Neutralizing antibodies during treatment of multiple sclerosis with interferon beta-1b

Interferon beta-1b (IFNB) was the first approved treatment for multiple sclerosis (MS) in the United States. This compound and other interferon beta preparations are now in wide use for MS worldwide. About 40,000 MS patients have been treated with IFNB in the United States. The acceptance of IFNB for the treatment of MS is based on evidence from a multicenter, blinded, randomized, placebo-controlled study of two dosage regimens of IFNB in 372 patients. The drug reduced exacerbation frequency and markedly decreased new lesion formation as measured by serial MRI scans. ^[1,2] An independent study recently supported the latter finding by measuring the frequency of gadolinium enhancement, a method not used in the multicenter trial. ^[3]

The therapeutic effectiveness of IFNB in MS was, however, diminished in the 35% of patients who became neutralizing antibody-positive (NAB+), as defined in this manuscript. This first became apparent during a detailed analysis of the effects of the presence of NABs completed in early 1995 and was mentioned in the final report of the multicenter trial of IFNB. ^[4] Most of this decrease in therapeutic effectiveness in NAB+ patients occurs late in the second year of treatment.

The purpose of this report is threefold: (1) To present evidence validating the accuracy of the NAB assay used during the multicenter trial. This standard assay for NABs, the only assay method currently available, relies on quantitating inhibition of viral cytopathic effect in tissue cultures. ^[5] (2) To provide additional details of the clinical and MRI effects of NABs. (3) To discuss the implications of these findings.

Methods.

The general methodology in the US/Canadian multicenter trial from 1988-1993 has been outlined in previous publications. ^[1,2,4] Certain procedures deserve emphasis, however:

Results.

Effect of NAB positivity on relapse rates and disease progression.

After 18 months, in the 8-MIU treatment arm, NAB+ patients had relapse rates significantly higher than NAB- patients, and very similar to those of placebo-treated patients Table 2. On the other hand, there was less accumulation of disability in 8-MIU NAB+ patients Figure 1. Although the fractional changes in EDSS in Figure 1 are not significant, the difference in EDSS change between NAB+ and NAB- patients is of suggestive significance in the third year (p = 0.083). This difference was not seen in the 1.6-MIU treatment group: mean advance in EDSS at the year-3 endpoint for the 77 NAB- patients was 0.47, and for the 48 NAB+ patients it was 0.45. A Kaplan-Meier curve could not be derived for this data because patients changed status during the study--the number of NAB+ patients increased over time.

View this table:

• View inline
• View popup
• Download powerpoint

Table 2. Neutralizing antibodies and the annual relapse rate in the interferon-beta 1b trial

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Neutralizing antibodies and progression of disability. Mean change in EDSS score from baseline.

MRI effects.

In the 8-MIU group, the number of enlarging lesions in annual MRI scans was significantly larger for NAB+ patients, in comparison with NAB- patients, in the second and third years, p = 0.03 and p = 0.01 Table 3, but remained smaller than in the placebo arm. Similarly, Table 3 shows that there was a trend toward more new lesions in the NAB+ patients, compared with NAB- patients, in year 3, p = 0.067. These patterns were not noted in the 1.6-MIU treatment arm. Year-to-year increases in MRI lesion burden were also greater in NAB+ than in NAB- patients in both the 1.6- and 8-MIU treatment arms after the first year Figure 2, but these differences were not significant.

View this table:

• View inline
• View popup
• Download powerpoint

Table 3. NABs and mean number of enlarging and new lesions in annual MRIs in IFNB trial

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Neutralizing antibodies and median percent year-to-year change in MRI lesion burden.

Discussion.

Two types of antibodies occur in patients treated with interferons: binding antibodies and a subset of binding antibodies termed "neutralizing antibodies" (NABs). ^[7-9] Binding antibodies, commonly measured by enzyme-linked immunosorbent assays (ELISA) or the Western blot technique, develop at some time in the majority (97%) of treated patients. ^[10] Many of these antibodies do not impair therapeutic effectiveness and have no known function, presumably because they bind to a portion of the IFN molecule that is not involved in activating interferon receptors. Such binding antibodies do not necessarily predict later formation of NABs. ^[9] NABs, on the other hand, are formed in a minority of patients, but are capable of reducing or abrogating the normal biological and treatment effects of interferons. Currently they can only be measured by a viral cytopathic effect reduction assay in tissue cultures. ^[5,8]

The retesting of aliquots of representative sera for NABs in an independent reference laboratory indicates that the original assay used in the pivotal IFNB multicenter trial was reproducible. Further testing of 4th and 5th year sera for NABs is now under way.

