Neurology -- Correspondence for Koch-Henriksen et al., 66 (7) 1056-1060

Correspondence published:

A randomized study of two interferon- beta treatments in relapsing–remitting multiple sclerosis: Luca Durelli, Pierangelo Barbero, Marinella Clerico for the INCOMIN trial study group (14 July 2006)

Reply from the Authors: Nils Koch-Henriksen, Per Soelberg Sorensen (14 July 2006)

A randomized study of two interferon- beta treatments in relapsing–remitting multiple sclerosis: Douglas S. Goodin (22 May 2006)

Reply from the authors: Nils Koch-Henriksen, Per Soelberg Sorensen (22 May 2006)

A randomized study of two interferon- beta treatments in relapsing–remitting multiple sclerosis 14 July 2006

Luca Durelli,
Dipartimento di Scienze Cliniche e Biologiche, Universita' di Torino, Italy
Divisione di Neurologia, Ospedale S.Luigi, Regione Gonzole, 10, I-10043 Orbassano (TO), Italy,
Pierangelo Barbero, Marinella Clerico for the INCOMIN trial study group
Send Correspondence to journal:
Re: A randomized study of two interferon- beta treatments in relapsing–remitting multiple sclerosis

luca.durelli{at}unito.it Luca Durelli, et al.

Trials with inadequate methodologic approaches are associated with biased estimates of treatment effect. Koch-Henriksen et al recently reported that low-dose (22 µg) once-weekly (OW) interferon-beta-1a (IFNß-1a) was as effective as high-dose (250 µg) every other day (EOD) IFNß-1b. [1]
These results contradict two head-to-head trials, INCOMIN and EVIDENCE (both cited by Koch-Henriksen et al), several clinical studies (eg, OWIMS, PRISMS), pharmacological studies [2], and systematic reviews [3] pointing to a dose or a dose–frequency effect for IFNß.
The CONSORT statement [4] outlines the need for an intention-to-treat (ITT) analysis in randomized clinical trials. Both INCOMIN and EVIDENCE used ITT, while Koch-Henriksen et al used a per-protocol approach in which patients lost to follow-up were excluded.
The main reason for withdrawal in the IFNß-1a arm was treatment failure and for the IFNß-1b arm it was adverse events. The CONSORT statement [4] requires a flow-chart allowing readers to track the progress of patients through the trial. If a flow chart would have been provided in the article by Koch-Henriksen et al, [1] data could be controlled and recalculated using ITT analysis.
Annualized relapse rate was recalculated [1], assuming that dropouts would retain the same relapse rate as if they stayed on treatment, but no difference between treatments was observed. However, relapse rate is sensitive to relapses from patients at the extremes of the Gaussian curve; many relapses occurring in a few patients may substantially change the overall population relapse rate. If the proportion of patients with relapses is examined, as in INCOMIN and EVIDENCE, patients with many relapses are counted as one clinically active patient and all active patients have the same proportional influence on the final count of patients with or without relapses. Using this outcome measure and the ITT approach, the evidence-based medicine parameters of the OW intramuscular IFNß-1a and the EOD subcutaneous IFNß-1b pivotal trials, and of INCOMIN and EVIDENCE, were statistically significant for all but the OW intramuscular IFNß-1a pivotal trial (Table).
This further confirms the AAN and MS Council for Clinical Practice Guideilines conclusions [3] point to a dose, or a dose-frequency effect for IFNB; a statement which has been contradicted by the Koch-Henriksen et al article. [1]
References
1. Koch-Henrikson N, Sørensen PS, Christensen T, et al. A randomized study of two interferon-beta treatments in relapsing–remitting multiple sclerosis. Neurology 2006;66:1056–1060.
2. Rothuizen LE, Buclin T, Spertini F, et al. Influence of interferon beta-1a dose frequency on PBMC cytokine secretion and biological effect markers. J Neuroimmunol. 1999;99:131–41.
3. Goodin DS, Frohman EM, Garmany GP Jr, et al. Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology and the MS Council for Clinical Practice Guidelines. Disease modifying therapies in multiple sclerosis: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology and the MS Council for Clinical Practice Guidelines. Neurology 2002;58:169–78.
4. Altman DG, Schulz KF, Moher D, et al, for the CONSORT Group. The Revised CONSORT Statement for Reporting Randomized Trials: Explanation and Elaboration. Annals Internal Med 2001;134:663–694.
Disclosure: Professor Luca Durelli received honoraria for speaking and for consultancy from Industria Farmaceutica Serono, Italy; Schering AG, Germany; and Dompe’ Biotec, Italy. Dr Marinella Clerico received honoraria for speaking from Schering AG, Germany.

Table Evidence based medicine parameters calculated with the intention-to-treat approach for published trials with IFNß in relapsing remitting multiple sclerosis.

IM IFNß-1a pivotal trial IFNß-1b pivotal trial INCOMIN EVIDENCE*

IFNß-1a 30 µg Placebo IFNß-1b 250 µg Placebo IFNß-1b 250 µg IFNß-1a 30 µg IFNß-1a 44 µg IFNß-1a 30 µg

Number with relapses 126/158 120/143 97/124 116/123 47/96 59/92 126/339 156/338

EBM
RR(95%CI) 0.95(0.85-1.06) 0.83(0.75-0.92) 0.76(0.59-0.99) 0.81(0.68-0.97)
ARR(95%CI) 4%(-4.5-12.9%) 16%(7.7-24.4%) 15%(1.2-19.2%) 9%(1.6-16.4%)
NNT(95%CI) 24(8-INF) 7(5-14) 7(3-85) 12(7-63)

Number with MRI activity 83/120 78/113 84/111 103/115 37/76 55/73 134/339 192/338

EBM
RR(95%CI) 1.0(0.84-1.19) 0.84(0.75-0.96) 0.65(0.50-0.84) 0.70(0.59-0.82)
ARR(95%CI) 0%(-12-12%) 14%(4-24%) 26.7%(11.7-41.6%) 17.3%(9.9-24.7%)
NNT(95%CI) -714(-INF-8) 8(5-25) 4(3-9) 6(4-10)

IM: intramuscular; EBM: evidence based medicine; RR: relative risk; ARR: absolute risk reduction; NNT: number needed to treat; CI: confidence interval; *Data at Year 1; INF: infinite.

Reply from the Authors 14 July 2006

Nils Koch-Henriksen ,
Per Soelberg Sorensen
Send Correspondence to journal:
Re: Reply from the Authors

koch-henriksen{at}stofanet.dk Nils Koch-Henriksen, et al.

We agree with Durelli et al that inadequate methodology leads to biased estimates of treatment effect but our own major concern was lack of blinding rather than the per protocol analysis as opposed to ITT analysis. As we stated, ITT could bias the results in the direction of the null hypothesis as patients dropping out of the IFN-beta 1a arm with treatment failure had the option of continuing treatment with other IFN-beta preparations.
To compensate for the per protocol bias, we recalculated relapse rates under an assumption that cases prematurely dropping out from the study would retain the relapse rate observed under the study. This correction moved the results slightly in the direction of favoring the IFN-beta 1b arm of the study but without significance.
Durelli et al have found this correction unsatisfactory, as they probably are under the impression that we used means of the relapse rates calculated for each individual case. The distribution of such rates would be skewed, and a few patients with many relapses could disproportionately increase the mean of such parameters. However, we did not use this method but pooled all relapses for a group and divided them by the grand sum of patient years of treatment for the same group, a measure far less sensitive to individual deviations.
Durelli et al favor comparing proportions of relapse-free patients, which is free from the above-mentioned bias. However, on the expense of some loss of information.
The proportions of patients free of relapses appear indirectly from our results section, in which we gave the numbers of patients meeting their first relapse on study.
We found no significant difference in proportions of relapse free patients (IFN-beta 1a: 40.8%; IFN-beta 1b: 45.2%).
Again assuming that a group of patients dropping out from the study would retain its relapse rate after discontinuation, an additional nine patients from the IFN-beta 1a arm and seven from the IFN-beta 1b arm would experience a relapse within the 24 months study period. The proportion of relapse free patients would then be 35% for the IFN-beta 1a arm and 41% for the IFN-beta 1b arm (p = 0.13). It is also our opinion that ITT analysis is mandatory in placebo-controlled studies, but using the per protocol analysis rather then the ITT analysis did not cause significant bias in this particular study.
We are aware of the requirement in the CONSORT statement of the flow chart which was included in our originally submitted manuscript. We later removed it due to limited space. However, most of the exact figures are included in the text.
Disclosure: The authors report no conflicts of interest.

A randomized study of two interferon- beta treatments in relapsing–remitting multiple sclerosis 22 May 2006

Douglas S. Goodin,
University of California, San Francisco
505 Parnassus Ave M794, San Francisco, CA 94143-0114
Send Correspondence to journal:
Re: A randomized study of two interferon- beta treatments in relapsing–remitting multiple sclerosis

