Patient subgroup analyses of the treatment effect of subcutaneous interferon β-1a on development of multiple sclerosis in the randomized controlled REFLEX study

The REFLEX study (NCT00404352) established that subcutaneous (sc) interferon (IFN) β-1a reduced the risks of McDonald MS (2005 criteria) and clinically definite multiple sclerosis (CDMS) in patients with a first clinical demyelinating event suggestive of MS. The aim of this subgroup analysis was to assess the treatment effect of sc IFN β-1a in patient subgroups defined by baseline disease and demographic characteristics (age, sex, use of steroids at the first event, classification of first event as mono- or multifocal, presence/absence of gadolinium-enhancing lesions, count of <9 or ≥9 T2 lesions), and by diagnosis of MS using the revised McDonald 2010 MS criteria. Patients were randomized to the serum-free formulation of IFN β-1a, 44 μg sc three times weekly or once weekly, or placebo, for 24 months or until diagnosis of CDMS. Treatment effects of sc IFN β-1a on McDonald 2005 MS and CDMS in the predefined subgroups were similar to effects found in the intent-to-treat population. McDonald 2010 MS was retrospectively diagnosed in 37.7 % of patients at baseline. Both regimens of sc IFN β-1a significantly reduced the risk versus placebo of McDonald 2005 MS and CDMS, irrespective of McDonald 2010 status at baseline (risk reductions between 29 and 51 %). The effect of sc IFN β-1a was not substantially influenced by baseline patient demographic and disease characteristics, or baseline presence/absence of McDonald 2010 MS.


Introduction
The first presentation of relapsing multiple sclerosis (MS) is the occurrence of an acute first clinical demyelinating event. The risk of new clinical and/or magnetic resonance imaging (MRI) events can be determined by a number of findings, such as the presence of specific oligoclonal bands in the cerebrospinal fluid, subclinical abnormalities in visual evoked potentials, the number of T2 lesions, the number of gadolinium-enhancing (Gd?) lesions and the type of clinical presentation (monofocal or multifocal) [1][2][3][4][5][6][7]. Occasionally, lesions within the central nervous system (CNS) characteristic of MS may be detected by MRI in the absence of clinical symptoms (radiologically isolated syndrome) [8].
The McDonald criteria for the diagnosis of MS have been continually reviewed and updated as necessary since their initial publication in 2001 [9,10]. The most recent version of the criteria (McDonald 2010 [11]) permits the diagnosis of MS at the time of the first clinical demyelinating event without a requirement for a second attack, provided certain MRI findings are met. This means that some patients who would not have been diagnosed as having MS at the time of the first attack (clinically isolated syndrome) using the 2005 criteria would now be diagnosed as having MS on the basis of having both one or more Gd? lesions (not related to the attack) and non-enhancing lesions on MRI.
The REFLEX (REbif FLEXible dosing in early MS; clinicaltrials.gov identifier NCT00404352) trial demonstrated that, in patients with a first clinical demyelinating event at high risk of conversion to MS, interferon (IFN) b-1a, 44 lg subcutaneously (sc) three times weekly (tiw) or once weekly (qw), significantly delayed conversion to McDonald MS (defined by the 2005 criteria) and clinically definite MS (CDMS), and significantly reduced MRI measures of disease activity versus placebo over 24 months [12]. The efficacy of tiw dosing was significantly greater than that of qw dosing for conversion to McDonald MS and MRI outcomes [12]. This paper will report the result of analyses of prespecified patient subgroups in the REFLEX trial. The patient subgroups included those factors used to balance the population during randomization and others that have previously been identified as risk factors for McDonald MS or CDMS. With the revision to the McDonald guidelines [11], it was also important to establish whether sc IFN b-1a was effective in the subgroup of patients who would still have been classified as not having MS by the updated criteria, and also to monitor the time to the second clinical event in patients who would have been diagnosed as already having MS at the time of their inclusion in the REFLEX study.

Methods
The full methodology of this trial has been published previously [12]. Briefly, patients with a first clinical demyelinating event suggestive of MS were randomized 1:1:1 to the serum-free formulation of IFN b-1a, 44 lg sc tiw or qw, or placebo, for 24 months or until a diagnosis of CDMS. An independent adjudication committee was responsible for assessing the eligibility of all patients, confirming or reclassifying each patient's first event as mono-or multifocal and reviewing all relapses that contributed to a diagnosis of CDMS. This committee was blinded to treatment group and comprised neurologists and a neuroradiologist with expertise in MS. The reporting and evaluation of relapses were continuous and independent of scheduled visits. A relapse was defined as new or worsening neurological symptoms, in the absence of fever, lasting for at least 24 h, accompanied by an objective change in the relevant (i.e., symptomatic) functional system, and preceded by 30 days or more of clinical stability or improvement. MRI scans were evaluated centrally at the VU Medical Center, Amsterdam, the Netherlands. The study was undertaken in compliance with the Declaration of Helsinki and standards of Good Clinical Practice according to the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use. Before initiation of the trial at each centre, the relevant institutional review board or independent ethics committee reviewed and approved the trial protocol, patient information leaflet, informed consent forms, and investigator brochure.
The predefined subgroups in this analysis were those used to stratify the randomization process: age\30 or C30 years, use of steroids at first attack, multi-or monofocal presentation of first clinical demyelinating event, presence or absence of Gd? lesions at baseline. Other predefined subgroups were sex and the presence of\9 or C9 T2 lesions at baseline.
In a post hoc analysis, a simulated retrospective diagnosis at baseline of dissemination in time and space according to the McDonald 2010 MS criteria was conducted for each patient. Evidence of dissemination in time was defined as the simultaneous presence of Gd? and nonenhancing lesions (with the exception of patients with brainstem syndromes). Dissemination in space was defined as the presence of C1 T2 lesions in C2 of three MS-typical regions of the CNS; in the original REFLEX MRI analysis, spinal cord location was not assessed and, therefore, only the juxtacortical, periventricular, and infratentorial regions were analyzed [12]. Patients were divided into McDonald 2010-positive and -negative subgroups.
All subgroup analyses were considered to be exploratory.

Statistical analyses
The primary endpoint of the REFLEX study was time to conversion to McDonald MS as defined by the 2005 version of the criteria [10]. The main secondary endpoint was time to conversion to CDMS. As described previously [12], the probability of patients remaining event free over time (from randomization) in each of the three treatment groups was estimated using the nonparametric Kaplan-Meier method for the primary and main secondary endpoints. The hazard ratios (HRs) for between-group comparisons with corresponding two-sided 95 % confidence intervals (CIs) in the prespecified subgroups were estimated using an unadjusted univariate Cox's proportional hazard model with treatment as the only covariate. p values for the comparison between treatment groups within each subgroup were calculated using the same model. Treatment effects within McDonald 2010 MS status strata were analyzed post hoc using Cox's proportional hazard regression models, with treatment as a covariate.

Results
The intent-to-treat (ITT) population comprised all 517 patients randomized to treatment. As previously described, baseline demographic and disease characteristics were similar across treatment groups [12]. Baseline characteristics of the ITT population, stratified by retrospective McDonald 2010 status, are summarized in Table 1.

McDonald MS 2005
Subgroups defined by patient demographics (age \30 vs. In general, the treatment effects of sc IFN b-1a were similar between predefined subgroups and similar to those found in the overall ITT population (Fig. 1a). A significant treatment effect of sc IFN b-1a tiw versus placebo was found in all subgroups apart from the subgroup of male patients. A significant treatment effect of sc IFN b-1a qw versus placebo was observed in the following subgroups: age C30 years, female patients, patients who used steroids for the first clinical event, patients with multifocal presentation, patients with C1 Gd? lesion at baseline, and patients with \9 and C9 T2 lesions at baseline (Fig. 1a, b).
Presence of Gd? lesions at baseline, n (%)

CDMS
In subgroups defined by patient demographics and steroid treatment, estimates of CDMS in placebo-treated patients at 2 years were in the range of 34-41 % (Fig. 2a). In subgroups defined by MRI findings at baseline, estimates of CDMS at 2 years were in the range of 31-46 % in placebo-treated patients (Fig. 2b)   As previously reported [12], time to CDMS compared with placebo was significantly reduced by both dose-frequencies of sc IFN b-1a, with no significant difference between tiw and qw dosing. The treatment effects of sc IFN b-1a in the predefined subgroups were similar to those in the ITT population (Fig. 2).  Table 2).
Baseline characteristics of patients with retrospective McDonald 2010 diagnosis were generally similar to those of the ITT population, with the expected exceptions of a higher mean number of Gd? and T2 lesions and a lower proportion of patients with monofocal presentation (Table 1).

Discussion
Previous studies of patients with a first clinical demyelinating event suggestive of MS have found that certain patient characteristics, such as younger age, more active disease as assessed by MRI, and steroid use during the first event, are predictive of CDMS [7,13]. In the REFLEX study, the primary outcome was The treatment effect on McDonald 2005 MS with both dose-frequencies of sc IFN b-1a in the predefined subgroups was broadly consistent with the findings of the REFLEX ITT analysis [12]. This suggests that the treatment effect of sc IFN b-1a is not dependent on demographic or disease characteristics. An interesting exception to this was the reduced treatment effect in male compared to female patients. However, the small number of male patients in each treatment group might have limited the ability to detect a significant treatment effect.
The treatment effect of sc IFN b-1a on McDonald 2005 MS was generally greater with tiw than with qw dosing, the difference reaching statistical significance in several subgroups. This is consistent with the clear additional benefit of tiw over qw dose-frequency on McDonald 2005 MS, as seen in the ITT population [12].       The results of these subgroup analyses of data from the REFLEX study show that the treatment effects of sc IFN b-1a at both dose frequencies across all subgroups were consistent with the treatment effects in the ITT population, and that there was a clear additional benefit of tiw over qw dose frequency for time to McDonald MS. These treatment effects were also similar across subgroups in which patients were stratified by retrospective diagnosis of McDonald MS by the 2010 criteria, suggesting that had these criteria been available at the time of study design, the main conclusions of the REFLEX study, that sc IFN b-1a reduced the risk of McDonald 2005 MS and CDMS in patients with a first clinical demyelinating event, would not have been affected. Taken together, these analyses add to the previously reported findings of the REFLEX study, in that a consistent treatment effect of both dosing frequencies is seen across all patient groups, with a more robust effect of tiw over qw dosing on McDonald 2005 MS and MRI lesion counts. This further supports the initiation of treatment with sc IFN b-1a at the time of the first clinical demyelinating event.
Conflicts of interest MS Freedman has received personal compensation from Bayer HealthCare, Biogen Idec, EMD Serono (Canada), Genzyme, Novartis, Opexa, Sanofi-Aventis, and Teva Canada Innovation. N De Stefano has received honoraria from Schering,