Abstract
Background: A European (EU) and a North American (NA) placebo-controlled study with interferon beta-1b (IFNB-1b) in secondary progressive multiple sclerosis (SPMS) showed divergent results with regard to their primary outcome of sustained Expanded Disability Status Scale (EDSS) progression, while effects were similar on relapse and MRI-related endpoints. Reasons for this discrepancy were explored in the combined dataset.
Methods: Baseline characteristics and variability in EDSS assessments were compared. Retrospective combined analyses for time to confirmed progression were performed to assess treatment effects overall and in subgroups defined by pre-study disease activity criteria and other key baseline variables.
Results: The variance of EDSS measurements was 6.5% higher in the NA-SPMS study. The EU study included patients in an earlier phase of SPMS and with more active disease both pre-study (relapses, MRI) as well as on study (EDSS, relapses, and MRI variables as assessed in the placebo groups). The pooled analysis showed an overall risk reduction by about 20% in patients treated with 8 MIU (250 mcg) IFNB-1b for EDSS progression confirmed at 6 months (p = 0.008). Risk reduction by 30% to 40% was found for patients with at least one relapse or change in EDSS by >1 in the 2 years prior to study entry. No other consistent across-studies relation of clinical and MRI variables at baseline to potential treatment response was found.
Conclusions: Although post hoc, this combined analysis of the two large studies with IFNB-1b in secondary progressive multiple sclerosis suggests that both pronounced disability progression and continuing relapse activity might help in identifying those patients in the secondary progressive phase of the disease who are more likely to benefit from treatment.
Since the initial trial of interferon beta-1b (IFNB-1b)1 and subsequent studies with IFNB-1a,2,3⇓ a considerable amount has been learned about the effect of IFNB on multiple sclerosis (MS). It is now well established that the beta interferons reduce the frequency of relapses in patients with relapsing-remitting MS (RRMS).4 MRI data show that IFNB has a profound effect on the occurrence of measures of new disease activity seen on T2-weighted plane and on T1-weighted post contrast enhanced images.3,5–7⇓⇓⇓ Since lesions visualized with these sequences are generally thought to reflect the acute, inflammatory stage of lesion development it has been postulated that the effect of IFNB is primarily on the early inflammation phase of lesion formation.8 The impact of IFNB on disability progression is less clear: an effect on progression was observed in studies of IFNB-1a in RRMS2,3⇓; however, this may have been related to the reduction of relapses and their direct sequelae.
Two large-scale placebo-controlled studies addressed the ability of IFNB-1b to modify progression in the secondary progressive phase of disease (SPMS), where disability progresses irrespective of relapses. The first trial was initiated in Europe in 1994 (EU-SPMS trial).9 This study showed that IFNB-1b significantly slowed the progression of disability in SPMS patients as measured by the Expanded Disability Status Scale (EDSS).10 The study was stopped after all patients had completed at least 2 years in the study due to compelling evidence of efficacy, having exceeded all minimum criteria for early study discontinuation established for the prospectively planned interim efficacy analysis. In 1995, prior to the completion of the European study, a second study of IFNB-1b was begun in North America (NA-SPMS trial). The NA-SPMS study was also halted prior to the planned completion date but in this case because the outstanding clinical data would not have substantially altered the conclusion about lack of effect on the primary outcome measure.11
Subsequently, a third trial studying IFNB-1a in SPMS also failed to demonstrate a significant treatment effect on progression for the entire treatment cohort.12 A fourth study with once weekly IFNB-1a in SPMS has shown a small significant effect on a potentially more sensitive but less easy to understand disability measure, the MS Functional Composite (MSFC), which had been chosen as primary outcome for this study. This study failed to show any effect on sustained EDSS progression, the standard measure used by the three other studies of IFNB in SPMS.13
What could be the reasons for the divergent results of these studies in the effect of IFNB on progression in SPMS? The two similarly designed studies of IFNB-1b using the same drug and administration scheme provide an opportunity to examine possible answers to this question.
Methods.
Comparison of EU- and NA-SPMS study designs.
