Antibodies to beta-interferons in multiple sclerosis

Beta-interferon (β-IFN) has become an accepted treatment for patients with relapsing-remitting multiple sclerosis (MS). Physicians in the United States now have two FDA-approved β-IFNs, Betaseron (Berlex Laboratories; Wayne, NJ) and Avonex (Biogen; Cambridge, MA), to offer their patients.^1,2 Treatment with either agent decreases the relapse rate and the severity of individual relapses. β-IFNs are proteins administered by injection and, as with other parenterally administered proteins such as insulin, IFN injections can induce the development of antibodies. Not all antibodies are the same. Some antibodies may merely bind to the protein, whereas others may neutralize the action of the protein. In the case of insulin, antibodies that block its desired treatment effects (referred to as neutralizing antibodies [NAB]) are relatively easy to monitor. In the case of the β-IFNs, the issue of whether antibodies truly neutralize the efficacy of β-IFN for the MS disease process is more complex.

The article by Rudick et al.³ in this issue of Neurology focuses on the questions "To what extent are antibodies formed to β-IFN 1a in patients with MS?" and "Do these antibodies interfere with the clinical efficacy of the drug?" In a selected, well-studied cohort of patients being treated with β-IFN 1a, Rudick et al. report that a minority of patients develop antibodies that may neutralizeβ-IFN function. There was no correlation between progression of disability or relapse rate and the presence of NAB. However, a trend was noted of increased MS disease activity as measured by MRI. Patients with a single NAB titer greater than 20 were considered positive. Patients with NAB displayed suggestively more contrast-enhancing brain lesions on MRI at the end of the second year than did NAB-negative patients (p = 0.062).

β-IFN induces many genes. Rudick et al. report that the presence of NAB correlated with decreased levels of two genes induced by β-IFN: neopterin and beta-microglobulin. These data suggest that NAB affect the biologic activity of β-IFN in vivo. Interestingly, neopterin levels have also been reported to increase during MS disease activity, unrelated toβ-IFN therapy.⁴ Whether neopterin or beta-2-microglobulin might themselves have a beneficial effect, a detrimental effect, or no effect on the course of MS is not known. Because there is no evidence that these two molecules are directly linked to disease pathogenesis, the full meaning of the correlation of NAB with changes in each is unknown.

Rudick et al. also compared the incidence of NAB in patients treated with Avonex (β-IFN 1a) with that of patients treated for comparable lengths of time with Betaseron (β-IFN 1b). They found a lower incidence of NAB in patients taking Avonex. These comparisons, however, involved two patient groups that were different in an important respect. Those treated withβ-IFN 1b were studied after they decided to discontinue treatment, whereas the group treated with β-IFN 1a were beginningβ-IFN therapy for the first time. This difference brings into question whether the existence of NAB in the group discontinuing β-IFN 1b might have somehow influenced the decision to stop the drug. Thus, these data cannot be regarded as fully conclusive with regard to comparisons of NAB incidence between the two agents.

The data reported by Rudick et al. can also be considered along with data derived from a larger cohort of patients taking β-IFN 1b. In 1996, a retrospective analysis indicated that the presence of NAB was associated with a decrease in clinical and MRI efficacy during the pivotal trial of Betaseron.⁵ Patients in the high-dose group (8 million IU[MIU] every other day) who developed NAB to β-IFN 1b appeared to receive less treatment effect with regard to relapse rate and the number of new or enlarging MRI lesions when compared with patients who did not develop such antibodies. In contrast, there was a trend (p = 0.08) of less progression of disability in the NAB-positive patients. Being NAB-positive was defined as having two consecutive titers of at least 20 neutralizing units in an assay based on neutralization of the cytopathic effect of encephalomyocarditis virus on a human lung carcinoma cell line.

In 1997, Petkau and associates at the University of British Columbia used a different approach-a longitudinal approach-to analyze that same data from the Betaseron trial.⁶ These statisticians noted that conclusions drawn from cross-sectional analyses in which NAB-negative patients are compared with NAB-positive patients should be made with caution, because these two groups might have inherent differences. They argued that a more definitive analysis of the data would be a longitudinal analysis to determine whether the change from NAB-negative to NAB-positive in individual patients was associated with lowered clinical efficacy. When up to 5 years of data from the trial of Betaseron was analyzed in this fashion, different conclusions were reached regarding effects of NAB on drug efficacy. A statistically significant increase in annual exacerbation rate (the primary outcome in that trial) was observed in patients developing NAB only in the low dose arm, but not in the higher 8-MIU dose group. A nonsignificant trend of increased relapse rate was noted in the high-dose group. Furthermore, a change from NAB-negative to NAB-positive status was associated with a statistically suggestive improvement percentage change in MRI lesion burden for the high-dose group (p = 0.10). Thus conclusions reached after analysis of these same data using different approaches have not consistently indicated that changing from NAB-negative to NAB-positive status correlates with diminished clinical efficacy.

