Article Text

Download PDFPDF

Interferon β for secondary progressive multiple sclerosis: a systematic review
  1. Loredana La Mantia1,
  2. Laura Vacchi2,
  3. Marco Rovaris1,
  4. Carlo Di Pietrantonj3,
  5. George Ebers4,
  6. Sten Fredrikson5,
  7. Graziella Filippini2
  1. 1Neurorehabilitation Unit, Multiple Sclerosis Centre, IRCCS S Maria Nascente Fondazione Don Gnocchi, Milano, Italy
  2. 2Neuroepidemiology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milano, Italy
  3. 3Regional Reference Centre for Epidemiology, Cochrane Vaccines Field, Local Health Unit, Alessandria, Italy
  4. 4Clinical Neurology, University of Oxford, Oxford, UK
  5. 5Division of Neurology R54, Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
  1. Correspondence to Dr L La Mantia, Multiple Sclerosis Centre, Neurorehabilitation Unit, IRCCS Santa Maria Nascente Fondazione Don Gnocchi, Via Capecelatro 66, I-20148 Milano, Italy; lamantialore{at}


Background It is unclear whether recombinant β interferons (IFNβ) can be effective in secondary progressive multiple sclerosis (SPMS). The aim was to determine whether IFNβ can reduce the risk of disability and cognitive impairment progression in SPMS.

Methods Using Cochrane methodology, we reviewed all randomised placebo controlled trials of IFNβ in SPMS patients (1995–March 2012).

Results 5 trials (3082 patients) were included. After 3 years, interferons did not reduce disability progression, confirmed at 6 months (RR 0.98, 95% CI 0.82 to 1.16). A small reduction in the number of patients who had relapses during the first 3 years of treatment (RR 0.91, 0.84 to 0.97) was found. No analysis of cognitive data was possible. More treated than placebo patients dropped out for adverse events.

Conclusion 3 year treatment with IFNβ does not delay permanent disability in SPMS but reduces relapse risk, indicating that the anti-inflammatory effect of IFNβ is unable to prevent MS progression once it has become established.

Statistics from


Recombinant interferons β-1a (IFNβ-1a) and β-1b (IFNβ-1b) are established treatments for relapsing remitting multiple sclerosis (MS).1 Treatment with recombinant interferon β-1b (Betaferon) was approved by European and US regulatory agencies for secondary progressive MS (SPMS) early in the millennium.2 ,3 However, in 2005, the European Agency for the Evaluation of Medicinal Products limited its indication to SPMS patients with active disease, evidenced by relapses.4 Extavia, a recombinant IFNβ-1b produced by another company, was approved with the same indications in 2008 (Europe) and 2009 (USA).5 ,6

Nevertheless, recombinant IFNβ is currently the firstline treatment for SPMS, as mitoxantrone, the other approved agent, has a worse risk/benefit profile.7 It therefore seemed important to evaluate all of the available evidence regarding the use of these expensive drugs8 in patients with SPMS. We used the Cochrane methodology to systematically review all randomised controlled studies that compared IFNβ with placebo in SPMS. Our primary objectives were to determine (a) whether IFNβ is more effective than placebo in reducing the progression of disability and cognitive impairment in SPMS patients and (b) whether these agents are safe. Secondary aims were to determine the effects of IFNβ on clinical relapses, quality of life and mood, and MRI measures of disease activity/progression. This is an up to date and abridged version of the Cochrane systematic review.9


We applied the Cochrane Collaboration methodology10 and followed our predefined protocol, as reported in the Cochrane review.9 Criteria for inclusion of the studies in the review were as follows: randomised, double or single blind, placebo controlled trials of recombinant IFNβ for patients with a diagnosis of MS according to predefined criteria and affected by SP—that is, patients with an initial relapsing remitting phase followed by progression with or without superimposed relapses, according to Lublin criteria.11 We searched for randomised controlled trials in the Cochrane Multiple Sclerosis Group's Specialised Register (1995 to 15 March 2012), MEDLINE (1996–2012), EMBASE (1974–2012), clinical trial registries,12 Food and Drug Administration (FDA) reports and tried to retrieve missing data from investigators or sponsors.

Two authors scrutinised all articles found and decided independently whether a trial met the criteria for inclusion in the review. They independently extracted trial data onto an Excel worksheet. Trial arms involving the same agent at different dosages were converted into a single arm by summing the number of events and sample size. Any disagreements were resolved by discussion among the two authors and a third reviewer.