Key elements of study design are summarized in table 1. Although both studies required a diagnosis of SPMS, the entry criteria differed slightly with respect to verification of recent disease activity in the 2 years prior to enrollment: A one-point increase in EDSS was required for the NA-SPMS trial, whereas either a one-point increase in EDSS or a history of two relapses was required for the EU-SPMS trial. Both pre-study relapses and EDSS progression had to documented in the patients’ charts in the study centers. In the EU-SPMS study, if patients were referred by other physicians to the center, a letter of the referring neurologist, confirming EDSS progression by at least one point, was required in order to accept this criterion of inclusion as fulfilled. In the NA-SPMS at least one prior relapse had to be documented in the patients’ charts. This could include a doctor’s office visit, where the physician described the relapse in a progress note, or a hospital discharge summary describing the patient’s course during a relapse. In order to document progression there had to be a complete neurologic examination, conducted no more than 2 years prior to study entry, and the documentation had to be detailed enough to allow the neurologist of the study center to determine an EDSS score, and to demonstrate >1.0 step increase of the score over the 2 years preceding study entry (0.5 for initial EDSS of 6.0). The primary endpoint was time-to-confirmed progression for both studies; however, the definition of a confirmed progression differed somewhat. The definition employed for the EU-SPMS trial was 3 months sustained progression with censoring of EDSS scores obtained during relapses and that employed for the NA-SPMS trial was 6 months sustained progression without censoring of EDSS scores obtained during relapses. In both studies, an EDSS change of 0.5 points from a baseline EDSS score of 6 and 6.5 was considered a one-point change. The results for disability progression for the respective trials remain the same irrespective of the definition of progression applied. In contrast to the EU-SPMS study which tested 8 MIU IFNB-1b against placebo, the NA-SPMS trial also investigated the effects of flexible dosing according to body surface area in a second active treatment arm, and the placebo group was split in a fixed and flexible “dosing” scheme.
Table 1 Comparison of study design elements in the EU-SPMS and NA-SPMS studies of interferon beta 1b
Datasets and definitions used.
For better comparability, the analysis presented here is based on the respective ITT populations and uses for both studies the NA-SPMS criteria for sustained progression, which according to one recent publication may be more robust.14 Of the NA-SPMS study, for direct comparability only the group with standard dosing of 8 MIU IFNB-1b was used. Tests for heterogeneity of the two NA-SPMS placebo groups, both for baseline characteristics and behavior during the trial, did not yield significant differences, therefore the two groups were pooled. For the EU-SPMS study, the primary data set of the prospectively planned interim analysis which led to early termination was used.
EDSS measurement variability.
In both studies investigators responsible for EDSS assessments were trained in standardized training sessions conducted by one of the authors (L.K.), including discussion of definitions for different gradings of the Functional Systems and EDSS that were then applied to teaching videos. While in the EU-SPMS study initial training was conducted in three separate groups and then repeated twice during the course of the study, many of the North American EDSS physicians were trained on the basis of a video taken from the first initial training session, conducted by L.K. The source documents used were nearly identical and included standardized definitions of terms and gradings.15 Definition of grade 6.0 and 6.5 of the EDSS in the NA-SPMS was only based on the dependence on one (6.0) vs two (6.5) crutches/canes for walking ≥100 m according to a decision of the study’s steering committee; in the EU-SPMS both dependence on uni- and bilateral help and maximal (observed) walking distance were taken into consideration allowing patients who were able to walk >120 m with bilateral assistance to have EDSS 6.0.
Differences in the variability of EDSS measurements between studies were explored. The time-adjusted within-patient variability of all on-study EDSS scores of the respective placebo groups was assessed using a linear regression model with EDSS as dependent variable and time as the independent variable. The difference between the observed EDSS and that predicted from the model was used to compute the within-patient variability for each patient and the average EDSS measurement variability (in terms of the model mean square error) for the placebo groups.
Comparison of baseline characteristics.
Differences of key baseline characteristics of the respective study populations were explored using the t-test or χ2 test, as appropriate.
Retrospective combined analyses.
A combined analysis based on individual patient data for time to confirmed progression was performed using a Cox model in its stratified form, thus allowing for different hazard functions across studies.16 First, the stratified Cox model was fitted assuming a common treatment effect. The fit of this model was compared with the model which allows separate coefficients for the treatment effect using twice the change in maximized log likelihood. The treatment effect was considered to be heterogeneous between the studies in case of a p value (χ2 test) less than 0.10. Hazard ratio estimates were derived together with 95% CIs for the individual and the combined studies, and p values (χ2 test) were provided for treatment group comparisons in the combined analysis.
