Multiple Sclerosis and Related Disorders 9 (2016) 23–30

Contents lists available at ScienceDirect

Multiple Sclerosis and Related Disorders

journal homepage: www.elsevier.com/locate/msard

Comparative efficacy of disease-modifying therapies for patients

with relapsing remitting multiple sclerosis: Systematic review
and network meta-analysis
Emer Fogarty a,n, Susanne Schmitz b, Niall Tubridy c, Cathal Walsh d, Michael Barry e
a
National Centre for Pharmacoeconomics, Dublin, Ireland
b
Health Economics and Evidence Synthesis Research Unit, Department of Population Health, Luxembourg Institute of Health, Luxembourg
c
Department of Neurology, St. Vincent's University Hospital, Dublin, Ireland
d
Department of Mathematics and Statistics, University of Limerick, Ireland
e
National Centre for Pharmacoeconomics, Dublin, Ireland

art ic l e i nf o a b s t r a c t

Article history: Introduction: Randomised studies have demonstrated efficacy of disease-modifying therapies in relap-
Received 24 April 2015 sing remitting multiple sclerosis (RRMS). However it is unclear how the magnitude of treatment efficacy
Received in revised form varies across all currently available therapies.
29 April 2016
Objective: To perform a systematic review and network meta-analysis to evaluate the comparative ef-
Accepted 7 June 2016
ficacy of available therapies in reducing relapses and disability progression in RRMS.
Methods: A systematic review identified 28 randomised, placebo-controlled and direct comparative
Keywords: trials. A network meta-analysis was conducted within a Bayesian framework to estimate comparative
Network meta-analysis annualised relapse rates (ARR) and risks of disability progression (defined by both a 3-month, and
Systematic review
6-month confirmation interval). Potential sources of treatment-effect modification from study-level
Disease-modifying therapy
covariates and baseline risk were evaluated through meta-regression methods. The Surface Under the
Multiple sclerosis
Disability progression Cumulative RAnking curve (SUCRA) method was used to provide a ranking of treatments for each out-
Relapse come.
Results: The magnitude of ARR reduction varied between 15–36% for all interferon-beta products, gla-
tiramer acetate and teriflunomide, and from 50 to 69% for alemtuzumab, dimethyl fumarate, fingolimod
and natalizumab. The risk of disability progression (3-month) was reduced by 19–28% with interferon-
beta products, glatiramer acetate, fingolimod and teriflunomide, by 38–45% for pegylated interferon-
beta, dimethyl fumarate and natalizumab and by 68% with alemtuzumab. Broadly similar estimates for
the risk of disability progression (6-month), with the exception of interferon-beta-1b 250 mcg which was
much more efficacious based on this definition. Alemtuzumab and natalizumab had the highest SUCRA
scores (97% and 95% respectively) for ARR, while ranking for disability progression varied depending on
the definition of the outcome. Interferon-beta-1b 250 mcg ranked among the most efficacious treat-
ments for disability progression confirmed after six months (92%) and among the least efficacious when
the outcome was confirmed after three months (30%). No significant modification of relative treatment
effects was identified from study-level covariates or baseline risk.
Conclusion: Compared with placebo, clear reductions in ARR with disease-modifying therapies were
accompanied by more uncertain changes in disability progression. The magnitude of the reduction and
the uncertainty associated with treatment effects varied between DMTs. While natalizumab and alem-
tuzumab demonstrated consistently high ranking across outcomes, with older interferon-beta and gla-
tiramer acetate products ranking lowest, variation in disability progression definitions lead to variation
in the relative ranking of treatments. Rigorously conducted comparative studies are required to fully
evaluate the comparative treatment effects of disease modifying therapies for RRMS.
& 2016 Elsevier B.V. All rights reserved.

n
Corresponding author.
E-mail address: efogarty@stjames.ie (E. Fogarty).

http://dx.doi.org/10.1016/j.msard.2016.06.001
2211-0348/& 2016 Elsevier B.V. All rights reserved.
24 E. Fogarty et al. / Multiple Sclerosis and Related Disorders 9 (2016) 23–30
1. Introduction
Multiple sclerosis (MS) is a chronic, autoimmune, disabling
disease of the central nervous system. Relapses and disability
progression are the clinical hallmarks of MS and the two most
commonly assessed clinical endpoints used to evaluate ther-
apeutic interventions in clinical trials (European Medicines Agency
(EMA) Committee for Medicinal Products for Human Use (CHMP),
2006). Relapses represent focal, acute, recurrent inflammation,
and are characterised by a gradual onset of symptoms which sta-
bilise over days or weeks and resolve gradually, either completely
or partially (Coles, 2009). Progression refers to the steady and ir-
reversible worsening of symptoms and signs independent of the
occurrence of relapses resulting from diffuse, early, chronic, pro-
gressive neurodegeneration (Confavreux and Vukusic, 2006a). The
efficacy of disease-modifying therapies (DMTs) on these clinical
endpoints has been demonstrated in clinical trials over the last
two decades, however few trials have assessed the comparative
effectiveness of DMTs. We conducted a systematic review and
Bayesian network meta-analysis considering all placebo-con-
trolled and comparative trials of DMTs, to evaluate their com-
parative efficacy in reducing the risk of relapse and disability
progression in MS.
