Influence of the HLA-DRB1 Genotype on Antibody Development to Interferon Beta in Multiple Sclerosis

Objective  To determine relevant HLA-DRB1 alleles associated with the susceptibility of anti–interferon beta antibody development in a large patient cohort.

Design  In a case-control study, HLA-DRB1 genotyping was performed in a discovery cohort (n = 268) and a validation cohort (n = 825).

Setting  Patients were recruited in Germany by primary care physicians and neurologists and were mainly of Northern European heritage.

Patients  All patients had a diagnosis of multiple sclerosis and were receiving long-term interferon beta therapy.

Main Outcome Measures  The antibody status to interferon beta was determined in all patients by capture enzyme-linked immunosorbent assay and in vivo myxovirus protein A assay and correlated with the HLA-DRB1 genotype.

Results  In the discovery and validation cohorts, HLA-DRB1*04:01, *04:08, *16:01 were identified as genetic markers that are associated with an increased risk of anti–interferon beta antibody development (P < .05). In addition, alleles with a protective potential were identified, including HLA-DRB1*03:01, *04:04, *11:04. However, after correction for multiple testing, protective alleles did not reach statistical significance.

Conclusion  The HLA alleles identified in this study seem to be the major genetic determinant of antibody development, allowing the prediction of the individual risk of patients before initiation of therapy.

Multiple sclerosis (MS) is considered a chronic inflammatory disease of the central nervous system of autoimmune origin.¹ Both genetic and environmental factors are supposed to contribute to the pathogenesis of MS.^2-5 For many years, interferon beta has been a common first-line medication.^6-8 It is approved for the treatment of a clinically isolated syndrome, relapsing-remitting MS, and secondary progressive MS with superimposed relapses.⁹ At present, 3 different recombinant interferon beta preparations are on the market. Like other protein-based disease-modifying agents, interferon beta exhibits immunogenicity.¹⁰ Up to 40% of patients will develop antibodies against interferon beta (binding antibodies), most of them with neutralizing capacity (neutralizing antibodies [NAbs]).^11-13 Binding antibodies may bind to antigenic epitopes of the interferon beta molecule that are not necessarily involved in the interferon receptor activation.¹⁴ In contrast, NAbs interfere with interferon beta binding to its receptor and thus abrogate the bioactivity of interferon beta in vitro and in vivo.¹⁵ As a consequence, therapeutic efficacy of interferon beta diminishes dramatically,^16-20 and to date, no approach has been validated to reverse antibody production.^11,21 It is known that preparation and route of administration can strongly influence antibody development.²² These exogenous factors, however, do not fully explain why some patients will develop antibodies against interferon beta and some will not. Thus, additional host factors determine individual susceptibility to antibody development. The genetic background may be pivotal and immune genes influencing B-cell activation are likely candidates.

Antibody production and generation of B-cell responses to proteins depend on the help of antigen-specific CD4⁺ T cells.²³ The development of antigen-specific CD4⁺ T-cell responses itself is strongly influenced by the individual repertoire of HLA antigen class II molecules, which present peptide antigens for recognition to the T-cell receptor.²⁴ HLA antigen class II alleles are highly polymorphic and therefore likely candidates influencing the development of NAbs.

In a previous study, we performed medium-resolution typing for HLA-A, HLA-B, HLA-C, HLA-DRB1, and HLA-DQB1 alleles in patients with MS receiving long-term interferon beta therapy to investigate the role of HLA alleles in the development of anti–interferon beta antibodies.²⁵ We observed a strong association between HLA-DRB1*04:01 and HLA-DRB1*04:08 and the development of anti–interferon beta antibodies. Furthermore, we observed higher anti–interferon beta antibody titers in HLA-DRB1*16:01 allele carriers whereas HLA-DRB1*11:01 and HLA-DRB1*11:04 allele carriers had lower titers. However, because of the low patient number of our pilot study, no significant correlation was found for these alleles in the initial cohort. In the present study, we investigated the role of these alleles in a large independent validation cohort consisting of 825 patients receiving long-term interferon beta therapy. All patients received high-resolution typing for HLA-DRB1 alleles and were analyzed for the presence of NAbs by capture enzyme-linked immunosorbent assay (cELISA) and in vivo myxovirus protein A (MxA) assay.