Retrospective combined analyses in subgroups defined on clinical grounds.
For subgroups defined on clinical grounds (disease activity defined by relapses or rapid disease progression), the question of homogeneous effects across studies within subgroups was evaluated using the same approach as described above, and treatment group comparisons were performed within these subgroups on the basis of the data set.
Post hoc subgroup analysis per study.
Additionally, post hoc subgroup analyses were performed for groups according to key baseline variables. For each subgroup, hazard ratio estimates were derived for each of the two studies, and the question of homogeneous effects across studies within these subgroups was evaluated.
Results.
Comparison of study outcomes.
Outcomes related to disability progression.
The life tables for time to confirmed progression (according to NA definition of progression) for the two studies are presented in figure 1, showing a difference between treatment groups for the EU-SPMS study (p = 0.003, log-rank test with stratification adjustment for baseline EDSS categories [≤3.5; 4.0 to 5.5; ≥6.0]) and no treatment effect in the NA-SPMS study (p = 0.59).
Figure 1. Time to confirmed progression (NA definition of progression). 8 MIU Interferon beta 1b vs placebo (pooled placebo in NA SPMS study). Life table estimates. Figures in brackets denote the number of patients with confirmed progression of the total number of patients.
Both the proportion of patients on placebo progressing and the effect of treatment as compared to placebo differed between the two studies: in the EU-SPMS trial, 46% of the placebo patients had sustained progression as compared to 34% in the NA-SPMS trial. In the EU-SPMS trial, treatment reduced the proportion of patients progressing to 36%, which represents a 21% reduction compared to the placebo group (p = 0.012, Mantel-Haenszel test with stratification adjustment for baseline EDSS category ≤3.5; 4.0 to 5.5; ≥6.0). In contrast, only a 6% (NS) reduction in progression was seen in the 8 MIU treatment arm compared to the placebo arm in the NA-SPMS trial.
Relapse and MRI-related outcomes.
Comparison of relapse and MRI-related outcomes of the two studies is shown in table 2. In distinction to the differences seen in disability-related outcome measures, IFNB-1b appears equally effective in both studies in reducing relapse rate and disease activity as measured by MRI: the relapse rate in the placebo group of the NA-SPMS trial was lower than that in the EU-SPMS (p < 0.001). However, the magnitude of the reduction in relapse rate in the two studies was similar. Also, contrast enhancing lesions were more frequent at baseline in the EU-SPMS cohort than in the NA-SPMS cohort (table 3), but again the relative reduction in acute contrast enhancing lesions was similar.
Table 2 Selected secondary outcomes in the EU-SPMS and NA-SPMS studies
Table 3 Selected baseline characteristics of the EU-SPMS and NA-SPMS study cohorts
In summary, treatment with IFNB-1b yielded similar results in both studies with regard to relapse and MRI activity, but remarkably different outcomes with regard to delay of progression. In a second step, reasons for the discrepancies between studies were explored. One possible explanation could be a less sensitive measurement of disability progression in the NA-SPMS trial. Alternatively, the different outcomes could be due to biologic differences between the two study cohorts, which might impact on the efficacy of IFNB-1b on disability progression. In this case, the profile of putative treatment responders could possibly be described.
Is the difference in the primary outcome of the EU-SPMS and NA-SPMS trials due to variability in EDSS measurements?
Both intra- and inter-rater variabilities in EDSS scoring are well known to clinical investigators working in MS.17 Therefore both clinical trials attempted to minimize variability by providing standardized training to the evaluators.
The possible influence of the slightly different definition of EDSS grades 6.0 and 6.5 that might have resulted in a lower sensitivity for change in the NA-SPMS study was explored by separated analyses of the behavior of patients who had values of 5.5 to 6.5 at baseline or during the trial, and did not reveal any major effect (data not shown).
The examination of the EDSS variability between the two studies by means of a linear regression model showed that the variance was 6.5% higher in the NA-SPMS study as compared to the EU-SPMS study. However, this higher variability in EDSS measurements is not sufficient to explain the differences in progression rates (data not shown).
Are the differences in the primary outcome of the EU-SPMS and NA-SPMS trials due to biologic differences in two study cohorts?
Initial evidence suggesting a biologic difference between the two study cohorts was found in the baseline characteristics of the two populations (see table 3).
As seen in table 3, the study cohort in the EU-SPMS study differed significantly from that in the NA-SPMS study in many measures. Inspection of the characteristics in which the populations differed indicates that the study population entered into the EU-SPMS study had somewhat earlier and more active disease from the standpoint of relapses and MRI activity. One exception was the change in EDSS over the previous 2 years, which was greater in the NA-SPMS cohort.
Differences in the populations were further explored by examining on-study behavior of the placebo groups. The significantly larger proportion of patients progressing in the EU-SPMS placebo group as compared to the NA-SPMS placebo cohort (46% vs 34%, p = 0.004) can also be noted from figure 1. Also, the on-study relapse rate was higher in the EU-SPMS placebo group as compared to the NA-SPMS placebo group (0.63/year vs 0.28/year, p < 0.001, see table 2).
Retrospective analyses across the studies.
Overall treatment effect on disability progression.
The question whether there is an overall treatment effect on disability progression if data of both studies are combined was addressed by means of a retrospective pooled analysis using a Cox model.
There was an overall risk reduction by about 20% in the IFNB-1b treated patients for disability progression, which was significant (p = 0.008) for the pooled analysis. However, and as expected, there was a trend toward heterogeneity of treatment effects between the two studies (table 4 and figure 2).
Table 4 Retrospective analyses in subgroups defined by baseline disease activity: time to confirmed progression
Figure 2. Retrospective combined analyses in subgroups defined by disease activity. Time to confirmed progression. Hazard ratio estimates derived from stratified Cox model.
Relationship of clinical pre-study disease activity and treatment effect on disability progression.
Comparison of baseline characteristics and on study behavior of the placebo groups suggested that there was a preponderance of patients with active disease in the EU-SPMS study where treatment effectively delayed progression. In order to explore whether pre-study disease activity influences treatment effects on disability progression, patients were categorized in subgroups according to pre-study relapses and EDSS progression. Treatment effects were investigated across studies within subgroups, as well as in combined subgroups along the same lines as above.
In both the EU-SPMS and NA-SPMS subgroups with active disease, a consistent pattern for a treatment effect on progression emerged (see table 4 and figure 2). Patients with at least one relapse in the 2 years before inclusion or pronounced EDSS progression (change in EDSS >1 in the 2 years prior to study entry) seemed to benefit from treatment in both studies with a reduction of the risk to experience disability progression by around 30% to 40% in the IFNB-1b treatment arm. The effect was similar in patients who had one or more relapses prior to enrollment and most prominent in patients with most active disease, i.e., who had both relapses and pronounced progression before enrollment. A treatment effect was also suggested for subgroups with pronounced progression without relapses both across studies and for the combined data, although CIs were rather wide due to the smaller number of patients in the respective subgroups (see table 4).
In contrast, patients with rather low pre-study disease activity (patients with relapses and little EDSS progression or patients with no history of relapses and EDSS changes not greater than 1) did not benefit equally from treatment in the two studies. While a modest treatment effect seemed still apparent in the low activity EU-SPMS subgroup, IFNB-1b patients in the respective NA-SPMS subgroup did worse than those on placebo (see table 4).
Relationship of other baseline variables to treatment effect on disability progression across studies.
We also explored the relation of other patient characteristics at baseline to potential treatment response across studies. Subgroups according to age, sex, disease duration, EDSS status at baseline, as well as MRI status at baseline (presence of gadolinium enhancing lesions, restricted to frequent MRI subgroups, and T2 lesion load, all patients) were formed using the median of overall data, where appropriate, to split groups. No consistent picture arose across studies for any of these variables, including the MRI variables. In general, treatment responses were consistent in the EU-SPMS subgroups, but less so across the studies (see figure E-1, on the Neurology Web site at www.neurology.org).
Discussion.
The differences observed in the primary outcome of two clinical trials with similar design and studying the same treatment, IFNB-1b, on patients in the same stage of disease could be related to differences in the conduct of the trial or differences in the study populations. Although the first possibility may have contributed to the differences observed in time to sustained progression both the variability in measurements and the somewhat different definition of EDSS steps 6.0 and 6.5 do not explain the range of difference across studies. The similarity in the effect of treatment on relapse and MRI outcomes and the results of other clinical trials of interferon treatment in SPMS indicate that baseline differences and differences in disease activity in the two study populations contributed the most important part to the disparity in the effect of IFNB-1b on progression.