We initially defined sustained disability progression (primary outcome measure) as an increase in Expanded Disability Status Scale (EDSS) score13 by at least 1 point (0.5 points if baseline EDSS ≥5.5), confirmed at the 6 month follow-up visit. However, as this endpoint was not available in all studies, we also adopted a less rigorous definition of progression (ie, progression confirmed at 3 months). Cognitive impairment was assessed by the instruments used in the published studies, provided they had been validated.

Safety (primary outcome measure) was evaluated in terms of: (a) serious adverse events (SAE) defined according to the WHO14; and (b) the proportion of patients who withdrew or dropped out because of adverse events. For all studies we extracted all adverse events for which it was reasonable to assume a causal relation with the treatment. Quality of life and mood (secondary outcome measures) were assessed using the scale scores reported in the studies, provided the scales had been validated. To assess the effect of treatment on brain MRI outcomes (secondary outcome measure), we considered the number of combined unique (CU) lesions, defined as new or enhancing or enlarging lesions, and measures of cerebral atrophy (normalised or absolute changes in brain volume).

The two authors evaluated trial quality independently, assessing six quality characteristics related to risk of bias according to Higgins15: (i) generation of randomisation sequences (selection bias); (ii) concealment of allocation (selection bias); (iii) blinding of outcome assessment (detection bias); (iv) blinding of participants and personnel (performance bias); (v) how patients who withdrew after randomisation or were lost to follow-up were treated in the analysis (attrition bias); and (vi) other factors.

The editors of the Cochrane Multiple Sclerosis and Rare CNS Diseases Group, and of the Cochrane Library, gave permission to co-publish this submitted version of the Cochrane review.

Statistical methods

The meta-analysis was performed using Review Manager (RevMan) software, V.5.1 (Copenhagen, The Nordic Cochrane Centre, The Cochrane Collaboration, 2011).16 Binary outcomes were assessed by estimating RR with 95% CIs, using the Mantel–Haenszel method. Continuous outcomes were assessed as mean differences between the intervention and control groups. The analysis always adhered to the original randomisation, as presented in trial publications. Between study heterogeneity of findings was assessed using the I2 statistic. If I2 exceeded 50%, we considered the heterogeneity to be substantial and used a random effects approach17 to assess its effect on treatment efficacy. Between study differences in baseline clinical characteristics (age, EDSS, disease duration, duration of progressive disease phase and pre-study number of relapses) were assessed by one way ANOVA and Sheffé's post hoc multiple comparison test.

A within studies subgroup analysis was done considering separately patients with pre-study relapses and patients without pre-study relapses, in order to assess whether the effect of treatment on disability progression was modified by pre-study relapses. A post hoc subgroup analysis was done on subsets of studies characterised by differing patient age and disease duration.


We provisionally identified 129 records (figure 1); 70 of these were discarded based on examination of titles or abstracts. Full papers pertaining to the remaining 59 abstracts were retrieved; 18 of these were discarded because they did not fulfil predefined inclusion criteria. The reasons for exclusion are reported in figure 1.9

Figure 1

Flow diagram illustrating study selection. MS, multiple sclerosis; SP, secondary progressive.

The following five trials were eligible: the European Study Group trial of 1998 (European SG)18; the Secondary Progressive Efficacy Clinical Trial of Recombinant Interferon-Beta-1a in MS study of 2001 (SPECTRIMS)19; the IMPACT study of 2002 (IMPACT)20; the Nordic Study Group trial of 2004 (Nordic)21; and the North American Study Group trial of 2004 (North American SG).22 The remaining 36 studies were related to these five trials.

The characteristics of the included trials are reported in table 1.

Table 1

Characteristics of the trials included

Two studies (North American SG,22 European SG18) evaluated IFNβ-1b. The other three studies were conducted with IFNβ-1a at various doses and preparations (SPECTRIMS,19 IMPACT,20 Nordic21). Treatment duration and follow-up were 2 years for IMPACT20 and 3 years for the remaining studies.

A total of 3082 patients from these five studies are analysed in this review; of these, 1809 were given IFNβ and 1273 were given placebo. Inclusion criteria varied between studies (table 1).

In four studies, diagnostic criteria of SPMS were defined as a period of deterioration sustained for at least 6 months, occurring after a period of relapsing remitting disease; while the other study (IMPACT)20 included patients with at least 1 year of deterioration.

In four studies the primary outcome was time to disability progression, defined as 3 months of sustained EDSS increase for the European SG18 and SPECTRIMS19 and 6 months sustained EDSS increase for the European SG,23 North American SG22 and Nordic.21 IMPACT20 used worsening of the MS Functional Composite as the primary outcome and 3 months of sustained EDSS increase as the secondary outcome. We retrieved additional data for the North American SG22 from an FDA report,3 and data on 3 months of sustained EDSS increase at 3 years in SPECTRIMS from Serono (manufacturer of IFNβ-1a).