2. Methods
2.1. Literature search and study selection
A systematic search of the published literature was conducted
from inception to March 2016 to identify eligible studies using
EMBASE, MEDLINE (via PubMed) and CENTRAL (via Cochrane Li-
brary) databases. In addition, the Food and Drug Administration
(FDA) and European Medicines Agency (EMA) websites were
searched for reviews and assessment reports on interventions of
interest. For inclusion in the analysis, randomised controlled trials
(RCTs) were required to include adult patients with relapsing re-
mitting MS (RRMS), report at least one of the primary outcome
measures of interest, and compare DMTs which are authorised by
US and/or European regulatory agencies for the treatment of
RRMS. Eleven DMTs met this criterion including interferon beta-1b
(IFN β-1b) subcutaneous (SC) 250 mcg, IFN β-1a SC 22 mcg and
IFN β-1a SC 44 mcg, IFN β-1a intramuscular (IM) 30 mcg, pegy-
lated IFN β-1a SC 125 mcg, glatiramer acetate 20 mg, glatiramer
acetate 40 mg, natalizumab, alemtuzumab, fingolimod, teri-
flunomide, and dimethyl fumarate. The Preferred Reporting Items
for Systematic Reviews and Meta-Analyses (PRISMA) statement
was followed for our network-meta analysis protocol (Moher et al.,
2009). The full search strategy, inclusion and exclusion criteria are
detailed in Additional File 1.
2.2. Outcome measures
Annualised relapse rate (ARR) and confirmed disability pro-
gression were selected as the most commonly reported clinical
outcomes in RRMS trials. ARR is defined as the mean number of
confirmed relapses per patient adjusted for the duration of follow-
up. Definitions of relapse and disability progression were as per
the individual trial(s), and are detailed in Additional File 2. The
definition of ‘relapse’ was subject to slight variation across trials
but it was commonly defined as new or worsening symptoms that
last 24 h, occurring in the absence of fever or infection (Schu-
macher et al., 1965). Definitions of disability progression varied
between trials, but it was commonly defined as at least 1 point
increase on the Expanded Disability Status Scale (EDSS, an am-
bulation-centred scale from 0 to 10), or a 0.5 point increase if the
baseline EDSS was Z5.5, confirmed during two subsequent neu-
rological examinations separated by an interval of at least three to
six months free of relapses (Kurtzke, 1983). Disability progression
confirmed after three months and after six months were included
in the analysis as two separate outcomes.
2.3. Data extraction and quality assessment
Total number of treated patients, duration of follow-up, inter-
vention and control details, ARR, proportion of patients free of
disability progression (confirmed after three months and/or six
months) at the end of the study-period, and key population
baseline characteristics (age, gender, baseline EDSS score, duration
of disease, pre-trial relapse rate, and proportion of patients pre-
viously treated with DMT) were extracted from all studies, where
reported. FDA and EMA reports were used as the primary data
sources, where available. Where ARR and/or patient years of fol-
low-up were not reported, rates were derived from total number
of patients recruited, total number of relapses, mean number of
relapses per person and total patient years of follow-up, where
reported. The quality of included RCTs was assessed using Co-
chrane Collaboration's Risk of bias tool focussing on sequence
generation, allocation concealment, blinding of participants and
personnel, blinding of outcome assessment, incomplete outcome
data, selective outcome reporting and ‘other issues’ (Higgins and
Green, 2011). The overall risk of bias associated with each study
was classified as low, medium or high. High risk of bias was at-
tributed to trials with lack of blinding among participants, per-
sonnel or outcome assessors, open treatment allocation, sig-
nificant attrition bias due to incomplete outcome reporting.
Medium risk of bias was attributed to trials with unclear methods
of random sequence generation, and moderate attrition bias due to
incomplete outcome reporting. Low risk of bias was attributed to
trials without any bias detected in any of the seven domains, and
in trials with unclear methods of allocation concealment which
were otherwise judged to have been randomised appropriately.
2.4. Data analysis
A network meta-analysis was conducted combining data from
trials comparing DMTs with placebo, as well as direct comparative
trials. Combining direct and indirect (via a common comparator)
evidence in this way increases the precision of relative effect es-
timates and allows treatments to be ranked for each of the out-
comes studied. The network meta-analysis was conducted using
Bayesian Markov Chain Monte Carlo methods and fitted in R to
WinBUGS (Sturtz et al., 2005; Lunn et al., 2000). Vague or non-
informative priors were used. Convergence and lack of auto-
correlation were confirmed with autocorrelation plots after a
150,000-simulation burn-in phase. Posterior means were based on
an additional 400,000 simulation phase. The model assumed fixed
effects as the network largely consisted of single-study connec-
tions. A non-informative vague prior was used for means and a
uniform prior on the standard deviation parameter was assumed
for the baseline effects (Gelman, 2006). For all pairwise compar-
isons the model estimates relative hazard rates (HR) of disability
progression (assuming progression-free survival follows an ex-
ponential distribution) and relative ARR (assuming a poisson dis-
tribution for the number of relapses within one study arm). The
probabilistic analysis does not rely on statistical significance; ra-
ther comparative effectiveness is based on probabilities of being
among the best treatments for each outcome. The Surface under
the cumulative ranking curve (SUCRA) is used to provide a hier-
archy of treatments for each outcome, accounting for both the
location and the variance of all relative treatment effects (Salanti
et al., 2011). Values are expressed as a percentage and show the
 E. Fogarty et al. / Multiple Sclerosis and Related Disorders 9 (2016) 23–30
cumulative probability of a treatment being among the top r
treatments. The larger the SUCRA value the better the treatment
(Salanti et al., 2011). A ranking plot may be used to illustrate the
treatments’ hierarchy and the differences in SUCRA values (Chai-
mani et al., 2013). Potential sources of treatment-effect modifica-
tion from baseline population differences were investigated by
extending the model to a meta-regression including age, gender,
baseline EDSS, duration of disease, number of relapses in the
previous two years, and proportion of patients previously treated
with DMT. Treatment effects were estimated at the study-mean
covariate level. By modelling relative effect differences, the model
accounts for random variation in baseline risks. However, to ex-
plain a potential systematic impact of baseline risk on response,
baseline risk i.e. outcome in the placebo arm, was included
alongside other potential confounders as a covariate. Only trials
which include a placebo arm contribute to the estimation of this
regression parameter. Inconsistency between direct and indirect
evidence for each mean treatment effect was examined by com-
paring the model to a model relaxing the consistency assumption.