douglas.goodin{at}ucsf.edu Douglas S. Goodin

I read with great interest the results of the randomized-trial by Koch-Henriksen et al [1] which compared weekly IFNβ-1a 22μg, s.c. (Rebif) with every other day IFNβ-1b 250μg, s.c. in the treatment of relapsing/remitting (RR) MS. There are, however, several additional points of clarification and discussion that the authors need to provide so that the readership can place this study into proper context.
First, this 2-year trial finished enrollment in October 1997 and yet the results of this trial were not received by Neurology until February 2005. [1] The authors need to explain the reason for the inordinate delay in the publication of this material.
Second, the authors currently (and inappropriately) frame their discussion in the context of the EVIDENCE and INCOMIN trials [2,3], each of which compared high-dose and low-dose IFNβ products of proven efficacy. By contrast, in the present study, the authors have compared high-dose IFNβ-1b against a product that was demonstrated in the OWIMS trial [4] to be ineffective (a 0% relapse-rate reduction) in the treatment of RRMS. Thus, their study seems to show that IFNβ-1b 250μg, s.c. is an ineffective agent – a result that contrasts markedly with the placebo-controlled data in RRMS. [5] Consequently, the authors need to focus their discussion on why they feel that their trial failed to replicate this earlier placebo-controlled experience. [4,5]
Third, the group of patients who chose not to be part of the randomized trial (and who received every other day IFNβ-1b 250μg, s.c.) seemed similar at baseline in all respects to trial- participants (Table E1; original paper). Nevertheless, these patients had a significantly greater relapse-rate (p<0.009) and more disease progression (p=0.031) compared to patients treated in the trial. Thus, in this group of patients, treatment with IFNβ-1b actually seemed to be harmful. The authors need to discuss why they feel this might be so and, again, why their experience is so different from the placebo-controlled data. [5]
Fourth, the authors suggest that neutralizing antibodies (NAbs) may have played a role in these results. Actually NAbs are considered in only a very cursory manner in the paper so that the validity of this proposal is unclear. However, if NAbs are the cause, the authors need to explain why the differences (or lack thereof) between the groups were apparent from the very beginning of the trial (Figure 1; original paper). Surely, NAbs did not evolve at the start of the study.
References
1. Koch-Henriksen N, Sørensen PS, Christensen T, et al. A randomized study of two interferon-beta treatments in relapsing-emitting multiple sclerosis. Neurology 2006;66:1056-1060.
2. Durelli L, Verdun E, Barbero P, et al. Every-other-day interferon beta-1b versus once-weekly interferon beta-1a for multiple sclerosis: results of a 2-year prospective rendomised multicentre study (INCOMIN). Lancet 2002;359:1453-1460.
3. Panitch H, Goodin DS, Francis, G, et al. Benefits of high-dose, high-frequency interferon beta-1a in relapsing MS are sustained to 16 months: final comparative results of the EVIDENCE trial. J Neurol Sci 2005;239:67-74.
4. The Once Weekly Interferon for MS Study Group. Evidence of interferon beta-1a dose response in relapsing-remitting MS: the OWIMS Study. Neurology, 1999;53:679-86.
5. The IFNB Multiple Sclerosis Study Group and the UBC MS/MRI Analysis Group. Interferon Beta-1b in the Treatment of MS: Final Outcome of the Randomized Controlled Trial. Neurology 1995;45:1277-1285.
Disclosure: The author reports no conflicts of interest.

Reply from the authors 22 May 2006

Nils Koch-Henriksen,
Department of Neurology, Aalborg Hospital
DK 9000 Aalborg, Denmark,
Per Soelberg Sorensen
Send Correspondence to journal:
Re: Reply from the authors

koch-henriksen{at}stofanet.dk Nils Koch-Henriksen, et al.

Dr Goodin has raised some important questions giving us opportunity to discuss and clarify our views.
The reason for the delay of our publication was due to unanticipated technical difficulties in transferring MRI scans from different image file formats used at the participating centers to a format which could be processed by the semiautomatic software. We used the time to find and engage qualified software engineers. When started, this task proved to be extremely costly and time consuming. Secondly, the authors spent some time discussing how to interpret the unexpected results of the study.
In the process of revising the paper, we had to sacrifice a reference to the OWIMS study [4] and remove parts of the discussion. We believe that the odd results of the present study can be partially ascribed to lack of blinding, and the study may be regarded as a warning against uncritical interpretation of unblinded comparative studies. Lack of blinding may distort the results in the direction of what is expected by the designers of the studies. In the present study patients and investigators may, consciously or unconsciously, have favored the more convenient and far less expensive treatment with once-a-week subcutaneous admini¬stration of IFN-beta 1a (Rebif®), which based on the reported results from the pivotal study of IFN-beta-1a 30 µg i.m. once weekly could be expected to be at least as effective regarding progression of disability as the established treatment with IFN-beta 1b 250 µg every other day. Regarding the OWIMS study, the patients of the IFN-beta 1a qw arm of the OWIMS study had, by chance, a higher baseline mean CU lesion score and a greater BOD than the patients of the placebo arm, indicating more disease activity. In spite of the OWIMS study it is our opinion that IFN-beta 1a 22 mcg qw at least has some effect, and the effect of IFN-beta 1b every other day may be even higher, as we found a trend, however insignificant, in the MRI parameters.
The reason why non-randomized patients treated with INF-beta 1b every other day fared worse than the similar treated patients from the INF-beta 1b arm of the study may be a matter of self-selection. Even if the non-randomized patients on average did not differ significantly from the randomized patients at baseline, decline from randomization may have indicated more pronounced cognitive problems with less capability to make decisions, which may be a sign of a more active disease. It was not a consequence of our study, that treatment is harmful for these patients as suggested by Dr. Goodin. In the non-randomized patients of our study, the annualized relapse rate of 0.85 is in agreement with the 8 MIU IFN- beta 1b treated arm of the original placebo-controlled IFN-beta 1b-study. [5]
Dr. Goodin may have misunderstood our findings and conclusion as to Nabs: We have only called attention to a potential confounding role of NAbs, but NAbs proved not to affect our results (Table 1).
Disclosure: The authors report no conflicts of interest.