The baseline characteristics of the patients are shown in table 2.

Table 2

Baseline characteristics of the patients in the included trials

Patients recruited to the European SG18 and SPECTRIMS19 studies were significantly younger (p<0.001) and had significantly shorter disease duration (p<0.05) than those in the other three trials.20–22 Mean number of pre-study relapses also varied between studies (range 0.8–1.7), as did the pre-study observation period (4 years for Nordic,21 1 year for IMPACT20 and 2 years for the other three trials18 ,19 ,22).

Trial quality assessment

Random sequence generation was adequate for four trials18–20 ,22 and unclear for one21; allocation concealment was adequate for three trials18 ,19 ,22 and unclear for two20 ,21; blinding of outcome assessment was adequate for four trials18–20 ,22 and unclear for one21; complete outcome data were adequate in two trials19 ,20 and unclear in three studies.18 ,21 ,22 Two trials had other limitations: one stopped early for benefit18 and one20 used an inappropriate scale to assess primary outcome.24

Effects of interventions

Primary outcomes

Number of patients with sustained progression of disability

The numbers of patients with 6 months of sustained disability progression assessed at 3 years were available for three trials18 ,21 ,22: 1177 (65% of total) patients treated with IFNβ and 849 (67% of total) treated with placebo. From these data, the RR of sustained progression in IFNβ treated patients was 0.98 (95% CI 0.82 to 1.16; p=0.79), showing no benefit for interferon (figure 2). However, between study heterogeneity was high (I2=57%), with only the European SG18 having a significant treatment effect. One study19 reported the numbers of patients with progression only by 3 month confirmation; the RR for IFNβ treated patients was 0.91 (95% CI 0.80 to 1.04; p=0.17); the RR of 3 months of sustained progression in the European SG18 was 0.84 (95% CI 0.72 to 0.97; p=0.02).

Figure 2

Patients with a sustained (6 months) Expanded Disability Status Scale increase during the first 3 years of treatment. In the Forest plot, each trial is represented by a square, with the centre denoting the RR for that trial and the extremities of the horizontal bars denoting 95% CI. Square size is directly proportional to the weight of the trial within the group of trials. The diamond gives the overall RR for all trials: its centre denotes the RR and extremities the 95% CI. Trials are ordered chronologically. Risk ratios were estimated by the Mantel–Haenszel approach using a random effects model. *Data extracted from Kappos 2001.23 n=number of patients with outcome; N=number of randomised patients.

Data from two trials19 ,20 with 1054 patients (34%) were available to calculate the number of patients who had progressed (confirmed at 3 months) at 2 years; the RR (0.94, 0.81 to 1.08; p=0.36) indicated no benefit for interferon.

Subgroup analyses of progression at 3 years in patients with pre-study relapses (RR 0.90, 95% CI 0.75 to 1.09; p=0.28) and in patients without pre-study relapses (RR 1.05, 95% CI 0.83 to 1.33; p=0.70) showed no treatment effect in either group (figure 3). However, a high level of heterogeneity was found (I2=51%) for the subgroup without pre-study relapses.

Figure 3

Patients with sustained (3 or 6 months) Expanded Disability Status Scale increase during the first 3 years of treatment. Patients with versus patients without pre-study relapses. Explanation of the Forest plots as in figure 2. Risk ratios were estimated by the Mantel–Haenszel approach using a random effects model. *Data extracted from Kappos 2001.23 n=number of patients with outcome; N=number of randomised patients.

We compared the European SG18 and SPECTRIMS19 trials with the other three studies because the former were similar in terms of patient age and disease duration at study entry (see table 2). There was a tendency to an effect in younger patients and shorter disease duration (RR 0.92, 95% CI 0.83 to 1.02; p=0.10) compared with older patients and longer disease duration (RR 1.03, 95% CI 0.90 to 1.18; p=0.64) (figure 4). A high level of heterogeneity was found (I2= 61%) for the subgroup of younger patients and shorter disease duration.

Figure 4

Patients with sustained (3 or 6 months) Expanded Disability Status Scale increase during the first 3 years of treatment according to patient age and disease duration. Explanation of the Forest plots as in figure 2. Risk ratios were estimated by the Mantel–Haenszel approach using a random effects model. n=number of patients with outcome; N=number of randomised patients.


The RR (1.00, 95% CI 0.83 to 1.19; p=0.97) of an SAE in treated patients did not differ from that in controls. However, significantly more patients treated with IFNβ (RR 2.62, 95% CI 1.92 to 3.57; p<0.001) withdrew or dropped out for adverse events (including SAEs) than controls. Sixteen deaths occurred in treated patients (four suicides, three for pulmonary embolism, three for cardiac arrest, one for cancer, one for intracerebral haemorrhage, one for brainstem infarction, one for urosepsis and two for unknown cause). Seven deaths occurred in control patients (two suicides, one for subarachnoid haemorrhage, one for ‘arteriosclerosis’, one for pneumonia and two for unknown cause). The following adverse events were significantly related to IFNβ treatment: cutaneous necrosis, injection site reactions, influenza-like syndrome and leukopenia.

Secondary outcomes

Number of patients with at least one relapse during follow-up

We were able to obtain information on relapses for 1590 (88% of total) treated and 1049 (82% of total) placebo patients. IFNβ treatment significantly reduced the risk of relapses over 3 years of follow-up (RR 0.91, 95% CI 0.84 to 0.97; p=0.007) (figure 5). One study20 reported the number of patients who had relapses at 2 years: the RR for treated patients was 0.72 (95% CI 0.54 to 0.95; p=0.02).

Figure 5

Patients with at least one relapse during the first 3 years of treatment. Explanation of the Forest plot as in figure 2. Risk ratios were estimated by the Mantel–Haenszel approach using a random effects model. n=number of patients with outcome; N=number of randomised patients.

Relapse rate during follow-up

This outcome was available after 2 years for one study20 and after 3 years for three studies.18 ,19 ,22 The mean difference in relapse rate between IFNβ and placebo was −0.12 (95% CI −0.20 to −0.04; p=0.005) after 2 years and −0.16 (95% CI −0.21 to −0.10; p<0.001) after 3 years.

Changes in MRI markers of disease activity

The Nordic study21 did not include MRI assessment and, therefore, did not contribute to this outcome analysis. At baseline, contrast enhancing lesions were absent in about 60% of IMPACT20 patients and in 50% of patients from the other three trials.18 ,19 ,22 MRI derived measures were available for small samples of participants and MRI analysis criteria, scanning schedules varied across studies. As a consequence, we were able to perform only a qualitative analysis of the MRI findings. Data on patients with new or enhancing or enlarging lesions (CU) were available from the European SG study25 for 17% and 15% of participants after 6 and 24 months of treatment, respectively, and from the SPECTRIM study26 for 43% of participants after 9 months. Fewer patients had CU lesions during interferon treatment than controls at 6 months (RR 0.70, 95% CI 0.52 to 0.95; p=0.02), at 9 months (RR 0.51, 95% CI 0.42 to 0.62; p<0.001) and at 24 months (RR 0.52, 95% CI 0.33 to 0.80; p=0.003). No analysis of cerebral atrophy data was possible.

Other outcomes

Because data were not comparable between studies (different scales used, different assessment time points) meta-analyses for cognitive impairment, quality of life and mood in relation to treatment were not possible.

Quality of life was evaluated by the Sickness Impact Profile in the European SG,18 Multiple Sclerosis Quality of Life Inventory in North American SG22 and IMPACT,20 ,27 General Health Questionnaire in SPECTRIMS19 and the Nottingham Health Profile Part I in Nordic.21 A significant benefit related to IFNβ was found only in two studies.18 ,27

Psychological aspects were evaluated in three studies, with different scales: Montgomery Asberg Depression Rating Scale assessing mood changes and suicidal risk,18 changes in Beck Depression Inventory score between baseline and end of follow-up,22 the Center for Epidemiologic Studies Depression Rating Scale and Beck Hopelessness Scale.19 Patients on β-1b had no increased incidence of new or worsened depression, according to the results of the studies.18 ,19 ,22 Depression was reported in 29% of patients receiving placebo, in 32% receiving low dose IFNβ and in 35% receiving high dose IFNβ.19


The results of this systematic review show that recombinant interferons are not effective in reducing the risk of disease progression in SPMS. This finding is based on the demonstration of no between group differences in disability progression confirmed at 318 ,19 or 6 months21 ,22 in four studies. Preventing disability progression is the key therapeutic goal in SPMS, and EDSS increase is a validated measure of unremitting disability in this condition, provided it is confirmed after a sufficient period of time (at least 6 months is necessary and 1 year would be better).28

Nevertheless, our analysis revealed that IFNβ treatment was associated with a significant reduction in the risk of relapse, as measured by numbers of patients experiencing relapse and by relapse rate during follow-up. Only one study18 showed a reduced short term disability progression risk with IFNβ treatment, possibly reflecting an effect on relapse related disability.29

Our review has a number of limitations. First, because data between trials were not comparable, it was not possible to perform quantitative analyses of important outcomes, such as cognitive impairment and brain atrophy. MRI measures of MS activity (CU lesions) were available only for small subpopulations of two studies18 ,19 that reported a favourable impact of IFNβ on those measures. Overall, these findings are consistent with IFNβ being effective in reducing the risk of clinical and paraclinical inflammatory events during the SP stage of MS.