In this study, 1093 patients with MS receiving long-term interferon beta treatment were included. Patients were recruited in Germany by primary care physicians and neurologists and were mainly of Northern European heritage. Two patient cohorts were analyzed. The discovery cohort comprised 268 patients and the validation cohort, 825 patients. Written informed consent was obtained from all patients. The study protocol was approved by the local ethics committees of the University of Düsseldorf and the Technische Universität München. The demographic details of the cohorts are displayed in Table 1.

In the discovery cohort, HLA genotyping for DRB1 was performed by hybridization of sequence-specific oligonucleotide probes immobilized on microspheres with amplified genomic DNA samples followed by flow fluorescence-intensity analysis according to the instructions of the manufacturer (One Lambda, Canoga Park, California). In the validation cohort, HLA-DRB1 alleles were determined by either sequence-specific primer polymerase chain reaction using a nested approach, as previously described,²⁶ followed by group-specific sequencing of exon 2 polymerase chain reaction products (Protrans, Hockenheim, Germany) or by sequencing-based typing using generic sequencing of exon 2 of DRB1 followed by ambiguity resolution using group-specific sequencing primers (Invitrogen Life Technologies, Carlsbad, California).

A standardized cELISA, as described by Pachner,²⁷ was performed. Defined positive and negative samples were included in each assay to obtain a standard curve for comparison of different assays. The standard curve ranged from 100% (highest positive control) to 0% (no antibodies). Serum positivity of all tested samples was normalized on the basis of these standard curves. Reactivity higher than 25% of the highest positive control was considered to be antibody positive (the cutoff value corresponded to the median of 50 control donors + 5 SD). The biological in vivo activity of interferon beta was measured by MxA (MX1) gene expression using TaqMan real-time polymerase chain reaction (Applied Biosystems, Foster City, California).²⁸ Twelve hours after interferon beta injection, blood samples were drawn from subjects and messenger RNA was isolated according to the manufacturer's instructions. TaqMan reverse-transcription reagents (Applied Biosystems) were used to convert messenger RNA to complementary DNA. Complementary DNA levels were detected with Applied Biosystems GeneAmp via the protocols described by the manufacturer. The housekeeping gene GAPDH was used as an internal control. Normalization was done according to the delta-delta Ct method. In each assay, the same positive and negative samples were included to ensure the quality of the assay. Antibodies were considered biologically active when the MxA induction was decreased by more than 50% compared with newly treated antibody-negative control donors. Patients were classified as being anti–interferon beta negative, having binding antibodies, and having NAbs. Therefore, patients who were anti–interferon beta negative had no increased reactivity in the cELISA and showed normal MxA induction, whereas binding antibody–positive patients had a positive cELISA result and normal MxA levels. Neutralizing antibody–positive patients had a positive result for the cELISA and a reduced MxA response.

To analyze the difference of antibody levels and MxA induction levels between allele carriers and nonallele carriers, a 2-tailed nonparametric Mann-Whitney test was applied. A P value <.05 was considered to be statistically significant. For comparison of antibody levels between groups, the P values were adjusted for multiple testing with the false discovery rate (FDR).

The discovery cohort included 268 patients who were selected according to their antibody status based on the results of the cELISA and the in vivo MxA assay. The validation cohort consisted of 825 patients who were all tested for their antibody status by both methods (Table 1). All 1093 patients were typed for HLA-DRB1 (Table 2). Heterozygous allele carriers were counted twice and homozygous allele carriers were counted only once (phenotype frequencies). DRB1*15:01, which is highly associated with genetic susceptibility to MS across almost all populations,^26,29 was by far the most frequent allele in both cohorts with 153 allele carriers in the discovery cohort and 434 allele carriers in the validation cohort. This allele had no effect on the development of antibodies to interferon beta or on in vivo bioactivity. However, several HLA-DRB1 alleles had a strong influence on the occurrence of antibodies in interferon beta–treated patients.

Although DRB1*03:01 carriers were rather frequent in both cohorts (48 allele carriers in the discovery cohort and 169 allele carriers in the validation cohort), the results of anti–interferon beta reactivity were inconsistent for both cohorts. In the discovery cohort, the results of the cELISA were not different for DRB1*03:01-positive and DRB1*03:01-negative patients. However, carriers of DRB1*03:01 in the validation cohort showed a lower anti–interferon beta reactivity, with a median of 4.3% (range, −34.1 to 140.3), than nonallele carriers (median, 7.5%; range, −21.5 to 123.1) (Figure 1A). Significance was kept not only for the single test (P value = .006) but also after the FDR procedure (P value = .04). When considering all patients from both groups, the odds ratio (OR) for developing antibodies against interferon beta in DRB1*03:01-positive patients was 0.68 (P value = .04; FDR P value = .05).