The Deviance Information Criterion was used to assess model fit.
3. Results
3.1. Evidence base
The systematic search identified 6086 potentially relevant
publications, of which 28 trials met the inclusion criteria (Fig. 1)
(Cohen et al., 2010, 2012a; Confavreux et al., 2014; Vermersch
et al., 2013; O’Connor et al., 2009, 2011; Saida et al., 2012; Mikol
et al., 2008; PRISMS (Prevention of Relapses and Disability by In-
terferon beta-1a Subcutaneously in Multiple Sclerosis) Study
Group, 1998; Jacobs et al., 1996; Kappos et al., 2010, 2011; Johnson
et al., 1995; Durelli et al., 2002; Duquette et al., 1993; Calabresi
et al., 2014a, 2014b; Panitch et al., 2002; Etemadifar et al., 2006;
Fig. 1. PRISMA flowchart – Identification and selection of randomised controlled trials of disease-modifying therapies in patients with relapsing remitting multiple sclerosis.
25
Gold et al., 2012; Fox et al., 2012; Comi et al., 2001; Lublin et al.,
2013; Coles et al., 2008, 2012; Vollmer et al., 2014; Polman et al.,
2006; Khan et al., 2013). Trials were published between 1993 and
2014, were of 1.75 years mean duration and comprised a total of
17,040 patients. The evidence network is illustrated in Fig. 2. ARR
outcomes were obtained from all 28 trials, while data on disability
progression confirmed after three months and six months were
available from 16 trials. Overall the included studies were broadly
comparable in terms of mean age (31.8–40.4 years), percentage
female (63–79%), and mean baseline EDSS score (1.9–3.2). Greater
variation was observed in the mean number of relapses in the
previous two years (1.7–3.5), mean duration of disease prior to
recruitment (1.2–10.5 years), and in the proportion of patients who
had received prior treatment with a DMT (0–100%) (see Additional
file 3 for full details of the baseline characteristics of the study po-
pulations). The overall risk of bias within included studies was
judged to be low in 14 studies (50%), medium in one study (4%)
and high in 13 studies (46%). High risk of bias was predominantly
due to the single-blind nature of many trials. All but one study
employed outcome-assessor blinding (Durelli et al., 2002) but
participants were not blinded to treatment allocation in a further
12 RCTs (Vermersch et al., 2013; Mikol et al., 2008; Kappos et al.,
2011; Durelli et al., 2002; Panitch et al., 2002; Etemadifar et al.,
2006; Fox et al., 2012; Coles et al., 2008, 2012; Cohen et al., 2012a;
Vollmer et al., 2014; O’Connor et al., 2009). Incomplete outcome
data due to loss to follow-up or imbalance in discontinuations
across treatment groups was identified in three studies, (Jacobs
et al., 1996; Coles et al., 2008, 2012) with high attrition bias
identified in one study (Jacobs et al., 1996) (see Additional file 4 for
full details of quality assessment).
3.2. Network meta-analysis
Fig. 3a–c presents the network meta-analysis results for each
DMT versus placebo for the three outcomes analysed. Each of the
26 E. Fogarty et al. / Multiple Sclerosis and Related Disorders 9 (2016) 23–30
Fig. 2. Network diagram of trials – Each node in the network represents a disease-modifying therapy. Links between nodes represent pairwise treatment comparisons
extracted from randomised controlled trials. Nodes are weighted according to the number of studies included for respective therapies. IFN β ¼interferon beta, IM -
¼ intramuscular, SC ¼ subcutaneous.
DMTs reduced ARR versus placebo. The magnitude of ARR reduc-
tion varied between 15–36% for all IFN β products, glatiramer
acetate and teriflunomide, and from 50 to 69% for alemtuzumab,
dimethyl fumarate, fingolimod and natalizumab. The risk of dis-
ability progression confirmed after three months was reduced by
19–28% with IFN β products, glatiramer acetate, fingolimod and
teriflunomide, by 38–45% for pegylated IFN β, dimethyl fumarate
and natalizumab and by 68% with alemtuzumab. Superiority over
placebo was less certain for IFN β
1a 30 mcg, IFN β
1a 22 mcg,
IFN β
1b 250 mcg and glatiramer acetate 20 mg compared with
other therapies. Results for disability progression confirmed after
six months varied slightly from disability progression confirmed
after three months for each DMT, with the exception of IFN β
1b
250 mcg which was much more efficacious when the outcome was
defined by a six-month confirmation interval (HR 0.31, 95% CI
0.15–0.62 versus HR 0.83, 95% CI 0.62–1.12). A cumulative ranking
analysis (Fig. 3d–f) revealed that alemtuzumab and natalizumab
had the highest SUCRA scores for ARR and disability progression
confirmed after three months, while IFN β
1a 30 mcg ranked
lowest and second-lowest among active treatments for these
outcomes (SUCRA scores of 9% and 32% respectively). Ranking of
treatments was affected by the definition of disability progression
largely due to the conflicting results of IFN β
1b 250 mcg, ranking
as the most efficacious treatment for disability progression con-
firmed after six months (92%) and as the least efficacious for dis-
ability progression confirmed after three months (30%). Alemtu-
zumab and natalizumab both scored relatively highly for disability
progression confirmed after six months. Notable variation in
ranking across outcomes was observed for fingolimod (81% for
ARR, 39–46% for the disability progression outcomes).
3.3. Covariate analysis
Meta-regression models including seven study-level covariates
were used to evaluate the impact of potential treatment effect
modifiers. Six covariates represented population baseline char-
acteristics (age, gender, baseline EDSS, duration of disease, number
of relapses in the previous two years, and proportion of patients
previously treated with DMT), and one covariate represented
baseline risk. The different meta-regression analyses suggested
negligible covariate effects from each of the included covariates
(See additional file 6). Adjustment of the models for baseline risk
(based on trial-specific placebo arms in each trial) similarly had
negligible impact on the results and the overall ranking of thera-
pies (See additional file 7).
DIC statistics were lower for the consistency model than the
inconsistency model for all outcomes indicating a better fit to the
data and no evidence of inconsistency between the direct and
indirect data (See Additional File 5).
4. Discussion
The last decade has seen major breakthroughs in the devel-
opment of new therapeutic strategies for RRMS. However, the
extent to which they present efficacy advantages compared to
established treatments has been unclear, as many trials are pla-
cebo-controlled, or don’t include all comparators of interest. The
Bayesian network meta-analysis framework allows the estimation
of relative treatment effects through the combination of placebo-
controlled and direct comparative studies while preserving with-
in-trial randomised treatment comparisons (Lunn et al., 2000).
The objective of this study was to compare the efficacy of in-
dividual DMTs for patients with RRMS in terms of the most
commonly used clinical trial outcomes, relapse and disability
progression. Both relapses and progression are the clinical hall-
marks of the MS disease process. However evidence from natural
history studies has shown dissociation between relapses and dis-
ability progression pathologies (Confavreux and Vukusic, 2006b).
This dissociation has been borne out in individual clinical trials
and in this network meta-analysis where clear reductions in ARR
are accompanied by more uncertain changes in disability pro-
gression. For many patients relapses are the initial defining feature
of their disease. Relapses can have a significant physical, psycho-
logical and social impact on patients, and place a substantial cost
burden on patients, their families and healthcare systems (O’Brien
 E. Fogarty et al. / Multiple Sclerosis and Related Disorders 9 (2016) 23–30 27

Fig. 3. a-c. Forest plots of treatments versus placebo for a) Annualised relapse rate, b) Disability progression confirmed at three months c) Disability progression confirmed at
six months. Relative treatment effects (rate ratios for annualised relapse rate, hazard ratios for disability progression) are represented by coloured nodes, and corresponding
95% credible intervals are represented by solid lines. Bars to the left of the central vertical line of no difference indicate superiority of the treatment over placebo. Hollow
nodes represent pairwise comparisons where the 95% credible interval spans the central vertical line of no difference. d-f. Network ranking plots for d) Annualised relapse
rate, e) Disability progression confirmed at three months c) Disability progression confirmed at six months. Treatments are ranked according to the surface under the
cumulative ranking curve (SUCRA). SUCRA values provide the hierarchy for the treatments and placebo, and show the cumulative probability (expressed as a percentage) of a
treatment being among the best options. The y-axis shows the possible ranks from r ¼ 1 up to r ¼ 11 and the x-axis shows the cumulative probabilities that the corresponding
treatment is among the top r treatments. The larger the SUCRA value the better the treatment. IFN β ¼interferon beta; IM ¼ intramuscular; SC ¼ subcutaneous; RR ¼rate ratio;
HR ¼hazard ratio; CI ¼credible interval.
28 E. Fogarty et al. / Multiple Sclerosis and Related Disorders 9 (2016) 23–30
et al., 2003; Kalb, 2007; Halper, 2007). However, it is the accu-
mulation of disability which has the greatest long-term clinical,
social and economic impact on patients and society (Fogarty et al.,
2013, 2014).
Generally, DMTs were superior to placebo in reducing MS re-
lapse rates and disability progression. However the magnitude of
the reduction and the uncertainty associated with treatment ef-
fects varied between DMTs, and between the different outcomes
included in the analysis, leading to variation in the relative ranking
of treatments. The monoclonal antibody therapies alemtuzumab
and natalizumab were generally among the highest ranked treat-
ments for all outcomes. Among the oral therapies, fingolimod and
dimethyl fumarate ranked higher than other therapies for ARR,
while there was little difference between teriflunomide and other
first-line DMTs for this outcome. Dimethyl fumarate, pegylated IFN
β and IFN β 44 mcg occupied higher rankings than other DMTs for
disability progression confirmed after three months and there was
little to distinguish between the rankings of other treatments.
European guidelines on the clinical investigation of medicinal
products for the treatment of MS recommend a six month con-
firmation interval for an accurate and reliable definition of sus-
tained worsening as studies have shown that improvement in
EDSS scores after relapse can continue beyond three months
(European Medicines Agency (EMA) Committee for Medicinal
Products for Human Use (CHMP), 2006; Cohen et al., 2012b). De-
spite this, disability progression is often defined in RCTs on the
basis of a three-month confirmation interval rather than six
months. For this reason, disability progression defined on the basis
of both three and six month confirmation intervals were included
as two separate outcomes in the analysis. With the exception of
IFN β
1b 250 mcg SC, point estimates for the efficacy of DMTs
versus placebo in reducing disability progression were similar
between both definitions with credible intervals for the two out-
comes overlapping in all cases. A significant improvement in ef-
ficacy was observed for IFN β
1b 250 mcg SC when disability
progression was defined on the basis of a six-month confirmation
interval, ranking highest for this outcome compared with being
among the lowest ranked treatments on the basis of a three-
month confirmation interval. Just one study, the INCOMIN study,
contributed evidence for the efficacy of IFN β
1b 250 mcg SC on
disability progression confirmed after six months, finding it to be
superior to IFN β
1a 30 mcg IM for this outcome (Durelli et al.,
2002). This is in contrast to two studies, one placebo controlled
and one versus glatiramer acetate which did not show significant
improvements with IFN β
1b 250 mcg SC in disability progres-
sion confirmed after three months (Duquette et al., 1993; O’Connor
et al., 2009). It has been suggested that differences in baseline
disease characteristics indicate that patients treated with IFN
β
1a 30 mcg IM in the INCOMIN study may have had a poorer
prognosis than IFN β
1b 250 mcg SC treated patients (Vartanian,
2003). Bias may also have been introduced to the INCOMIN study
by the lack of blinding as this was the only study included in the
analysis which was not assessor-blinded.
