed loop.

Network meta-analysis – safety, all active relapsing–remitting multiple sclerosis populations

Twenty-five studies contributed to the NMA (random) for any AE (Supplementary Figure 3). The definition of any AE varied across studies with differences in cut-off limits (e.g. >10% of population) and the inclusion of relapse due to multiple sclerosis as an AE. A summary of the results of the safety NMA is provided in Table 6.

Table 6. Summary of adverse events results in the ITT population.

Drug	OR^a	95% CrI	NNT	95% CrI	Rank^b
Natalizumab 300 mg q4w	2.70	0.96, 7.93	11	NNH 107 to ∞ to NNB 135	1
Glatiramer acetate 20 mg qd	2.23	0.89, 5.69	18	NNH 169 to ∞ to NNB 210	2
Placebo	1.59	0.76, 3.34	NA	NA	3
Daclizumab HYP 150 mg q4w	1.53	0.62, 4.16	NNH 21	NNH 510 to ∞ to NNB 485	4
Teriflunomide 7 mg od	1.38	0.55, 3.43	NNH 29	NNH 476 to ∞ to NNB 456	5
Fingolimod 0.5 mg qd	1.31	0.52, 3.06	NNH 29	NNH 416 to ∞ to NNB 371	6
Teriflunomide 14 mg od	1.21	0.48, 3.05	NNH 27	NNH 302 to ∞ to NNB 258	7
IFN beta-1a 30 mcg q1w	1.13	0.49, 2.81	NNH 25	NNH 201 to ∞ to NNB 141	8
DMF 240 mg bid	1.02	0.39, 2.7	NNH 20	NNH 169 to ∞ to NNB 122	9
IFN beta-1b 250 mcg eod	0.81	0.04, 15.52	NNH 9	NNH 94 to ∞ to NNB 84	10
Cladribine tablets	NA	NA	NNH 19	NNH 154 to ∞ to NNB 111	11
Glatiramer acetate 40 mg tiw	0.98	0.35, 2.75	NNH 19	NNH 135 to ∞ to NNB 86	12
Ocrelizumab 600 mg once every 6 months	0.67	0.12, 3.31	NNH 12	NNH 104 to ∞ to NNB 79	13
IFN beta-1a 44 mcg tiw	0.67	0.13, 3.03	NNH 12	NNH 102 to ∞ to NNB 78	14
PEG IFN beta-1a 125 mcg q2w	0.51	0.17, 1.51	NNH 11	NNH 25 to ∞ to NNH 8	15
Alemtuzumab 12 mg qd	0.27	0.05, 1.47	NNH 9	NNH 26 to ∞ to NNH 7	16

^a Odds ratio versus cladribine tablets; OR >1 indicates lower risk of adverse events versus cladribine tablets and OR <1 indicates greater risk.

^b Based on OR.

Abbreviations. bid, twice daily; CrI, credible interval; DMF, dimethyl fumarate; eod, every other day; HYP, high yield process; IFN, interferon; NNB, number needed to benefit with active therapy instead of placebo; NNH, number needed to harm with active therapy instead of placebo; NNT, number needed to treat with active therapy instead of placebo; NA, not applicable; od, once daily; OR; odds ratio (random-effects model); qd, per day; q1w, once weekly; q2w, every 2 weeks; q4w, every 4 weeks; tiw, three times weekly.

The results indicate no statistically significant difference in overall AE risk for cladribine tablets versus placebo and most alternative DMTs. The overall AE risk of cladribine tablets was statistically comparable to other oral DMTs, all IFN-beta regimens and GA 40 mg. There was a numerical but non-significant increased overall AE risk with alemtuzumab (OR = 0.27), peg IFN-beta (OR = 0.51), and ocrelizumab versus cladribine tablets (OR = 0.67). The overall AE risk for cladribine tablets was numerically increased versus natalizumab (OR = 2.70) and GA 20 mg (OR = 2.23). When ranked according to lowest overall AE risk, cladribine tablets were eighth and ahead of alemtuzumab (sixteenth). The NNH for DMTs versus placebo ranged from 9 (alemtuzumab) to 29 (teriflunomide 7 mg and fingolimod), with natalizumab and GA 20 mg being associated with a reduced AE risk versus placebo and NNT to avoid any AE of 11 and 18 respectively. The NNH for cladribine tablets was 19, compared with 20 for DMF, and 29 for fingolimod and teriflunomide 7 mg.

Discussion

The results of the SLR and NMA indicate that cladribine tablets are a relatively effective and safe alternative to other DMTs in both active RRMS and HRA + DAT populations. The active RRMS analysis suggests cladribine tablets are amongst the most effective oral DMTs and have an overall AE risk profile that is not significantly different from alternative DMTs. In this population, infusion therapies including alemtuzumab and ocrelizumab were associated with numerical but non-significant benefits on ARR and CDP6M versus cladribine tablets, with a higher overall AE risk. In patients with HRA + DAT, the benefits of cladribine tablets versus placebo from CLARITY translated to an increased effect versus all other DMTs, with cladribine tablets having efficacy greater than or comparable to alemtuzumab and natalizumab. These analyses demonstrate cladribine tablets as an alternative to other DMTs in an active RRMS population, and further highlight the value of this treatment in those with high unmet need, such as the HRA + DAT population. These results are representative of the clinical trial evidence for cladribine tablets.

The choice of DMT depends on numerous factors including tolerability and safety profile, patient preference, ease of administration and efficacy. Cladribine tablets are the first oral therapy to be given as a fixed course treatment, providing persons with RRMS an alternative treatment approach without daily therapy or infusions. This is the first review and NMA that considers cladribine tablets in comparison with other DMTs, and it combines both efficacy and safety endpoints to inform clinical decisions on the use of this new treatment option. A novel aspect of this analysis was the consideration of NNT and NNH as a way of comparing the harms and benefits of different DMTs. A positive benefit–risk profile is characterized by a lower NNT than NNH, with greater differences in NNT and NNH indicating improved benefit–risk profile⁶². As expected, the NNT for relapse and progression for all DMTs including cladribine tablets was consistently lower than or equal to its corresponding NNH, supporting a positive benefit–risk profile for DMT. In general, the benefit–risk profile of cladribine tablets was comparable or favorable to alternative DMTs. For disease progression, the benefit risk of cladribine tablets (NNT/NNH: 11/19) in patients with active RRMS appeared to be more favorable than the numerically more effective strategies of alemtuzumab (NNT/NNH: 9/9) and ocrelizumab (NNT/NNH: 9/12). The AE risk profile differs across DMTs and hence consideration of the individual risk is an important aspect of treatment decision-making.

This is the first NMA to consider the endpoint of NEDA, which represents no activity on EDSS, relapse or lesion imaging. This endpoint has not been routinely included as part of RCTs and hence there is limited data currently available. Of the available evidence, cladribine tablets were significantly more effective than other oral DMTs (DMF and teriflunomide) and of comparable efficacy to natalizumab. These analyses should be revisited when further evidence on the effect of DMTs such as alemtuzumab, fingolimod, ocrelizumab and daclizumab on NEDA becomes available. This is particularly relevant as the goal of treatment in RRMS increasingly moves towards inducing no disease activity, and endpoints such as NEDA become increasingly relevant benchmarks of treatment effect.

The results of our NMA are comparable to a previous analysis⁶ in demonstrating that, on average, the most effective treatments according to CDP and ARR are infusion-based therapies, followed by oral treatments and injectable therapies. No previous NMA have included cladribine tablets, and none considered recently published evidence for ocrelizumab and daclizumab. The present NMA also includes unpublished post-hoc analysis of CDP6M for interferon beta-1a versus placebo from PRISMS. This analysis was performed to improve the level of evidence available for CDP6M in both active RRMS and HRA + DAT populations, and to improve the evidence connecting alemtuzumab and ocrelizumab, both of which were studied versus interferon beta-1a, to the rest of the network.

As in Fogarty et al.⁶, IFN beta-1b was found to be numerically superior to all other DMTs for CDP6M at 24 months, while the results for CDP3M indicated that cladribine tablets were better than IFN beta-1b (Supplementary file 1). The variation in our results might be due to inclusion in the CDP6M analyses of the INCOMIN trial³³, which reported a statistically significantly improvement in CDP for IFN beta-1b compared with IFN beta-1a (RR [95% CI]: 0.44 [0.25–0.80]). The improvement in CDP6M for IFN beta-1b reported in a single study by Durelli et al.³³ is in contrast to results seen for improvements in CDP3M for IFN beta-1b reported in two other studies that show comparable efficacy for IFN beta-1b versus placebo and glatiramer acetate^40,45.

There are limitations associated with this current review. As with all NMA, these analyses relied on the similarity of studies to generate exchangeable treatment effects. The trials included in this NMA were conducted over a 30 year period (1987 to 2017), during which time the treatment algorithm, as well as the diagnosis criteria, for RRMS has changed; this will impact the generalizability of our results⁶³. The trials also differed in terms of study characteristics (diagnostic criteria, study phase and blinding), the patient populations recruited (mean relapses in prior year, disease duration, treatment history – previously treated versus treatment naïve) and definitions of outcomes. The NMA for patients with HRA + DAT is also limited by variations in the definition of HRA + DAT across studies, a lack of subgroup-specific baseline characteristic data to assess potential for effect modifiers and uncertainty surrounding effect sizes, as highlighted in Huisman et al.⁷.

The safety findings must be interpreted with caution. The intention was to provide an overall AE risk profile using evidence available from RCTs. The reporting of any AE varied across studies leading to heterogeneity and potential bias in the analysis. This included the consideration of relapse of multiple sclerosis as an AE. In the AFFIRM study²⁹, this resulted in a higher overall risk of AE with placebo than active treatment, leading to natalizumab appearing safer than other DMTs in the NMA. Additionally, not all relevant AEs are captured in RCTs, such as the cases of progressive multifocal leukoencephalopathy (PML) with natalizumab, DMF and fingolimod reported from post-surveillance monitoring. Further, the overall “any AE” risk category does not differentiate among the individual AEs associated with each DMT, nor consider the effect of different AEs on overall patient wellbeing. Hence, when considering the potential for harm with DMTs, further information on the risk profile of each DMT should be taken into account alongside its overall risk profile.

Conclusions

This is the first NMA to consider recently approved treatments for RRMS, cladribine tablets, ocrelizumab and daclizumab. The results of this analysis show that cladribine tablets are a comparatively effective and safe alternative to other DMTs in both the active RRMS and HRA + DAT populations.

Transparency

Declaration of funding

This work was supported by Merck KGaA, Darmstadt, Germany.

Author contributions: All authors were involved in the conception and design of the study. All authors provided analysis and interpretation of the data, critically revised the manuscript for intellectual content and provided final approval of the submitted version.

Declaration of financial/other relationships

M.K.S. has disclosed that she/he is an employee of Parexel International, Chandigarh, India. I.S.K. has disclosed that she/he is an employee of Parexel International, Chandigarh, India. S.B. has disclosed that she/he is an employee of Parexel International, London, UK. R.H. has disclosed that she/he has no significant relationships with or financial interests in any commercial companies related to this study or article. G.H. has disclosed that she/he is an employee of Merck KGaA, Middlesex, UK. S.L.W. has disclosed that she/he is an employee of EMD Serono Ltd., Billerica, MA, USA and is also a stock shareholder of EMD Serono, a business of Merck KGaA, Darmstadt, Germany.

CMRO peer reviewers on this manuscript have received an honorarium from CMRO for their review work, but have no relevant financial or other relationships to disclose.

Acknowledgements

Medical writing support was provided by Mark OConnor and Caroline Spencer of inScience Communications, Springer Healthcare, London, UK, and was supported by Merck KGaA, Darmstadt, Germany.