Forty-two patients carried the allele HLA-DRB1*04:01 in the discovery cohort and exhibited a significantly elevated median anti–interferon beta antibody reactivity of 66.6% (range, −3.2 to 119.3). In the validation cohort with 102 allele carriers, the influence of DRB1*04:01 was less pronounced with a median anti–interferon beta reactivity of 15.6% (range, −14.2 to 120.4) (Figure 1B). After correction for multiple testing, the result was still significant, with a P value of .03. The OR for DRB1*04:01 in all patients was 2.95 (P value < .001, FDR P value < .001).

Neither DRB1*04:02 (13 patients in the validation cohort) nor DRB1*04:03 (20 patients in the validation cohort) revealed any difference in terms of anti–interferon beta reactivity.

For DRB1*04:04 carriers, the median cELISA levels were lower compared with DRB1*04:04-negative patients in both cohorts (Figure 1C). However, despite a significant single test (P value = .03 in the discovery cohort; P value = .02 in the validation cohort), the null hypothesis was kept for DRB1*04:04 after FDR correction (FDR P value = .16 in the discovery cohort and FDR P value = .10 in the validation cohort). The OR for DRB1*04:04 in all patients was 1.55 (P value = .27; FDR P value = .32).

DRB1*04:05 (7 allele carriers in the validation cohort) and DRB1*04:07 (6 allele carriers in the validation cohort) showed no impact on the cELISA.

In the discovery cohort, which comprised 4 patients with DRB1*04:08, the median anti–interferon beta reactivity was increased to 98.5% (range, 90.5 to 109.8) with a P value of .003 (FDR P value = .048). In the validation cohort, 9 patients were identified as carriers of DRB1*04:08. In these patients, anti–interferon beta reactivity was significantly elevated to 95% (range, −16.7 to 123.1) with a P value of .006 (FDR P value = .045) (Figure 1D). The OR for DRB1*04:08 in all patients was 15.07 (P value < .001; FDR P value < .001).

Genotyping revealed 44 allele carriers of HLA-DRB1*11 in the discovery and 137 allele carriers of HLA-DRB1*11 in the validation cohort. In the discovery cohort, a trend to lower anti–interferon beta reactivity was observed in DRB1*11:01-positive patients. However, in the validation cohort, no difference in anti–interferon beta reactivity was seen in DRB1*11:01 (84 subjects), *11:02 (4 subjects), *11:03 (13 subjects). In the discovery cohort, DRB1*11:04 allele carriers showed a trend toward lower anti–interferon beta reactivity (median, 4.4%; range, −0.8% to 82.8%). In the validation cohort, we identified 36 allele carriers of DRB1*11:04. Comparison of anti–interferon beta titers of HLA-DRB1*11:04 carriers and nonallele carriers showed reduced reactivity in HLA-DRB1*11:04-positive patients (Figure 1E). Median anti–interferon beta reactivity was 2.4% (range, −21.5% to 111.6%; P value = .02). However, after FDR correction, the null hypothesis was kept (P value = .10). The OR of DRB1*11:04 for all patients in both cohorts was 0.98 (P value > .99; FDR P value > .99).

Seventeen allele carriers of DRB1*16:01 were identified in the discovery cohort and 31 allele carriers of DRB1*16:01 were identified in the validation cohort. Comparison of anti–interferon beta antibody titers in HLA-DRB1*16:01-positive subjects and HLA-DRB1*16:01-negative subjects in the discovery cohort revealed an elevated median anti–interferon beta level in patients carrying a DRB1*16:01 allele (median, 90.8%; range, 0.1% to 102.9%; P value = .02). However, the null hypothesis was kept after the FDR procedure. In the validation cohort, the median anti–interferon beta reactivity in HLA-DRB1*16:01 carriers was 72.0% (range, −2.7% to 140.3%; P value < .001). After correction for multiple testing, anti–interferon beta reactivity in HLA-DRB1*16:01 carriers was still significantly elevated (Figure 1F). The OR for DRB1*16:01 was 10.8 (P value < .001; FDR P value<.001).

To further validate our findings and to confirm the impact of HLA alleles on the biological activity of interferon beta in vivo, we correlated the identified risk alleles of the HLA-DRB1 locus with the extent of MxA induction in vivo after administration of interferon beta. In line with the results of the ELISA reactivity, protective alleles showed a trend to higher in vivo MxA induction, whereas alleles with increased susceptibility to develop antibodies revealed lower in vivo MxA induction levels (Table 3). Statistical significance for in vivo MxA induction in the discovery cohort was observed for the alleles DRB1*04:08, *11:04, *16:01. In the validation cohort, only the 2 alleles DRB1*04:08 and DRB1*16:01 reached significance for MxA induction, the latter having also shown the strongest effect sizes in the cELISA assay. However, when all patients were analyzed together, the alleles DRB1*04:01 (P value = .03), *04:08 (P value < .001), *11:04 (P value = .002), *16:01 (P value < .001) reached statistical significance. To evaluate whether the DRB1* genotype itself has an influence on MxA induction capacity, we next performed a correlation of DRB1* alleles and in vivo MxA induction in antibody-negative patients. In the discovery cohort, a significant reduction was observed in DRB1*04:04 carriers (n = 9; median, 81.3%; range, 41.5% to 102.0%; P value = .04). However, after correction for multiple testing, the result did not reach statistical significance (FDR P value = .49) and no association could be observed in the validation cohort. In contrast, MxA induction in DRB1*09:01-positive patients was significantly reduced in the validation cohort but not in the discovery cohort (n = 7; median, 71.9%; range, −8.0% to 104.0%; P = .03). After correction for multiple testing, the result did not reach significance (FDR P value = .72). When we analyzed both cohorts together, we did not observe that the DRB1* genotype had any impact on MxA induction in interferon beta–treated antibody-negative patients with MS.

Finally, we performed a combined analysis taking into account all identified potentially protective alleles (DRB1*03:01, *04:04, *11:04) and risk alleles (DRB1*04:01, *04:08, *16:01) of both patient cohorts (n = 1093). Patients were divided into groups containing 1 or 2 protective alleles, 1 or 2 risk alleles, or none of both. A gene dose-dependent effect was observed for protective and risk alleles in terms of ELISA reactivity (Figure 2A). The OR for patients homozygous for potentially protective alleles was 0.49 (P value = .13; FDR P value = .16) compared with patients carrying neither a protective nor a risk allele. For patients heterozygous for protective alleles, the OR was 0.78 (P value = .19; FDR P value = .19). In contrast, the OR for patients carrying 1 risk allele was 3.8 (P value < .001; FDR P value < .001), and for patients carrying 2 risk alleles, the OR was 16.04 (P value = .004; FDR P value = .01). Combination of a protective and risk allele leads to a slight increase of ELISA reactivity. In this patient group, the OR for developing antibodies was 2.53 (P value = .03; FDR P value = .05).

Comparing MxA induction in these groups of allele carriers confirmed a preserved MxA induction for carriers with a protective allele whereas carriers with a risk allele showed a significantly reduced MxA response (Figure 2B). Patients carrying 2 protective alleles had an OR of 0.40 (P value = .03; FDR P value = .04) for reduced MxA induction compared with patients with no identified risk allele. The group of patients with 1 protective allele had an OR of 0.72 (P value = .06; FDR P value = .06). In patients with 1 risk allele and the other allele being protective, the OR for altered MxA induction was increased at 2.77 (P value = .01; FDR P value = .03). In patients with 1 risk allele and the other not being an identified risk allele, the OR was 1.79 (P value = .001; FDR P value = .006). The highest risk for impaired MxA induction was observed in patients with 2 risk alleles with an OR of 10.9 (P value = .02; FDR P value = .03).

Interferon beta belongs to an increasing number of protein-based pharmaceuticals. Immunogenicity of protein-based pharmaceuticals is a common problem because the development of antibodies might be induced.¹⁰ It has been shown in several studies that NAbs against interferon beta reverse the induction of interferon beta–induced molecules and thereby abolish the biological effects of interferon beta treatment.^15,30,31 As a consequence, therapeutic efficacy measured by magnetic resonance imaging lesion progression, relapse rate, and disability progression declines.^16,17,32

Neutralizing antibodies are supposed to be cross-reactive³³; thus, according to current guidelines, interferon beta therapy is no longer recommended for these patients.^34,35 Unfortunately, to date, no strategy has been established to reverse antibody production and restore the efficacy of interferon beta treatment.^11,21 Even after discontinuation of interferon beta therapy, antibodies will persist in most patients for years.^36,37 Immunogenicity of the interferon beta preparations differs.³⁸ Whereas only 5% to 10% of patients treated with intramuscular interferon beta-1a (Avonex) will develop NAbs, up to 30% of patients undergoing long-term treatment with subcutaneous interferon beta-1a (Rebif) or subcutaneous interferon beta-1b (Betaferon/Betaseron/Extavia) have positive NAb titers. However, the later interferon beta formulations seem to be more effective in suppressing clinical and paraclinical disease activity,^39,40 provided the patient does not develop NAbs.