The relevance of the clinical outcomes investigated in this
study in the setting of a chronic, heterogeneous condition which
evolves over time, is uncertain. While the mechanisms that un-
derlie relapses and gradual progression may be distinct, they may
sometimes present concurrently (Cohen et al., 2012b). Improve-
ment in EDSS scores after relapses may continue beyond three to
six months and variable recovery from relapses may lead to the
accrual of relapse-related disability. Relapses decrease over time
and disappear in many patients with the onset of secondary pro-
gression (Tremlett et al., 2008). The EDSS is an ordinal rather than
linear measure and differences between each level (e.g. a one
point change between 1.5 and 2.5, and between 3.5 and 4.5 may
not comparable. Other inadequacies of the EDSS scale have been
highlighted, and it is not clear whether either of the definitions of
progression assessed in this analysis truly represents irreversible
sustained disability progression and reliably reflect long-term
changes in disease progression (Cohen et al., 2012b). Despite its
limitations, at the moment the EDSS remains the only validated
outcome measurement to determine disability (determined by the
European Medicines Agency) and its use is recommended to
continue in the future alongside improved measures of disability
progression to facilitate comparison with other studies. Research is
ongoing into novel approaches to demonstrating efficacy. RCTs are
increasingly incorporating adjunctive imaging outcomes and
composite measures such as disease-activity-free-status (Bevan
and Cree, 2014). These and other novel outcomes will provide
additional evidence on the comparative efficacy of treatments but
are not developed sufficiently to serve as substitutes for the out-
comes investigated in this analysis.
The long-term relevance of treatment effects estimated on the
basis of short-term trials (6–33 months follow-up) may be limited.
RRMS may require treatment with DMT for many years, but there
is little evidence on the long-term efficacy of these agents. The
development of neutralising antibodies can complicate prolonged
IFN β therapy (Giovannoni et al., 2002). Results from some long-
term open-label extension studies suggest maintained efficacy of
DMTs on relapse rate over extended periods of use, however the
lack of a comparator arm limits the conclusions on clinical efficacy
which can be drawn from these studies (Khatri et al., 2011; Con-
favreux et al., 2012; Kappos et al., 2012a, 2012b; Gold et al., 2014;
Ford et al., 2010). Conflicting results have been reported for the
long-term efficacy of first line DMTs on reducing disability pro-
gression (Goodin et al., 2011; Shirani et al., 2012).
The study has a number of limitations. As with traditional
pairwise meta-analysis, the model requires that individual studies
are sufficiently similar, so as to generate exchangeable treatment
effects. The trials included in the analysis were conducted over a
period of over 25 years. Slight variation in outcome definitions
between studies and changes in patient populations were evident
in duration of disease, proportion of trial populations who had
received prior DMT, and also in the downward trend in relapse
rates over time. We explored potential sources of heterogeneity
from several known covariates to adjust for baseline imbalance in
underlying risk across studies. No significant association between
efficacy and baseline risk or other covariates were identified. Al-
though no significant confounding of treatment effect was iden-
tified, meta-regression based on aggregate study-level data is
prone to ecological bias and the risk of unknown imbalances in
treatment effect modifiers cannot be excluded (Cooper et al.,
2009). Consistency between the direct and indirect evidence is
dependent on the distribution of effect modifiers and where both
direct and indirect evidence was available no evidence of incon-
sistency was identified. There was variation in study quality
among the included trials. In particular, concerns regarding the
open-label design may affect our level of confidence in the esti-
mates of effect from some trials, as the potential for bias cannot be
excluded. In this study we did not investigate the relative ac-
ceptability or tolerability of the various treatment strategies.
Consideration of safety outcomes alongside treatment efficacy is
central to decision-making in healthcare and has been addressed
elsewhere (Saidha et al., 2012; Smith et al., 2010). Cost-effective-
ness, likewise, was not considered. This is increasingly a critical
determinant of reimbursement in many countries, due to the high
acquisition cost of many of these treatments.
5. Conclusion
The network meta-analysis approach is not a substitute for
large, well-designed RCTs comparing the comparators of interest.
 E. Fogarty et al. / Multiple Sclerosis and Related Disorders 9 (2016) 23–30
However, given the current limitations in the evidence base, this
study provides an alternative framework within which compara-
tive effectiveness can be estimated. This analysis integrated evi-
dence from a comprehensive network of trials of DMT in RRMS,
using both direct and indirect data to simultaneously evaluate
treatment effects of interventions not previously studied in head-
to-head trials. Treatment rankings have been presented for sepa-
rate outcomes, based on cumulative probabilities of being among
the most effective treatments. There is little difference (indicating
similarity in treatment effects) between these probabilities for
many treatments, with newer agents natalizumab and alemtuzu-
mab demonstrating consistently high ranking across outcomes.
Health policy and resource allocation decision-making frequently
requires quantification of comparative efficacy. Evidence synthesis
within this network meta-analysis framework is a powerful ap-
proach to dealing with complex evidence structures to facilitate
coherent, evidence-based decision-making.
Author contributions
EF conceived and designed the study, identified and acquired
trials, extracted data, contributed to data analysis, interpreted the
results drafted the manuscript and is guarantor. SS identified trials,
extracted data, performed statistical analysis and critically revised
the manuscript for important intellectual content. CW provided
statistical advice and critically revised the manuscript for im-
portant intellectual content. NT and MB critically revised the
manuscript for important intellectual content. All authors ap-
proved the final version of the manuscript.
Conflict of interest statement
EF, SS, CW and MB report no conflict of interest. Prof. Tubridy
has received unrestricted educational and research grants on be-
half of the Department Of Neurology at St. Vincent’s University
Hospital from Bayer Schering, Biogen Idec and Novartis.
Ethical approval
Not required for this study.
Funding
No funding was received for this study.
Data sharing
Statistical code is available from the corresponding author.
Appendix A. Supplementary material
Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.msard.2016.06.
001.
References
Bevan, C.J., Cree, B.C., 2014. Disease activity free status: a new end point for a new
era in multiple sclerosis clinical research? JAMA Neurol. 71 (3), 269–270.
29
Calabresi, P.A., Radue, E.-W., Goodin, D., Jeffery, D., Rammohan, K.W., Reder, A.T.,
et al., 2014a. Safety and efficacy of fingolimod in patients with relapsing-re-
mitting multiple sclerosis (FREEDOMS II): a double-blind, randomised, placebo-
controlled, phase 3 trial. Lancet Neurol. 13 (6), 545–556.
Calabresi, P.A., Kieseier, B.C., Arnold, D.L., Balcer, L.J., Boyko, A., Pelletier, J., et al.,
2014b. Pegylated interferon beta-1a for relapsing-remitting multiple sclerosis
(ADVANCE): a randomised, phase 3, double-blind study. Lancet Neurol. 13 (7),
657–665.
Chaimani, A., Higgins, J.P.T., Mavridis, D., Spyridonos, P., Salanti, G., 2013. Graphical
tools for network meta-analysis in STATA. PLoS One 8 (10), e76654.
Cohen, J.A., Coles, A.J., Arnold, D.L., Confavreux, C., Fox, E.J., Hartung, H.P., et al.,
2012a. Alemtuzumab versus interferon beta 1a as first-line treatment for pa-
tients with relapsing-remitting multiple sclerosis: a randomised controlled
phase 3 trial. Lancet 380 (9856), 1819–1828.
Cohen, J.A., Reingold, S.C., Polman, C.H., Wolinsky, J.S., 2012b. Disability outcome
measures in multiple sclerosis clinical trials: Current status and future pro-
spects. Lancet Neurol. 11 (5), 467–476.
Cohen, J.A., Barkhof, F., Comi, G., Hartung, H.P., Khatri, B.O., Montalban, X., et al.,
2010. Oral fingolimod or intramuscular interferon for relapsing multiple
sclerosis. New Engl. J. Med. 362 (5), 402–415.
Coles, A., 2009. Multiple sclerosis: the bare essentials. Pract. Neurol. 9 (2), 118–126.
Coles, A.J., Compston, D.A.S., Selmaj, K.W., Lake, S.L., Moran, S., Margolin, D.H., et al.,
2008. Alemtuzumab vs. interferon beta-1a in early multiple sclerosis. New Engl.
J. Med. 359 (17), 1786–1801.
Coles, A.J., Twyman, C.L., Arnold, D.L., Cohen, J.A., Confavreux, C., Fox, E.J., et al.,
2012. Alemtuzumab for patients with relapsing multiple sclerosis after disease-
modifying therapy: a randomised controlled phase 3 trial. Lancet 380 (9856),
1829–1839.
Comi, G., Filippi, M., Wolinsky, J.S., 2001. European/Canadian multicenter, double-
blind, randomized, placebo-controlled study of the effects of glatiramer acetate
on magnetic resonance imaging–measured disease activity and burden in pa-
tients with relapsing multiple sclerosis. European/Canadian Glatiramer Acetate
Study Group. Ann. Neurol. 49 (3), 290–297.
Confavreux, C., Vukusic, S., 2006a. Natural history of multiple sclerosis: a unifying
concept. Brain 129 (3), 606–616.
Confavreux, C., Vukusic, S., 2006b. Accumulation of irreversible disability in mul-
tiple sclerosis: from epidemiology to treatment. Clin. Neurol. Neurosurg. 108
(3), 327–332.
Confavreux, C., Li, D.K., Freedman, M.S., Truffinet, P., Benzerdjeb, H., Wang, D., et al.,
2012. Long-term follow-up of a phase 2 study of oral teriflunomide in relapsing
multiple sclerosis: safety and efficacy results up to 8.5 years. Mult. Scler. 18 (9),
1278–1289.
Confavreux, C., O’Connor, P., Comi, G., Freedman, M.S., Miller, A.E., Olsson, T.P., et al.,
2014. Oral teriflunomide for patients with relapsing multiple sclerosis (TOW-
ER): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet
Neurol. 13 (3), 247–256.
Cooper, N.J., Sutton, A.J., Morris, D., Ades, A.E., Welton, N.J., 2009. Addressing be-
tween-study heterogeneity and inconsistency in mixed treatment compar-
isons: application to stroke prevention treatments in individuals with non-
rheumatic atrial fibrillation. Stat. Med. 28 (14), 1861–1881.
Duquette, P., Girard, M., Despault, L., DuBois, R., Knobler, R.L., Lublin, F.D., et al.,
1993. Interferon beta-1b is effective in relapsing-remitting multiple sclerosis. I.
Clinical results of a multicenter, randomized, double-blind, placebo- controlled
trial. Neurology 43 (4 I), 655–661.
Durelli, L., Verdun, E., Barbero, P., Bergui, M., Versino, E., Ghezzi, A., et al., 2002.
Every-other-day interferon beta-1b versus once-weekly interferon beta-1a for
multiple sclerosis: results of a 2-year prospective randomised multicentre
study (INCOMIN). Lancet 359 (9316), 1453–1460.
Etemadifar, M., Janghorbani, M., Shaygannejad, V., 2006. Comparison of Betaferon,
Avonex, and Rebif in treatment of relapsing-remitting multiple sclerosis. Acta
Neurol. Scand. 113 (5), 283–287.
European Medicines Agency (EMA) Committee for Medicinal Products for Human
Use (CHMP). Guideline on Clinical Investigation of Medicinal Products for the
Treatment of) Multiple Sclerosis. London, 16 November 2006 Doc. Ref. CPMP/
EWP/561/98 Rev. 1.
Fogarty, E., Walsh, C., McGuigan, C., Tubridy, N., Barry, M., 2014. Direct and indirect
economic consequences of multiple sclerosis in Ireland. Appl. Health Econ.
Health Policy, 1–11.
Fogarty, E., Walsh, C., Adams, R., McGuigan, C., Barry, M., Tubridy, N., 2013. Relating
health-related quality of Life to disability progression in multiple sclerosis,
using the 5-level EQ-5D. Mult. Scler. 19 (9), 1190–1196.
Ford, C., Goodman, A.D., Johnson, K., Kachuck, N., Lindsey, J.W., Lisak, R., et al., 2010.
Continuous long-term immunomodulatory therapy in relapsing multiple
sclerosis: results from the 15-year analysis of the US prospective open-label
study of glatiramer acetate. Mult. Scler. 16 (3), 342–350.
Fox, R.J., Miller, D.H., Phillips, J.T., Hutchinson, M., Havrdova, E., Kita, M., et al., 2012.
Placebo-controlled phase 3 study of oral BG-12 or glatiramer in multiple
sclerosis. New Engl. J. Med. 367 (12), 1087–1097.
Gelman, A., 2006. Prior distributions for variance parameters in hierarchical
models. Bayesian Anal. 1 (3) 151–33.
Giovannoni, G., Munschauer, F.E., Deisenhammer, F., 2002. Neutralising antibodies
to interferon beta during the treatment of multiple sclerosis. J. Neurol., Neu-
rosurg. Psychiatry 73 (5), 465–469.
Gold, R., Kappos, L., Arnold, D.L., Bar-Or, A., Giovannoni, G., Selmaj, K., et al., 2012.
Placebo-controlled phase 3 study of oral BG-12 for relapsing multiple sclerosis.
New Engl. J. Med. 367 (12), 1098–1107.
30 E. Fogarty et al. / Multiple Sclerosis and Related Disorders 9 (2016) 23–30
Gold, R., Phillips, J.T., Bar-Or, A., Hutchinson, M., Kappos, L., Rana, J., et al., 2014. 3-
year follow-up of oral BG-12 (dimethyl fumarate) for relapsing-remitting
multiple sclerosis (RRM)S: Integrated clinical outcomes from the define, con-
firm, and endorse studies. Mult. Scler. 20 (7), 932.
Goodin, D.S., Jones, J., Li, D., Traboulsee, A., Reder, A.T., Beckmann, K., et al., 2011.
Establishing long-term efficacy in chronic disease: use of recursive partitioning
and propensity score adjustment to estimate outcome in MS. PLoS One 6 (11),
e22444.
Halper, J., 2007. The psychosocial effect of multiple sclerosis: the impact of relapses.
J. Neurol. Sci. 256 (SUPPL. 1), S34–S38.
Higgins, J. Green S, (editors). Cochrane Handbook for Systematic Reviews of Inter-
ventions. Version 5.1.0. the Cochrane Collaboration,. 2011.
Jacobs, L.D., Cookfair, D.L., Rudick, R.A., Herndon, R.M., Richert, J.R., Salazar, A.M.,
et al., 1996. Intramuscular interferon beta
1a for disease progression in re-
lapsing multiple sclerosis. Ann. Neurol. 39 (3), 285–294.
Johnson, K.P., Brooks, B.R., Cohen, J.A., Ford, C.C., Goldstein, J., Lisak, R.P., et al., 1995.
Copolymer 1 reduces relapse rate and improves disability in relapsing- remit-
ting multiple sclerosis: results of a phase III multicenter, double- blind, place-
bo-controlled trial. Neurology 45 (7), 1268–1276.
Kalb, R., 2007. The emotional and psychological impact of multiple sclerosis re-
lapses. J. Neurol. Sci. 256 (SUPPL. 1), S29–S33.
Kappos, L., Radue, E.W., O’Connor, P., Polman, C., Hohlfeld, R., Calabresi, P., et al.,
2010. A placebo-controlled trial of oral fingolimod in relapsing multiple
sclerosis. New Engl. J. Med. 362 (5), 387–401.
Kappos, L., Li, D., Calabresi, P.A., O’Connor, P., Bar-Or, A., Barkhof, F., et al., 2011.
Ocrelizumab in relapsing-remitting multiple sclerosis: a phase 2, randomised,
placebo-controlled, multicentre trial. Lancet 378 (9805), 1779–1787.
Kappos, L., Radue, E.W., O’Connor, P., Polman, C., Hohlfeld, R., Calabresi, P., et al.,
2012a. Long-term efficacy and safety of fingolimod (FTY720) in relapsing-re-
mitting multiple sclerosis (RRMS): results from the extension of the phase III
FREEDOMS study. Neurology 78, 1.
Kappos, L., Belachew, S., Butzkueven, H., Pellegrini, F., Trojano, M., Wiendl, H., et al.,
2012b. Long-term safety and efficacy and association between baseline treat-
ment history and postbaseline relapses in multiple sclerosis patients treated
with natalizumab in the tysabri(registered trademark) observational program
(top). Neurology 78 (1).
Khan, O., Rieckmann, P., Boyko, A., Selmaj, K., Zivadinov, R., Group, G.S., 2013. Three
times weekly glatiramer acetate in relapsing-remitting multiple sclerosis. Ann.
Neurol. 73 (6), 705–713.
Khatri, B., Barkhof, F., Comi, G., Hartung, H.-P., Kappos, L., Montalban, X., et al., 2011.
Comparison of fingolimod with interferon beta-1a in relapsing-remitting
multiple sclerosis: a randomised extension of the TRANSFORMS study. Lancet
Neurol. 10 (6), 520–529.
Kurtzke, J.F., 1983. Rating neurologic impairment in multiple sclerosis: an expanded
disability status scale (EDSS). Neurology 33 (11), 1444–1452.
Lublin, F.D., Cofield, S.S., Cutter, G.R., Conwit, R., Narayana, P.A., Nelson, F., et al.,
2013. Randomized study combining interferon and glatiramer acetate in mul-
tiple sclerosis. Ann. Neurol. 73 (3), 327–340.
Lunn, D.J., Thomas, A., Best, N., Spiegelhalter, D., 2000. WinBUGS – a Bayesian
modelling framework: concepts, structure, and extensibility. Stat. Comput. 10,
325–337.
Mikol, D.D., Barkhof, F., Chang, P., Coyle, P.K., Jeffery, D.R., Schwid, S.R., et al., 2008.
Comparison of subcutaneous interferon beta-1a with glatiramer acetate in
patients with relapsing multiple sclerosis (the REbif vs Glatiramer Acetate in
Relapsing MS Disease [REGAR)D] (stud)y): (a multicentre, randomised, parallel,
open-label trial). Lancet Neurol. 7 (10), 903–914.
Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., 2009. Preferred reporting items for
systematic reviews and meta-analyses: the PRISMA statement. BMJ, 339.
O’Brien J.A., Ward, A.J., Patrick, A.R., Caro, J., 2003. Cost of managing an episode of
relapse in multiple sclerosis in the United States. BMC Health Services Research
3(O’Brien J.A., jobrien@caroresearch.com; Ward A.J., alexward@caroresearch.
com; Patrick A.R., apatrick@hsph.harvard.edu; Caro J., jcaro@caroresearch.com).
Caro Research Institute, Concord, MA 01742, United States, pp. 1–12.
O’Connor, P., Filippi, M., Arnason, B., Comi, G., Cook, S., Goodin, D., et al., 2009. 250
(mu)g or 500 (mu)g interferon beta-1b versus 20 mg glatiramer acetate in re-
lapsing-remitting multiple sclerosis: a prospective, randomised, multicentre
study. Lancet Neurol. 8 (10), 889–897.
O’Connor, P., Wolinsky, J.S., Confavreux, C., Comi, G., Kappos, L., Olsson, T.P., et al.,
2011. Randomized trial of oral teriflunomide for relapsing multiple sclerosis.
New Engl. J. Med. 365 (14), 1293–1303.
Panitch, H., Goodin, D.S., Francis, G., Chang, P., Coyle, P.K., O’Connor, P., et al., 2002.
Randomized, comparative study of interferon (beta)-1a treatment regimens in
MS: the evidence trial. Neurology 59 (10), 1496–1506.
Polman, C.H., O’Connor, P.W., Havrdova, E., Hutchinson, M., Kappos, L., Miller, D.H.,
et al., 2006. A randomized, placebo-controlled trial of natalizumab for relapsing
multiple sclerosis. New Engl. J. Med. 354 (9), 899–910.
PRISMS (Prevention of Relapses and Disability by Interferon beta-1a Sub-
cutaneously in Multiple Sclerosis) Study Group, 1998. Randomised double-
blind placebo-controlled study of interferon beta-1a in relapsing/remitting
multiple sclerosis. Lancet 352 (9139), 1498–1504.
Saida, T., Kikuchi, S., Itoyama, Y., Hao, Q., Kurosawa, T., Nagato, K., et al., 2012. A
randomized, controlled trial of fingolimod (FTY720) in Japanese patients with
multiple sclerosis. Mult. Scler. 18 (9), 1269–1277.
Saidha, S., Eckstein, C., Calabresi, P.A., 2012. New and emerging disease modifying
therapies for multiple sclerosis. Ann. N. Y. Acad. Sci. 1247 (1), 117–137.
Salanti, G., Ades, A.E., Ioannidis, J.P.A., 2011. Graphical methods and numerical
summaries for presenting results from multiple-treatment meta-analysis: an
overview and tutorial. J. Clin. Epidemiol. 64 (2), 163–171.
Schumacher, G.A., Beebe, G.W., Kibler, R.F., Kurland, L.T., Kurtzke, J.F., Fea, Mc.
Dowell, 1965. Problems of experimental trials of therapy in multiple sclerosis:
report by the panel on the evaluation of experimental trials of therapy in
multiple sclerosis. Ann. N. Y. Acad. Sci. 122, 552–568.
Shirani, A., Zhao, Y., Karim, M.E., Evans, C., Kingwell, E., Van Der Kop, M.L., et al.,
2012. Association between use of interferon beta and progression of disability
in patients with relapsing-remitting multiple sclerosis. JAMA: J. Am. Med. As-
soc. 308 (3), 247–256.
Smith, B., Carson, S., Fu, R., et al. Drug class review: disease-modifying drugs for
multiple sclerosis: Final Update 1 Report [Internet]. Portland (OR): Oregon
Health & Science University; 2010 Aug. Available from: 〈http://www.ncbi.nlm.
nih.gov/books/NBK50570/〉.
Sturtz, S., Ligges, U., Gelman, A., 2005. R2WinBUGS: a package for running Win-
BUGS from R. J. Stat. Softw. 12 (3), 1–16.
Tremlett, H., Zhao, Y., Joseph, J., Devonshire, V., Adams, D., Craig, D., et al., 2008.
Relapses in multiple sclerosis are age- and time-dependent. J. Neurol. Neuro-
surg. Psychiatry 79 (12), 1368–1374.
Vartanian, T., 2003. An examination of the results of the EVIDENCE, INCOMIN, and
phase III studies of interferon beta products in the treatment of multiple
sclerosis. Clin. Ther. 25 (1), 105–118.
Vermersch, P., Czlonkowska, A., Grimaldi, L.M., Confavreux, C., Comi, G., Kappos, L.,
2013. Teriflunomide versus subcutaneous interferon beta-1a in patients with
relapsing multiple sclerosis: a randomised, controlled phase 3 trial. Mult. Scler.
J.
Vollmer, T.L., Sorensen, P.S., Selmaj, K., Zipp, F., Havrdova, E., Cohen, J.A., et al., 2014.
A randomized placebo-controlled phase III trial of oral laquinimod for multiple
sclerosis. J. Neurol. 261 (4), 773–783.