Differences in the study populations are evidenced by differences in their baseline characteristics. The patients in the EU-SPMS trial were younger, had a slightly shorter duration of disease, had a higher relapse rate in the 2 years prior to entry, and, in the frequent MRI subgroup, a greater proportion had contrast enhancing lesions. Nevertheless, the baseline characteristics of all studies in SPMS clearly reflect characteristics of patients with SPMS (see table E-1 on the Neurology Web site at www.neurology.org).
Several factors may have contributed to the selection of different study populations. Both studies required the diagnosis of SPMS at entry. However, the entry criterion describing recent disease activity differed between the two studies. For the EU-SPMS study, patients were considered to have recent disease activity on the basis of either a history of two relapses or progression of one point on the EDSS scale over the past 2 years. In the NA-SPMS study, patients were required to have at least one point deterioration in EDSS over the past 2 years. Therefore, it is not surprising that the study population in the EU-SPMS study had a higher frequency of relapses over the prior 2 years. Consistent with the higher frequency of pre-study relapses was that the subset of patients who had frequent MRI evaluations in the EU-SPMS study also had a higher number of contrast enhancing lesions at baseline than the respective subset of patients in the NA-SPMS study (see table 3).
There remains the somewhat counterintuitive finding that NA-SP patients seemed to have a higher pre-study progression than EU-SP patients. However, the EU-SPMS placebo group had a higher EDSS progression rate during study as compared to the NA-SPMS study. There are three possible explanations for the observed discrepancy:
The different entry criteria defining recent disease activity in both studies: approximately 40% of patients in the EU-SPMS study were enrolled upon having met the pre-study disease activity criterion of ≥2 documented relapses in the 2 years prior to study entry (see footnote of table 3) and their true previous progression is unknown;
Regression to the mean;
Neurologists recruiting patients for trials may have preferentially documented those aspects of the disease that were important for inclusion in the trial resulting in underestimating the true pre-study progression rate in the EU-SPMS study and in overestimating the true pre-study progression rate in the NA-SPMS study.
Such discrepancies are not unique to these two SPMS studies. If we compare the relapse behavior pre- and post baseline in the SPECTRIMS study12 with that of the EU-SPMS study, relapse rates before entry were clearly lower in the SPECTRIMS study while the relapse rate in the placebo groups during the two studies—documented under GCP conditions—was identical18 (see table E-1).
An additional reason for differences in the selection of patients must be sought in the different conditions the two studies enrolled: the EU-SPMS study enrolled patients prior to approval of IFNB or other treatments for RRMS in Europe. In contrast, IFNB-1b was approved in the United States for RRMS 2 years prior to the start of the NA-SPMS study, and IFNB1a and glatiramer acetate were approved during the enrollment phase. Further, a large North American multicenter trial targeting SPMS was enrolling patients at the same time.19 Thus the conditions under which the two studies enrolled may have made it less likely that the NA-SPMS study would include patients who were in the SPMS stage of disease but still having relapsing activity or aggressive disease; these patients would have been more likely to have been treated open label with IFNB as standard care. Similar effects have probably impacted on patient enrollment in other, subsequent studies of IFNB in SPMS.12,13⇓ The patient baseline characteristics in the SPECTRIMS study of subcutaneous IFNB-1a,12 which also failed to demonstrate a significant treatment effect on EDSS progression, were somewhere between the NA- and EU-SPMS studies, while the IMPACT study of IM IFNB-1a,13 which showed an effect on the MSFC but not on the EDSS, recruited patients more similar to the NA-SPMS study (see table E-1).