Because the development of NAbs is one of the major factors affecting the efficacy of interferon beta treatment and has long-standing effects on the use of these drugs, it is important to identify patients who are prone to develop NAbs before initiation of therapy.

In the current study, we evaluated the role of HLA antigen class II molecules for the development of antibodies to interferon beta in a large cohort of patients with MS receiving long-term interferon beta therapy. We have demonstrated that the alleles HLA-DRB1*04:01, *04:08, *16:01 are associated with the development of antibodies against interferon beta therapy in MS. In contrast, HLA-DRB1*03:01, *04:04, *11:04 seem to protect against antibody development. Interestingly, we observed a gene dose effect when taking into account both HLA alleles and the effect on interferon beta antibodies. Patients with 2 risk alleles had higher antibody titers than those with 1 or no risk allele. Patients with 2 protective alleles had lower titers than those with 1 or no protective allele. The presence of a risk allele together with a protective allele lowered the risk of interferon beta antibody development, although titers were still higher compared with those who did not carry any risk allele.

Overall, our findings demonstrate a strong impact of HLA alleles on the development of NAbs. Given the difference in efficacy and immunogenicity of interferon beta drugs, knowledge of the HLA haplotype of the patient might be helpful to guide treatment decisions in patients with MS.

Correspondence: Bernhard Hemmer, MD, Department of Neurology, Klinikum rechts der Isar, Technische Universität München, Ismaninger Strasse 22, 81675 Munich, Germany (hemmer@lrz.tu-muenchen.de).

Accepted for Publication: November 24, 2010.

Author Contributions: The authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Cepok, Hartung, and Hemmer. Acquisition of data: Buck, Cepok, Grummel, Jochim, Wassmuth, and Hemmer. Analysis and interpretation of data: Buck, Cepok, Hoffmann, Berthele, and Hemmer. Drafting of the manuscript: Buck, Cepok, and Hemmer. Critical revision of the manuscript for important intellectual content: Cepok, Hoffmann, Grummel, Jochim, Berthele, Hartung, and Wassmuth. Statistical analysis: Buck, Cepok, and Hoffmann. Obtained funding: Hemmer. Administrative, technical, and material support: Grummel, Jochim, Hartung, Wassmuth, and Hemmer. Study supervision: Hemmer.

Financial Disclosure: Dr Buck has received personal compensation as a speaker from Biogen Idec and Merck Serono and research support from Merck Serono. Dr Cepok has received personal compensation as a speaker from Bayer Vital and research support from Bayer Vital. Dr Berthele received lecture fees and consulting honoraries from Biogen Idec, Bayer Healthcare, Novartis, Merck Serono, and Teva and grant support from Bayer Healthcare. Dr Hartung received honoraria for consulting and speaking and travel support from Bayer Healthcare, Biogen Idec, BioMS, Genzyme, Merck Serono, Novartis, and Teva/sanofi-aventis with approval by the Rector of Heinrich-Heine-Universität Düsseldorf. The Department of Neurology, Heinrich-Heine-Universität Düsseldorf, with approval of the Rector of Heinrich-Heine-Universität Düsseldorf, received unrestricted research grants from Bayer Schering, Biogen Idec, Novartis, and Teva to determine NAb levels in sera from treated patients with MS. Dr Hemmer has received in the past 3 years board membership and/or consulting fees from Novartis, Bayer Schering, Biogen Idec, Merck Serono, Roche, Micromet, and Novartis. He has also received grant support from Novartis, Bayer Schering, Biogen Idec, Merck Serono, and Teva.

Funding/Support: The study was supported by the Federal Ministry of Education and Research (BMBF) (Projects Biobanking and Omics in Control/MS as part of the krankheitsbezogenes Kompetenznetz Multiple Sklerose), DFG grant He2386/7-1, grants from Biogen Idec, Bayer, Serono, and Leipzig Research Center for Civilization Diseases, Universität Leipzig, which is funded by means of the European Union, the European Regional Development Fund, and the Free State of Saxony within the framework of the excellence initiative.