To use the information presented by Rudick and others, physicians need the answers to several important questions:

• What is a "neutralizing" antibody against β-IFN in MS? This is a complex issue. Numerous mechanisms have been proposed, but precisely how β-IFNs act to benefit MS is unknown.⁷ Because the mechanisms of β-IFN action are not known, we do not have an in vitro assay of true "neutralizing" antibodies and must rely on surrogate assays. Previous and current assays are based on different biologic effects of IFNs. Assays now being used include the cytopathic effect (CPE) assays, myxovirus resistance gene A (MxA) assays, and antiproliferative neutralization assays. When the addition of a serum sample inhibits the effect of β-IFN in each of these assays, it may indicate the presence of NAB. Unless specific steps are taken to detect only antibodies, other inhibitors in serum which are not antibodies can also yield positive results in these assays.

One of the commonly used assays to detect β-IFN NAB is the CPE assay, a measure of viral inhibition. This type of assay was used in the Rudick study and in the Betaseron trial. However, whether viral inhibition confers the beneficial effect of β-IFN on MS is speculative at best. Data collected thus far suggest that β-IFN does not act to benefit MS by inhibiting viral infections.⁸ The MxA assay measures the response of a β-IFN-sensitive gene⁹ and is said to correlate with CPE assay results.¹⁰ Antiproliferative neutralization assays measure inhibition of cell proliferation by β-IFN, usually employing neoplastic cells. Of concern is whether these assays are all equivalent.

• Are antibodies directed against a drug always bad? Antibodies that bind but do not alter function, or alter some but not all functions of a protein, may not be detrimental. In addition, immunologic studies have shown that NAB can have a positive effect by serving as carriers for cytokines, prolonging the half-life for biologic activity and allowing the cytokine to reach sites of action.¹¹ The clinical data suggesting positive effects of NAB on Betaseron efficacy would support such a prolonged duration of cytokine effects.

• Once induced, are NAB permanent, or can they be transient? Can one rely on a single positive NAB titer, or must it be confirmed by a second assay at a later date? Previous studies have suggested that NAB can be transient or intermittent. In the pivotal trial of Betaseron, 11% of patients in the placebo arm had one or more serum titers of 20 U/mL or greater, whereas less than 2% had two consecutive positive titers.⁵ Defining a patient as "NAB-positive," especially if one plans to use this information as part of a rationale to alter medication, may well require two or more positive titers. The 11% incidence of positive NAB titers in patients treated with placebo also suggests that some people develop antibodies to endogenously produced β-IFN.

• What constitutes a relevant level of NAB? Some assays of NAB are more sensitive than others. Both the MxA and antiproliferative neutralization assays are reported to detect NAB in some instances where the CPE assay indicates no NAB.^10,12 Therefore, if a clinically relevant level of NAB can be defined, it may differ from assay to assay.

• What are the long-term effects of NAB? Patients with relapsing-remitting MS are often young. MS is not a fatal disease. β-IFN therapies are not cures. β-IFN is produced naturally by the human body and has several distinct biologic effects, including decreasing cell proliferation, inhibiting viruses, and modulating the immune system. One might speculate that the development of NAB against endogenously producedβ-IFN is especially worrisome. β-IFN NAB have the potential to limit the patient's defenses against viruses and possibly cancers. Malignancy has been described as a late complication of α-IFN¹³ which is closely related to β-IFN. It is not clear whether the malignancies were due to α-IFN itself, its disturbance of the immune system or to antibodies produced in response to the IFN, or to other agents administered to the patients. Fortunately, no increase in malignancies or viral diseases have been associated with β-IFN or the presence ofβ-IFN NAB. Of course, we have used these agents in patients with MS for only 10 years.

The Rudick study provides important and needed additional information that we can add to our knowledge base. Nonetheless, we are left with more questions than answers. Can specific patient characteristics that predispose one to the development of NAB be identified? Do specific β-IFN preparations, dosages, or routes of administration alter the likelihood of developing NAB? Several other treatment studies using β-IFN preparations are being completed in Europe and Canada and may provide additional data. A complete analysis of all data from all studies will hopefully provide more insight. Stored sera should be made available for such an endeavor. In the absence of a prospective, longitudinal study of the effects of changing from NAB-negative to NAB-positive on the clinical outcome of β-IFN therapy, we will not have definitive answers.

What is the practicing neurologist to do? Decisions to discontinueβ-IFN or change therapies are still best made on a clinical basis. The presence of NAB may add support to a decision to alter therapy, but such decisions should not be based solely on NAB presence. If after a period of 6 or more months on β-IFN therapy an individual continues to have relapses of the same severity and frequency as before, a trial of another agent is advisable, particularly if high-titer NAB are persistently present. The possibility that the patient may be underdosed also should be considered. At this time, there is not enough information regarding interpretation of the results of existing β-IFN NAB assays to warrant routine NAB testing.

Acknowledgment

We thank Dr. John Petkau, Department of Statistics, University of British Columbia, for critical review and comments.