Diminution of treatment effect due to NABs is well known during alpha interferon treatment, ^[7] especially in those responsive hematologic malignancies such as hairy-cell leukemia, ^[11,12] and also in chronic hepatitis C. ^[13-15] Reduction of the biological effects of interferon beta was seen in 29% of patients with melanoma after development of NABs; in these patients, reduction of surrogate markers of interferon activity such as neopterin, 2 prime 5 prime oligoadenylate synthetase, and beta₂ microglobulins occurred in association with NAB positivity. ^[16]

Why some patients develop antibodies to interferons and others do not is poorly understood. Most studies show minimum formation with intravenous use. ^[8] In phase-I studies of IFNB-1b, neutralizing antibodies were present in 41% of patients treated by repeated intramuscular injections of interferon and in 38% of those treated by the subcutaneous route. ^[10] In our patients, we examined the hypothesis that perhaps the patients with the most rapidly progressive MS would be overrepresented among the antibody formers, but this could not be substantiated. The rate of progression of the disease prior to study entry, as judged by the ratio of EDSS to disease duration, was approximately similar in NAB+ and NAB- patients, as were the exacerbation rates in the 2 years prior to study entry. NAB formation was slightly more common in women than in men, but the difference was not statistically significant: in the 8-MIU treatment arm, 36% of female patients became NAB+ compared with 32% of male patients.

The precise level of a biologically significant serum NAB titer remains uncertain. However, the criterion of positivity used in this study (two successive levels >or=to20 NU/mL) gave a very low positive rate in the placebo arm (2/123) and served to discriminate for the primary outcome measure (relapse rate) between NAB+ and NAB- patients. The empirical adoption of this criterion of NAB positivity included the presumption "once NAB-positive, always NAB-positive." This hypothesis, and a possible differential clinical effect based on NAB titer, should be studied in the future in larger numbers of patients. Our study did not permit such a detailed subgroup analysis, according to titers, because of the relatively small numbers of patients in each subdivision.

Patients having two consecutive NAB titers of 20 NU/ml often had subsequent samples that were NAB-negative. Patients having titers of 500 NU/ml or higher seldom had subsequent negative titers. The reasons for fluctuation of titers between "positive" and "negative" in patients with lower NAB titers are not well understood. However, a short time interval between the last previous dose of IFNB and serum collection may be a factor in some cases. The optimum time to collect serum for NABs following the most recent IFNB injection has not been established; as noted, in this trial the postinjection interval varied from 12 to 36 hours.

The attenuation of treatment effect in MS patients who developed NABs (as defined in this study) during treatment with IFNB was most definite in regard to relapse rate, which reverted to the placebo level. However, there was also a statistically significant loss of treatment effect on enlarging MRI lesions in NAB+ patients compared with NAB- patients, and a trend toward increased new lesion formation.

Although treatment benefit is reduced or abolished in NAB+ patients, there is no evidence that NAB+ patients ever fare significantly worse than placebo-treated patients. In Table 2, one can see a gradual reduction in annual exacerbation rates in the placebo treatment arm with time. This is attributable to several factors. A gradual reduction in exacerbation rates is part of the natural history of MS patients. ^[17] A tendency to regress to the mean also plays a role. In addition, a major factor in this data was selective dropout of the patients with the highest exacerbation rates from the placebo group early in the trial, a factor mentioned in an earlier report; the mean annual exacerbation rate for dropouts in the placebo arm was 1.6 and in the 8-MIU arm was 1.02. ^[4] We believe that the latter is the main reason for seeing a somewhat less, though similar, decline in annual rates in the NAB+ patients in the 8-MIU treatment group in Table 2.

The tendency of NAB+ patients to show less progression of disability is difficult to explain, and perhaps does not need explanation since the finding was not statistically significant. The trend is counterintuitive in view of the increase in relapse rates and the MRI data in these same patients; also, it was not seen in the low-dose treatment arm. However, since a similar tendency was noted in the recent multicenter trial of interferon beta-1a (IFNB-1a), it is mentioned here and should not be dismissed without further study. In that trial, 17% of NAB- patients (20/119) developed sustained disability progression, in contrast to 11% of NAB+ participants (4/37). ^[18,19]

Diminution of the therapeutic effects of IFNB in some patients, associated with development of NAB positivity, should lead to attempts to overcome the problem. Theoretical possible approaches might be (1) a dosage increase, as in some cases of insulin-resistant diabetes mellitus, (2) intermittent therapy with corticosteroids, a method found effective in some cases of insulin-resistant diabetes, ^[20] (3) concomitant treatment with an immunosuppressive drug, or (4) a change to a different beta- or alphainterferon preparation. Investigation of the possible efficacy of any of these approaches will be impossible until an assay for IFNB NABs becomes readily available in the clinic. The first two methods have not been systematically studied at all, and cotreatment with IFNB and immunosuppressant drugs has not been reported.

The fourth possibility of changing to another type of interferon beta has also not been investigated. Both IFNB-1b and IFNB-1a are currently marketed for use in MS and are rather similar 166-amino-acid molecules. IFNB-1b has a serine substitution at position 17, and IFNB-1a is glycosylated at the asparagine residue at position 80. ^[21] Otherwise, the molecules would appear to be identical, especially in that portion that attaches to interferon cell receptors, the region spanning residues 32-56. ^[22] Redlich et al., ^[22] using mouse antibodies, found that antibodies neutralizing IFNB recognized this linear epitope. Based on this evidence, it would seem likely that NABs developed in response to IFNB-1b would also neutralize IFNB-1a (and vice versa), but direct tests to confirm this need to be done.

Finally, in some instances IFN NABs disappear with time, despite continuous treatment. Steis et al. ^[12] noted this phenomenon in 8 of 10 patients treated with interferon alpha.

A recent consensus conference on the treatment of MS patients with IFNB-1b suggested that patients not doing well on IFNB-1b therapy, because of frequent exacerbations or steady increase in disability, should discontinue the drug. ^[23] Decisions to discontinue treatment are difficult to make based on clinical criteria alone, especially for therapies that are only partially effective. Certainly one piece of evidence that might help clinicians make such a decision would be the serum titer of NABs. In making this judgment, a reliable NAB assay that measures NABs--not binding antibodies such as those measured by ELISA or Western blot methods--is essential. One cannot accurately guess which patients have significant levels of NABs on the basis of clinical symptoms or titers of binding antibodies. While clinical side effects are less marked in groups of NAB+ patients, variability is high and the rate of side effects is sufficiently low in both NAB+ and NAB- patients after the first year, so accurate estimates are impossible on this basis. Guessing also risks the possibility of removing patients from therapy who remain NAB- and continue to benefit fully from treatment, even if some MS activity may continue.

Recommendations of a Guidelines Committee for the proper use of an assay for NABs are included in this issue of Neurology. ^[24] An accurate assay for use in the clinic should be available for physicians having patients who take IFNB-1b in the very near future.

The emphasis on NABs in this publication should not obscure the fact that, for the majority (65%) of MS patients, IFNB therapy continues to confer a meaningful treatment benefit. For example, the relapse rate in NAB- patients is reduced to 35 to 50% of the placebo rate (see Table 2), and the average treated patient's MRI lesion burden does not significantly advance during 4 years of treatment. ^[4]

Appendix.

The IFNB MS Study Group comprised the following participating institutions, principal investigators (non-italic type), and investigative teams: Hopital de Notre Dame (Montreal)-Pierre Duquette, MD, M. Girard, and R. Dubois; Jefferson Medical College-Robert L. Knobler, MD, PhD, Fred D. Lublin, MD, and Leith Kelley, RN; Montreal Neurological Institute--Gordon S. Francis, MD, Mark Freedman, MD, Yves Lapierre, MD, and Stanley Hum; Temple University Hospital-Jeffrey I. Greenstein, MD, Bibhuti Mishra, MD, and Nina DeLillio, RN; University of Alabama School of Medicine-John Whitaker, MD, and Beverly Layton; University of Arizona-William A. Sibley, MD, Joan Laguna, and John Krikawa; University of British Columbia-Donald W. Paty, MD, Joel J. Oger, MD, Lorne F. Kastrukoff, MD, Wendy Morrison, RN, and Jill Nelson, RN; University of California at San Francisco-Douglas Goodin, MD, Steven M. Massa, MD, PhD, and Elena Gutteridge, RN; University of Chicago-Barry G.W. Arnason, MD, Avertano Noronha, MD, Anthony T. Reder, MD, and Roberta Martia, RN; University Hospital, London, Ontario-George C. Ebers, MD, George P.A. Rice, MD, and Jane Lesaux, RN; University of Maryland-Kenneth P. Johnson, MD, Hillel S. Panitch, MD, Christopher T. Bever, MD, and Kathleen Conway, RN.

The University of British Columbia MS/MRI Study Group includes David K.B. Li, MD; Donald W. Paty, MD; E.L. Tanton, MD; Guo Jun Zhao, MD; Borys Flak, MD; Andrew Riddebough, BSc; Keith Cover, MSc; Brenda Rhodes, BSc; Andrea Cook, BSc; Karen Smith, RT; Lesley Costley, RT; Sylvia Renneberg, RT; Trudy Shaw, RT; and Monique Genton, RT; Vancouver, BC, Canada. External Advisory Committee: Chairman, Stanley van den Noort, MD, University of California, Irvine; Brian Weinshenker, MD, Mayo Clinic; William Weiss; Stephen Reingold, PhD, National MS Society; Wayne Taylor, McMaster University. Berlex Laboratories Inc.: Joy Constine Wallenberg, MD; Lisa Bedell.

Data Analysis and Manuscript Committee: Chairman, William A. Sibley MD; George C. Ebers, MD; Fred D. Lublin, MD; Donald W. Paty, MD; Anthony T. Reder, MD.