These observations raise questions about the validity of baseline statements on previous disease activity and have major implications on attempts of pooling clinical pre-baseline data from different studies even if these are very similar in their design. In-depth analysis of explicit and implicit selection and documentation criteria and—where available—consideration of placebo group behavior during the first phase in the study is necessary before such pooled data can be used. This is important for ongoing work to define prognostic indicators or to create “virtual placebo groups” for future clinical trials as currently attempted at the Sylvia Lawry Centre for MS Research.20
Over the past several years considerable progress has been made in our understanding of the natural history of the MS lesion.21,22⇓ It is now well established that an early, if not initial event in the development of a new lesion is disruption of the blood–brain barrier (BBB) that can be identified on post contrast-enhanced MRI studies. Good evidence exists indicating that most acute lesions associated with disruption of the BBB are associated with an acute inflammatory process. The degree of tissue damage seems to differ between patients during the acute stage of lesion development: some patients have substantial resolution of clinical findings following an exacerbation while others are left with permanent deficits. Studies of the pathology of lesions indicate that some lesions are characterized by loss of the oligodendrocyte and irreversible myelin loss while other lesions are characterized by partial remyelination. Further, the extent of axonal damage may vary. Disruption of axonal integrity seems to occur very early in the disease as has been reported recently.23 In contrast to the acute phase of disease, MRI studies have demonstrated that patients in the progressive stage of the disease are less likely to have evidence of acute disruption of the BBB, suggesting that progression of tissue destruction may proceed independently of overt inflammatory processes.
With regard to the baseline characteristics of the two study populations in the EU-SPMS and NA-SPMS studies, it is likely that the EU-SPMS study population was slightly earlier in their disease and had progression more closely related to the acute inflammatory component of the disease. Consistent with this hypothesis is that differences between the study populations are even more accentuated if we consider study behavior of the placebo groups: both progression and relapse rate were considerably higher in the EU-SPMS placebo group compared with the NA-SPMS and IMPACT studies, and to a lesser extent the SPECTRIMS study (see table E-1).
Was there an overall treatment effect of IFNB-1b on disability progression? The retrospective pooled analysis for the primary endpoint using a Cox model showed a modest significant treatment effect across the studies, although the result is largely driven by the EU-SPMS cohort.
Can we use differences in the study populations to help understand the characteristics of a patient most likely to respond to interferon treatment? This question was addressed in subgroups defined by clinical disease activity criteria or other key baseline variables again using a Cox model. Two baseline characteristics seem to contribute consistently to greater treatment effect in both studies: evidence of relapse activity or progression of EDSS by more than one point over the prior 2 years. The hazard ratios for patients with these characteristics or combinations of these characteristics are very similar for each study individually as well as for the combined studies and treatment effects were homogenous. This indicates that IFNB-1b treatment reduced the risk of disease progression in patients with active clinical disease as defined above in both the EU-SPMS and the NA-SPMS study. However, in contrast to results of retrospective subgroup analyses of the SPECTRIMS study,12 a treatment effect of similar magnitude was seen in patients with or without a history of relapses in the prior 2 years in the EU-SPMS study.24 The differences between the studies in this regard indicate that the presence or absence of a treatment effect cannot simply be related to a recent history of relapses. A substantial treatment effect was also observed in the subgroup of patients with progression of greater than one point in EDSS regardless of history of relapses during the prior 2 years. Regression to the mean should not be a possible explanation for the greater treatment effect in the subgroup with more active disease over the prior 2 years, since the treatment effect was observed in studies with a parallel group randomized design. The placebo group with these characteristics had an even higher progression rate on study than that seen in the placebo group for the entire study. Possibly, pronounced progression also represents the inflammatory phase of the disease rather than the more degenerative phase.
Subgroups defined by other key baseline variables including MRI yielded results which were either inconsistent across the studies or showed heterogeneous treatment effects.
Although the history of more active disease prior to randomization appears to identify patients who are most likely to have a treatment effect, the differences between the EU-SPMS and NA-SPMS cohorts seem to be greater than captured by these characteristics. If the on-study behavior with respect to relapses is examined in the two studies after stratifying for relapses in the 2 years prior to entry, an important difference is still found. EU-SPMS placebo patients without a history of relapse were much more likely to experience a relapse on-study as compared to the NA-SPMS cohort (data not shown). Overall, the EU-SPMS study population seems to have disease characteristics that are probably associated with the more acute inflammatory component of the illness than that of the NA-SPMS cohort; however, these differences are not fully identified by easily quantified baseline clinical characteristics.