We did not set out, in the present review, to compare the effects of different types of IFNβ: the five studies included in the review used two different IFNβ types (β-1a and β-1b), at differing dosages. The contradictory results of the studies with IFNβ-1b18 ,22 have been discussed by the FDA.3 The efficacy reported by European SG18 was suggested to be related to recruitment of patients with earlier SPMS stage and with more active pre-study and on-study disease than North American SG.30 This interpretation is supported to some extent by our analysis, which revealed a tendency to better outcomes in patients who had experienced pre-study relapses compared with patients without pre-study relapses, and in younger patients compared with older patients with a longer disease duration. Thus the effects of IFNβ seem to be related to disease activity and not to IFNβ type. In this context, it is noteworthy that a recent observational study on relapsing remitting MS patients receiving IFNβ-1b for 16 years found that fewer developed SPMS (28.6% vs 44.4%) and that time from diagnosis to SPMS development was longer (13.8 vs 11.4 years) compared with those who discontinued treatment, although neither effect was significant.31 Other observational studies found that IFNβ treatment reduced the risk of conversion to SPMS after 7 and 9 years of follow-up.32 ,33 Furthermore, early treatment with IFNβ-1b (after a first clinical event suggesting MS) was reported to reduce the risk of clinically definite MS; the risk of confirmed disability progression was not significantly lower in the early treatment group (HR 0.76, 95% CI 0.52 to 1.11; p=0.17).34

Note, however, that all data on the effects on long term disability were from open label extension studies31 ,34 or case series.32 ,33 Such non-randomised studies are highly susceptible to bias, particularly because of differences between comparison groups (selection bias) and lack of study protocol (reporting bias) and their findings must be interpreted with caution.35

The results of this review suggest that the effects of IFNβ on relapse are independent of those on disability progression. An influential study indicated that progression in SPMS was influenced by relapses in the years soon after disease onset.36 More recent studies suggest that the severity of progression is independent of relapses preceding and subsequent to progression37 and that relapses during SPMS have no measurable impact on time to reaching EDSS 6.38 Finally, dissociation between relapse suppression and disability progression has also been shown in SPMS patients treated with immunosuppressive drugs.39

Our findings provide a clearer picture of the therapeutic profile of IFNβ in SPMS: it appears able to reduce relapse frequency but is unable to prevent deterioration over the medium term (3 years). The window of opportunity for IFNβ efficacy seems to be limited to the inflammatory phase of the disease. These conclusions support the European Agency for the Evaluation of Medicinal Products' recommendation that IFNβs are most likely to be useful in patients with active MS,4 and highlight the need for careful clinical evaluation of patients before starting treatment in order to identify those most likely to benefit.

It is unlikely that new trials to compare interferons with placebo in SPMS will be undertaken as attention is now focusing on new agents. Mitoxantrone and natalizumab are approved to control refractory disease in selected MS patients, although both can have potentially life threatening side effects. Until new agents become available for clinical use we believe that the findings of this review justify the use of IFNβ only in selected SPMS patients with active disease in the expectation that the treatment may reduce the risk of disabling superimposed relapses.


We thank Phil You (Biogen Idec, Wellesley, MA, USA), Peter Cornelisse and Elisabetta Verdun di Cantogno (Merk Serono, Geneva, Switzerland) for providing additional data from primary studies (IMPACT and SPECTRIMS, respectively). We also thank Dr Vasiliy Vlassov (Moscow Medical Academy) for translating and assisting with the analysis of the Russian papers. Finally, we thank Donald Ward for help with the English.



  • Competing interests LLM has participated in clinical trials sponsored by Schering and meetings sponsored by Schering, Bayer, Merck Serono and Biogen Idec. SF has received honoraria for lectures and educational activities sponsored by Bayer Schering Pharma, Biogen Idec and Merck Serono; he participated as an investigator in the European SG. MR has received honoraria for lectures and educational activities sponsored by Biogen Idec.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles