You are here
Long-term impact of interferon or Glatiramer acetate in multiple sclerosis: A systematic review and meta-analysis
Multiple Sclerosis and Related Disorders, Volume 6, March 2016, Pages 57 - 63
In recent years the impact of disease-modifying drugs on long-term progression in multiple sclerosis (MS) was assessed both in observational studies and in extension of randomized controlled trial (RCT). Aim of this work was to quantitatively summarize by a meta-analysis the long-term impact of immunomodulatory drugs (Interferon-Beta (IFN-β) or Glatiramer Acetate (GA)) in relapsing-remitting (RR) MS patients.
We collected all published observational studies reporting the long-term efficacy of IFN-β or GA in RRMS patients. The primary outcome was the treatment effect on progression to a sustained EDSS score of 6 or to the Secondary Progressive (SP) phase. A non-parametric approach was adopted to test the overall treatment effect significance, while a random effect model was used to obtain a pooled quantitative estimate of the treatment benefit, in terms of hazard-ratios (HR) or Relative Risks, with their 95% confidence interval (CI).
Fourteen studies, on a total of 13,238 RRMS patients, were included in the meta-analysis. All studies but two reported a consistent effect of immunomodulatory treatment on long-term disease progression; the pooled effect on progression to EDSS 6 or SP was significant (p<0.01) when tested by the non-parametric test. The quantitative estimate of the treatment effect in reducing progression to EDSS 6 in the subset of studies reporting this outcome was HRpooled=0.49 (95% CI: 0.34–0.69), p<0.001.
Treatment with immunomodulators seems to reduce long-term probability of disability progression. Additional well-designed observational studies could help to confirm these findings.
- Long-term efficacy of disease-modifying treatments on disability progression.
- Heterogeneity in study design of observational studies.
- Well-designed observational studies could confirm the long-term benefit of treatments.
Keywords: Disability progression, Long-term, Interferon-beta, Observational studies, Meta-analysis, Treatment effect.
A large number of observational studies has been recently conducted to verify whether the efficacy of the disease modifying drugs approved for the treatment of multiple sclerosis (MS), demonstrated in randomized clinical trials (RCT), can be translated into a clinically significant delay in disease progression over the long term in real-world settings (Trojano et al, 2007, Trojano et al, 2009, Veugelers et al, 2009, Patrucco et al, 2010, Bergamaschi et al, 2012, Shirani et al, 2012, Drulovich et al, 2013, Tedeholm et al, 2013, and Cocco et al, 2015). All of these observational studies evaluated the impact of different preparation of Interferon beta (IFN) or Glatiramer Acetate (GA), the drugs that accumulated more than 20 years of observation, on the risk of reaching high levels of disability. A recent review (Sormani and Bruzzi, 2015) summarized the evidence coming from observational studies in MS, focusing on a critical revision of all the several potential biases that can affect such studies, where subjects are not randomly assigned to the treatment. In the present study we tried to quantitatively summarize, through a formal meta-analysis, the effect of IFN or GA as estimated in these observational studies. Despite these studies are affected by biases that can be both in favor or against the treatment effect, according to the different designs used, and the results are somehow discordant, it could be useful to have a quantitative overall estimate of the average tendency emerging from the whole picture, in an era when new drugs started to be extensively used and their effects will accumulate in the near future sufficient follow up time to be studied in the long-term.
2.1. Search strategy and selection criteria
We searched electronic databases (Ovid MEDLINE [1950-19 March 2015], PubMed [1965-19 March 2015] and The Cochrane Library with the same time limits), to identify studies fulfilling the following inclusion criteria: observational studies or extension of randomized controlled trials (RCT) reporting information on the long-term effect of Interferon-β (IFN) or Glatiramer acetate (GA) in patients with Relapsing Remitting Multiple Sclerosis (RRMS).
We used search terms for the disease (“multiple sclerosis”) and treatment (“Interferon”, “glatiramer acetate”, “immunomodulating”, “immunomodulatory”, “disease-modifying”), combining then terms for indexing articles in Medline/Pubmed (Interferon OR immunomodulating OR immunomodulatory OR disease-modifying) AND (long-term OR disability progression) AND (multiple sclerosis).
No language restriction was used.
Abstracts were independently screened by two reviewers (AS and MPS) and relevant information were extracted from the full papers. To find any additional studies, reference lists of included studies and contingent systematic reviews were evaluated.
2.2. Data extraction
Data extraction was done independently by two authors (AS, MPS) and the accuracy of extraction was validated by consensus.
For each study, data were collected on year of publication, follow-up length, criteria to define the treatment and the control group, criteria to define the treatment start and the duration of treatment, endpoint and measures used to assess the treatment effect. In the extensions of RCT the proportion of patients evaluated at the end of the study as compared to those originally randomized (retention fraction) was also extracted. Treatment effect estimates along with their 95% Confidence Intervals (CI) were extracted where explicitly reported or derived from figures and Tables where possible.
The meta-analysis was based on two primary endpoints, that are those most frequently used and reported as primary outcomes in long-term assessment of treatment benefit: the time to an Expanded Disability Status Scale (EDSS) score of 6 and the time to progression to the Secondary Progressive (SP) phase of MS.
Time to reach an irreversible EDSS score of 4 was also considered as secondary endpoint.
4. Statistical methods
The pooled analysis was conducted at different levels: first a rough evaluation on the overall benefit of treatment vs no treatment was run including the largest available evidence, even if heterogeneous. In this evaluation the primary endpoint was the probability to reach EDSS 6 or SP MS during the study follow up; both observational studies and long-term extension of clinical trials were included, reporting treatment effect estimates as hazard ratios (HR) (when the time to EDSS 6 or SPMS was the endpoint), or relative risks (RR), (when the proportion of patients reaching EDSS 6 was the endpoint). The statistical analysis was based on a non-parametric sign test; this simple test gives a p-value testing the null hypothesis that superiority or inferiority of treated arms vs placebo were randomly distributed across studies. As a consequence, no pooled estimate of treatment effect is given in this analysis, but each study is classified according to the direction and statistical significance of treatment effect, as +1 when showing a significantly favorable treatment effect, as 0 when showing a treatment effect non significantly different from 0, or as −1 when the treatment was significantly unfavorable. A more quantitative assessment of statistical significance was run on the same set of studies using a one-sample Wilcoxon Signed Rank (WSR) test, based on the ranks of the log-transformation of original effect sizes (log Hazard Ratio or log Relative Risk), then weighted according to the inverse of their variance.
Separate quantitative analyses were run on studies based on the same endpoint (time to EDSS 6, time to SP conversion, time to EDSS4) and using the same treatment effect estimate (HR). HRs with their 95% Confidence Intervals (CI) were extracted from each study and a random effect model was applied to obtain the pooled estimates of treatment effect and its significance.
Since multiple arms from the same study creates “clustered data”, each contrast coming from a study with more than two arms was given a lower weight, according to a procedure previously described (Sormani et al., 2009).
Heterogeneity among studies was quantified by the I2 coefficient (Higgins and Thompson, 2002).
The statistical package Stata (v.11; IBM Corp.) was used (“metan” routine) to run the meta-analyses and draw the forest plots while the software R (v.3.0.3) was used to calculate the weighted ranks and perform the nonparametric tests.
5.1. Characteristics of included studies
10 observational studies (Trojano et al, 2007, Trojano et al, 2009, Veugelers et al, 2009, Patrucco et al, 2010, Bergamaschi et al, 2012, Shirani et al, 2012, Drulovich et al, 2013, Tedeholm et al, 2013, Cocco et al, 2015, and Goodin et al, 2011) (Table 1) and 4 long-term extension of RCT (Rudick et al, 2005, Kappos et al, 2006, Bermel et al, 2010, and Ebers et al, 2010) (Table 2), including a total of 13,238 patients were selected for the analysis (Fig. 1). The median follow up time was 8.5 years (range=4.5–21 years). All the studies evaluated the effect of IFN or GA vs no treatment; in the observational studies the control group was represented by contemporary untreated patients (6 studies (Trojano et al., 2007; Patrucco et al., 2010; Bergamaschi et al., 2012; Shirani et al., 2012; Drulovich et al., 2013; Cocco et al., 2015)), historical untreated patients (2 studies (Shirani et al., 2012; Tedeholm et al., 2013)), patients with a delayed start of treatment (Trojano et al., 2009) or low exposure to treatment (Goodin et al., 2011). One study had two (both contemporary and historical) control groups (Shirani et al., 2012) and one study (Veugelers et al., 2009) compared the EDSS accumulation before and after the treatment start.
|Study||Follow up||Treatment group||Control group||Survival time||Endpoint||Treatment effect|
|Trojano et al. (2007)||Median 5.7 years||IFNβ (n=1103)– sc-IFNβ-1b (n=180, 16.3%), im-IFNβ-1a (n=444, 40.3%), sc-IFNβ-1a (n=479, 43.4%)||Contemporary untreated (n=401)||Time from first visit||Time to EDSS 6 (n=1378)||Yes – statistically significant|
|Time to SP (n=1328)|
|Time to EDSS 4 (n=1246)|
|Time from date of birth||Time to EDSS 6 (n=1378)||Yes – statistically significant|
|Time to SP (n=1328)|
|Time to EDSS 4 (n=1246)|
|Trojano et al. (2009)||Median 4.5 years||Early IFNβ (n=310) –sc-IFNβ-1b (n=22, 7.1%), im-IFNβ-1a (n=124, 40%), sc-IFNβ-1a (n=164, 52.9%)||Late IFNβ (n=2260) – sc-IFNβ-1b (n=181, 8%), im-IFNβ-1a (n=814, 36%), sc-IFNβ-1a (n=1266, 56%)||Time from treatment initiation||Time to EDSS 6 (n=2570)||Yes – trend favouring IFN|
|Time to EDSS 4 (n=2277)||Yes – statistically significant|
|Time from date of birth||Time to EDSS 6 (n=2570)||Yes – trend favouring IFN|
|Time to EDSS 4 (n=2277)||Yes – statistically significant|
|Veugelers et al. (2009)||24 years observation period; Median: 5.8 yrs||After IFNβ/GA introduction –sc-IFNβ-1b (25.1%), im-IFNβ-1a (19.7%), sc-IFNβ-1a (25.9%), GA (21.8%)||Before IFNβ introduction||Time from onset||Time to EDSS 6||Yes – statistically significant|
|Time to EDSS 4||Yes – statistically significant|
|Patrucco et al. (2010)||9 years||IFNβ (n=160)||Contemporary untreated (n=50)||Time from onset||Time to EDSS 6||Yes – statistically significant|
|Time to SP||Yes – statistically significant|
|Bergamaschi et al. (2012)||Median 16.6 yrs (treated), 18.3 yrs (untreated)||IFNβ or GA (n=306)||Contemporary untreated (n=284)||Time from diagnosis||Time to SP||Yes – statistically significant|
|Time to EDSS 6||Yes – statistically significant|
|Shirani et al. (2012)||Median: 5.1 years (IFNβ), 4 years (contemporary), 10.8 years (historical)||IFNβ (n=868)||Contemporary untreated (n=829)||Time from IFNβ treatment eligibility||Time to EDSS 6||No – trend favouring untreated|
|Historical untreated (n=959)||Yes – trend vs IFN|
|Goodin et al. (2011)||21 years||sc-IFNβ-1b – high exposure||sc-IFNβ-1b – low exposure||Time from treatment initiation||Time to EDSS 6||Yes- statistically significant|
|Time to SP||Yes- statistically significant|
|Drulovich et al. (2013)||7 years observation period; Median follow-up: 6 yrs||IFNβ (n=236) – sc-IFNβ-1a (n=133; 56.4%), sc-IFNβ-1b (n=103, 43.6%)||Contemporary untreated (n=183)||Time from onset||Time to EDSS 6||Yes- statistically significant|
|Time to SP||Yes- statistically significant|
|Time to EDSS 4||Yes- statistically significant|
|Tedeholm et al. (2013)||12 yrs observation period||IFNβ (n=730)||Historical untreated (n=186)||Time from onset||Time to SP||Yes- statistically significant|
|Cocco et al. (2015)||Median 12 years||IFNβ/GA (n=1389)||Contemporary untreated (n=539)||Time from onset||Time to EDSS 6||Yes- statistically significant|
|Original RCT||Study||N analysed/total enrolled (retention fraction)||Follow up (years)||Treatment group||Control group||Endpoint||Treatment effect|
|Multiple Sclerosis Collaborative Research Group (MSCRG)||Rudick et al. (2005)||160/172 (93%)||8||im- IFNβ-1a||Placebo||proportion of patients with EDSS>=6||Treatment: 29.1%||Non-significant benefit of treatment|
|Treatment vs Placebo: p=0.09|
|Bermel et al. (2010)||136/172 (79%)||15||im- IFN-1a||Placebo||Proportion of patients reaching EDSS 6.0||Treatment: 47.8%||Non-significant benefit of treatment|
|Treatment vs Placebo: p=0.23|
|Proportion of patients reaching EDSS 4.0||Treatment: 73.9%;|
|Treatment vs Placebo: p=0.48|
|PRISMS||Kappos et al. (2006)||382/560 (68%)||8||sc- IFNβ-1a 44 μg||Placebo||Time to EDSS 6||Treatment: Time: 3.5 yrs||Significant benefit of treatment on delaying the time to progression to EDSS milestones|
|Placebo: Time: 1.7 yrs|
|Treatment vs Placebo:|
|Time to EDSS 4||Treatment: Time: 5.8 yrs|
|Placebo: Time: 3.5 yrs|
|Treatment vs Placebo:|
|The IFNβ Multiple Sclerosis Study Group||Ebers et al. (2010)||270/372 (72.6%)||16||IFNβ-1b 250 μg||Placebo||Proportion of patients reaching EDSS 6.0||Treatment: 44/96 (45.8%)||No treatment effect|
|Placebo: 36/79 (45.6%)|
|Treatment vs Placebo: p=0.98|
|Proportion of patients reaching SP||Treatment: 42/96 (43.8%)|
|Placebo: 34/79 (43%)|
|Treatment vs Placebo: p=0.91|
Time considered is the 10th percentile of time to progression for EDSS 6 and the 20th percentile of time to progression for EDSS 4.
In the extensions of RCT the experimental group was treated with different preparations of IFN-β while the control group was the one originally randomized to placebo; since after the study completion all the placebo patients were switched to IFN-β, in RCT extensions the comparison was between a delayed vs an immediate IFN-β treatment start.
The time to reach each milestone in observational studies was evaluated from disease onset (5 studies (Veugelers et al., 2009; Patrucco et al., 2010; Drulovich et al., 2013; Tedeholm et al., 2013; Cocco et al., 2015)), from disease diagnosis (1 study (Bergamaschi et al., 2012)), from time of treatment start (2 studies (Trojano et al., 2009; Goodin et al., 2011)), from time of treatment eligibility (1 study (Shirani et al., 2012)) or from time of first visit (1 study (Trojano et al., 2007)).
When multiple doses of IFN-β were reported, only the approved dose was considered in the analysis (Kappos et al, 2006 and Ebers et al, 2010). The effect of treatment on progression to EDSS 6 was reported in 13 studies (Trojano et al, 2007, Trojano et al, 2009, Veugelers et al, 2009, Patrucco et al, 2010, Bergamaschi et al, 2012, Shirani et al, 2012, Drulovich et al, 2013, Cocco et al, 2015, Goodin et al, 2011, Rudick et al, 2005, Kappos et al, 2006, Bermel et al, 2010, and Ebers et al, 2010), while the effect on conversion to SPMS was reported in seven studies (Trojano et al, 2007, Patrucco et al, 2010, Bergamaschi et al, 2012, Drulovich et al, 2013, Tedeholm et al, 2013, Goodin et al, 2011, and Ebers et al, 2010).
eTable 1 (Supporting information-Online only) show a quality assessment of observational studies included. The quality was overall good, especially in the statistical methods to account for differences between treatments groups. On the other hand, blinded evaluations or procedures for an objective assessment of outcomes were missing in all the examined observational studies.
5.2. Treatment effect estimate
5.2.1. Overall qualitative evidence
All observational studies but one (Shirani et al., 2012) reported a benefit of IFN-β or GA (both evaluating treated vs untreated cohort or an early vs a delayed treatment start) on the long-term probability to reach EDSS 6.
Among the four long-term extension of RCT, three (Rudick et al, 2005, Kappos et al, 2006, and Bermel et al, 2010) consistently reported a benefit of an early start with IFN-β (that reached statistical significance in one of them (Kappos et al., 2006)) on the probability to reach EDSS 6, while one (Ebers et al., 2010) detected no difference between the early and the delayed treatment start.
Overall the benefit of IFN-β resulted to be statistically significantly (Fig. 2), both using the qualitative approach (Sign test; p=0.0034) and using the test based on the weighted ranks of effect sizes (WSR=117.1; p<0.01).
5.2.2. Time to EDSS 6
Eight observational studies (Trojano et al, 2007, Trojano et al, 2009, Veugelers et al, 2009, Patrucco et al, 2010, Shirani et al, 2012, Drulovich et al, 2013, Cocco et al, 2015, and Goodin et al, 2011) and 1 RCT extension (Kappos et al., 2006), including a total of 11,261 patients, reported quantitative data of treatment effect on time to EDSS 6.
The pooled treatment effect estimate (HRpooled=0.49 (95% CI: 0.34–0.69); p<0.001; Fig. 3) indicated a significant reduction of 51% in the risk of reaching the EDSS 6 milestone due to treatment.
Removing from the analysis the RCT extension (Kappos et al., 2006) did not change the result (HRpooled=0.49 (95% CI: 0.33–0.72); p<0.001).
5.2.3. Time to SP
Six observational studies (Trojano et al, 2007, Patrucco et al, 2010, Bergamaschi et al, 2012, Drulovich et al, 2013, Tedeholm et al, 2013, and Goodin et al, 2011), including a total of 3723 patients, reported data on time to SP conversions. Results across studies were homogeneous (I2=0%) and the pooled estimate highlighted a significantly lower risk of reaching SPMS among treated patients HR=0.36 (95% CI: 0.29–0.44; p<0.001; Fig. 4).
5.2.4. Time to EDSS 4
5 studies (4 observational (Trojano et al., 2007, 2009; Veugelers et al., 2009; Drulovich et al., 2013)) reported quantitative data on treatment effect on time to EDSS 4, including 5682 patients. All the studies reported a consistent reduction in the risk of reaching this EDSS milestone:
The pooled effect indicated a 44% of risk reduction due to treatment (HRpooled=0.56 (95% CI: 0.46–0.68); p<0.001; eFig. 1).
Recently, many RCTs have demonstrated the efficacy of an increasing number of new drugs in MS. However, the methodological advantages of RCTs that guarantee a high internal validity of such studies in establishing the efficacy of new therapies, translates into limitations in terms of external validity, that is, the generalizability of the results to real life. In particular, in MS, it would be highly relevant a long-term evaluation of the effect of treatments, that is difficult to assess in randomized trials of long duration, both for practical and ethical reasons. A recent review highlighted the methodological challenges of the evaluation of the long-term effects of immunomodulatory drugs for RRMS in observational studies (Sormani and Bruzzi, 2015); the main problems are related to the biases arising when comparing concurrent treated and untreated subjects, drawn from the same population, in a context where all the more sophisticated statistical methods used to control them (multivariate analysis or propensity score matching) are not able to protect from unknown or unmeasured confounders (Sormani and Bruzzi, 2015). Nevertheless, many observational studies on the long-term effect of immunomodulatory drugs (IFN-β and GA) have been published up to now and the available evidence overall seems to provide support to the hypothesis of non-negligible long term benefits associated to an early use of these drugs, despite the potential biases, acting in both directions (favouring or unfavouring treatment).
A quantitative meta-analysis is challenging in this setting, due to the differences in study design, outcome definitions, measures of treatment effect across the published studies. Even when the endpoint is the same, it can be assessed in a different way across studies: the definition of progression to the SP phase of the disease is not uniquely defined and require time to be assessed; also EDSS milestones require a confirmation time that can vary in different studies. For these reasons we run meta-analyses at different levels, starting from a very conservative one, including all the studies and based only on the direction of treatment effect, using a sign test. This kind of analysis only indicate whether there is evidence of a benefit of an early start of IFN-β or GA, without quantifying such benefit. A more quantitative analysis, including only studies using the same outcomes, indicated a large treatment benefit, with a size that can be estimated as ranging from a 50% to a 60% relative reduction in the risk of reaching clinically significant EDSS milestones or converting to SPMS. Moreover, no distinction can be done in this analysis between the effect of GA and IFN, since the proportion of GA treated patients is low (Veugelers et al, 2009 and Bergamaschi et al, 2012), and sometimes not specified (Cocco et al., 2015).
It must be kept in mind that, even if meta-analyses are considered the highest level of evidence, pooling biased studies generates biased meta-analytic results. On the other hand, since many different biases have been recognized to affect the collected studies (Sormani and Bruzzi, 2015), and since these biases can have affected the treatment effect estimates in opposite directions, the overall size of the estimated treatment effect should be diluted rather than increased by this heterogeneity.
In conclusion, in an era when head to head observational studies comparing new with established therapies start to be published (Kalincik et al, 2015, Kalincik et al, 2015, He et al, 2015, Gajofatto et al, 2014, and Lanzillo et al, 2012), it would be useful to have a complete picture of the long term effect of injectable immunomodulatory therapies, that are now the new standard of care to refer to for the assessment of the efficacy of emerging therapies.
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Conflict of interest statement
AS, FG, FB, IM and NDT report no conflict of interest.
Sormani MP. reports personal fees from Merck Serono, Biogen, Teva, Novartis, Genzyme, Roche, Synthon, outside the submitted work.
- Bergamaschi et al., 2012 Bergamaschi R., Quaglini S., Tavazzi E., et al., 2012. Immunomodulatory Therapies Delay Disease Progression in Multiple Sclerosis, Mult Scler., (Epub ahead of print). doi: 10.1177/1352458512445941.
- Bermel et al., 2010 R.A. Bermel, B. Weinstock-Guttman, D. Bourdette, et al. Intramuscular interferon beta-1a therapy in patients with relapsing-remitting multiple sclerosis: a 15-year follow-up study. Mult. Scler.. 2010;16:588-596 Crossref
- Cocco et al., 2015 E. Cocco, C. Sardu, G. Spinicci, et al. Influence of treatments in multiple sclerosis disability: a cohort study. Mult. Scler.. 2015;21:433-441 Crossref
- Drulovich et al., 2013 J. Drulovich, j Kostic, S. Mesarosa, et al. Interferon-beta and disability progression in relapsing-remitting multiple sclerosis. Clin. Neurol. Neurosurg.. 2013;115(1):S65-S69
- Dreyer et al., 2014 A.N. Dreyer, P. Velentgas, K. Westrich, R. Dubois. The GRACE checklist for rating the quality of observational studies of comparative effectiveness: a tale of hope and caution. J. Manag. Care Pharm.. 2014;20:301-308
- Ebers et al., 2010 G.C. Ebers, A. Traboulsee, D. Li, et al. Investigators of the 16-year long-term follow-up study. Analysis of clinical outcomes according to original treatment groups 16 years after the pivotal IFNB-1b trial. J. Neurol. Neurosurg. Psychiatry. 2010;81:907-912 Crossref
- Goodin et al., 2011 D.S. Goodin, J. Jones, D. Li, et al. Establishing long-term efficacy in chronic disease: use of recursive partitioning and propensity score adjustment to estimate outcome in MS. PLoS One. 2011;11:e22444
- Gajofatto et al., 2014 A. Gajofatto, M.R. Bianchi, L. Deotto, M.D. Benedetti. Are Natalizumab and fingolimod analogous second-line options for the treatment of relapsing-remitting multiple sclerosis? A clinical practice observational study. Eur. Neurol.. 2014;72:173-180 Crossref
- Higgins and Thompson, 2002 J.P.T. Higgins, S.G. Thompson. Quantifying heterogeneity in a meta-analysis. Stat. Med.. 2002;21:1539-1558 Crossref
- He et al., 2015 A. He, T. Spelman, V. Jokubaitis, et al. Comparison of switch to fingolimod or interferon beta/glatiramer acetate in active multiple sclerosis. JAMA Neurol.. 2015;72:405-413 Crossref
- Kappos et al., 2006 L. Kappos, A. Traboulsee, C. Constantinescu, et al. Long-term subcutaneous interferon beta-1a therapy in patients with relapsing-remitting MS. Neurology. 2006;67:944-953 Crossref
- Kalincik et al., 2015 T. Kalincik, V. Jokubaitis, G. Izquierdo, et al. Comparative effectiveness of glatiramer acetate and interferon beta formulations in relapsing-remitting multiple sclerosis. Mult. Scler.. 2015;21:1159-1171 Crossref
- Kalincik et al., 2015 T. Kalincik, D. Horakova, T. Spelman, et al. Switch to natalizumab versus fingolimod in active relapsing-remitting multiple sclerosis. Ann. Neurol.. 2015;77:425-435 Crossref
- Lanzillo et al., 2012 R. Lanzillo, M. Quarantelli, S. Bonavita, et al. Natalizumab vs Interferon beta 1a in relapsing-remitting multiple sclerosis: a head-to-head retrospective study. Acta Neurol. Scand.. 2012;126:306-314 Crossref
- Patrucco et al., 2010 L. Patrucco, J.I. Rojas, E. Cristiano. Long term effect of interferon-beta on disease severity in relapsing-remitting multiple sclerosis patients. Rev Neurol.. 2010;50:529-532
- Rudick et al., 2005 R.A. Rudick, G.R. Cutter, M. Baier, et al. Estimating long-term effects of disease modifying drug therapy in multiple sclerosis. Mult. Scler.. 2005;11:626-634 Crossref
- Shirani et al., 2012 A. Shirani, Y. Zhao, M. Karim, et al. Association between use of interferon beta and progression of disability in patients with relapsing-remitting multiple sclerosis. JAMA. 2012;308:247-256
- Sormani and Bruzzi, 2015 M.P. Sormani, P. Bruzzi. Can we measure long-term treatment effects in multiple sclerosis?. Nat. Rev. Neurol.. 2015;11:176-182
- Sormani et al., 2009 M.P. Sormani, L. Bonzano, L. Roccatagliata, et al. Magnetic resonance imagin as a potential surrogate for relapses in multiple sclerosis: a meta-analytic approach. Ann. Neurol.. 2009;65:268-275 Crossref
- Trojano et al., 2007 M. Trojano, F. Pellegrini, A. Fuiani, et al. New natural history of interferon-beta-treated relapsing multiple sclerosis. Ann. Neurol.. 2007;61:300-306 Crossref
- Trojano et al., 2009 M. Trojano, F. Pellegrini, Paolicelli, et al. Real-Life Impact of Early Interferon Therapy in Relapsing Multiple Sclerosis. Ann. Neurol.. 2009;66:513-520 Crossref
- Tedeholm et al., 2013 H. Tedeholm, J. Lycke, B. Skoog, et al. Time to secondary progression in patients with multiple sclerosis treated with first generation immunomodulating drugs and historical controls, adjusted for clinical covariates. Mult. Scler.. 2013;19:765-774 Crossref
- Veugelers et al., 2009 P.J. Veugelers, J.D. Fisk, M.G. Brown, et al. Disease progression among multiple sclerosis patients before and during a disease-modifying drug program: a longitudinal population-based evaluation. Mult. Scler.. 2009;15:1286-1294 Crossref
- Wells et al., 2000 Wells G.A., Shea B., O’Connell D., et al., 2000. The Newcastle-Ottawa Scale (NOS) for assessing the quality of non-randomized studies in meta-analyses, Paper Presented at 3rd Symposium on Systematic Reviews: Beyond the Basics Oxford.
Biostatistics Unit, Department of Health Sciences, University of Genoa, Genoa, Italy
⁎ Correspondence to:, Dep. of Health Sciences (DISSAL), Via Pastore 1, Genova, Italy.
© 2016 Elsevier B.V., All rights reserved.