While there was a higher rate of baseline gadolinium enhancements and also a higher T2 lesion load in the EU-SPMS study, these two baseline MRI measures did not emerge as independent distinguishing characteristics for treatment response. The lower number of patients, where information about gadolinium enhancement was available, together with the well-known high variability of this measure, provides only part of the explanation. This finding is in accordance with the lacking correlation between gadolinium enhancement and short term disability progression that was already shown in a meta-analysis of contrast enhanced MRI studies.25 Our observation further supports the assumption that inflammation as depicted by focal contrast enhancements but also increased T2 lesion load, especially in the later phase of the disease, does not play an important role in the development of disability.21,22⇓
The results of the NA-SPMS study suggest that there might be a population of SPMS patients with rather slow progression and no history of recent relapses that may not have any substantial benefit from treatment. The possibility that patients without evidence of active disease may even do worse on treatment needs to be considered. The failure to define such a population of patients with possibly negative balance in the EU-SPMS study and the width of the CIs as depicted in supplementary figure E-1 do not support the theoretical possibility of an even harmful effect of treatment on subgroups of patients. The full characteristics that would allow the prospective identification of that population remain to be defined.
Finally, it needs to be stressed that reduction in progression even in patients who seem most likely to benefit was measured over a relatively short duration in relationship to the overall course of the disease; the effect of treatment on the long-term progression of disease in a patient is not known. As we have learned from studies with IFNB or studies of immunosuppressive therapies,26,27⇓ elimination of the inflammatory component of the disease does not necessarily stop progression. In fact, one of the most important lessons learned from the comparison of the two studies of IFNB-1b in SPMS might be that mechanisms of progression differ and that treatments that can delay progression related to high (inflammatory) disease activity may be ineffective in modifying progression related to ongoing destruction independent of, or remote to, the acute inflammatory component of the illness. The problem remains for the clinicians how to reliably identify possible treatment responders. Despite the known limitations of retrospective analyses, the data presented here suggest that pronounced disability progression or continuing superimposed relapses in the secondary progressive phase of the disease could help in identifying patients who may benefit from treatment. By analogy—but less clearly supported by the data due to the more diverging results in the two studies—the opposite features (no superimposed relapses, slow pace of disability progression) may help characterize patients who would not respond to treatment with IFNB1b.
Appendix
Members of the EU-SPMS Steering Committee: L. Kappos, University Hospital Basel, Switzerland; C. Polman, Free University VU Medical Centre, Amsterdam, Netherlands; C. Pozzilli, University of Rome “La Sapienza,” Rome, Italy; A.J. Thompson, The National Hospital Queen Square, London, UK; F. Dahlke, M. Ghazi, K. Wagner, K. Beckmann (Statistical Advisor), Schering AG, Berlin, Germany.
Members of the NA-SPMS Steering Committee: D. Goodkin, Mount Zion MS Center, San Francisco, CA; A. Miller, Maimonides Medical Center, New York, NY; L. Myers, UCLA Medical Center, Los Angeles, CA; H. Panitch, Fletcher Allen Health Care MS Center, Burlington, USA; D. Paty, University of British Columbia, Vancouver, Canada; W. Sibley, University of Arizona Health Sciences Center, Tucson; B. Weinshenker, Mayo Clinic, Rochester, MN.
Members of the EU-SPMS Advisory Board: H. McFarland, National Institute of Health, Bethesda, MD; J. Petkau, Vancouver, Canada; O. Sabouraud, Rennes, France; K. Toyka, Würzburg, Germany.
Members of the NA-SPMS Advisory Board: S. Cook, Newark, NJ; T.R. Fleming, Seattle, WA; H. McFarland, Bethesda, MD; S.C. Reingold, New York, NY; J. Wolinsky, Houston, TX.
Acknowledgments
The European and North American SPMS Studies with IFNBeta-1b were supported by Schering AG and Berlex Pharmaceuticals. The departments of L.K. and C.P. have received research grants and compensation for participation in clinical studies in excess of $10,000 USD. L.K. is supported by the Swiss Multiple Sclerosis Society.
The authors thank all clinical investigators of the two studies (for full list, see references 9, 11, and 20) and the participating patients.
Footnotes
-
Additional material related to this article can be found on the Neurology Web site. Go to www.neurology.org and scroll down the Table of Contents for the November 23 issue to find the title link for this article.
-
↵*Members of the EU-SPMS Steering Committee and the EU-SPMS Advisory Board are listed in the Appendix on page 1787.
- Received December 9, 2003.
- Accepted September 9, 2004.
References
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan