Anti-MDA5 Antibodies in a Large Mediterranean Population of Adults with Dermatomyositis

A new myositis-specific autoantibody directed against melanoma differentiation-associated gene 5 (anti-MDA5) has been described in patients with dermatomyositis (DM). We report the clinical characteristics of patients with anti-MDA5 in a large Mediterranean cohort of DM patients from a single center, and analyze the feasibility of detecting this autoantibody in patient sera using new assays with commercially available recombinant MDA5. The study included 117 white adult patients with DM, 15 (13%) of them classified as clinically amyopathic dermatomyositis (CADM). Clinical manifestations were analyzed, with special focus on interstitial lung disease and its severity. Determination of anti-MDA5 antibodies was performed by a new ELISA and immunoblot technique. In sera, from 14 (12%) DM patients (8 CADM), MDA5 was recognized by ELISA, and confirmed by immunoblot. Eight of the 14 anti-MDA5-positive patients (57.14%) presented rapidly-progressive interstitial lung disease (RP-ILD) versus 3 of 103 anti-MDA5-negative patients (2.91%) (P < 0.05; OR: 44.4, 95% CI 9.3–212). The cumulative survival rate was significantly lower in anti-MDA5-positive patients than in the remainder of the series (P < 0.05). Patients with anti-MDA5-associated ILD presented significantly lower 70-month cumulative survival than antisynthetase-associated ILD patients. Among the cutaneous manifestations, only panniculitis was significantly associated with the presence of anti-MDA5 antibodies (P < 0.05; OR: 3.85, 95% CI 1.11–13.27). These findings support the reliability of using commercially available recombinant MDA5 for detecting anti-MDA5 antibodies and confirm the association of these antibodies with RP-ILD in a large series of Mediterranean patients with DM.

MDA5, also known as interferon-induced helicase-1 (IFIH1), is a member of the retinoic acid-inducible gene Ilike helicase (RIG-I or RLH) family of proteins, [10] which function by recognizing single-stranded RNA viruses and are involved in the innate immune response, including type I IFN production [11].
The main drawback to routine use of this antibody for clinical purposes is that its determination is limited to techniques that are only available in research laboratories, such as immunoprecipitation of radioactive-labeled protein [8] or enzyme-linked immunoassay (ELISA) using in-house fabricated recombinant proteins [12,13].
Our objective was to evaluate the prevalence and clinical manifestations of anti-MDA5-positive patients in a large 2 Journal of Immunology Research cohort of DM patients from a single center in Barcelona, and to determine the feasibility of detecting this autoantibody with the use of more widely available techniques (ELISA and immunoblotting) with commercially available recombinant MDA5 as the antigen.

Patient Population.
This study was performed in 117 adult patients (92 women) with DM (15 with clinically amyopathic DM). In addition, 45 patients with polymyositis (PM), 30 with systemic sclerosis (SSc), and 25 with systemic lupus erythematosus (SLE) were included as controls. Twentyfive healthy controls were also included to determine the cut-off value for establishing the positive status by ELISA. Healthy and disease controls were age and sex matched to the DM patients. The median age of DM patients was 52 years (range 22-81). The patients studied belong to a historical cohort diagnosed with idiopathic inflammatory myopathy at Vall d'Hebron General Hospital in Barcelona (Spain), between 1983 and 2012. Our center is a single teaching hospital with approximately 700 acute care beds, attending a population of nearly 450,000 inhabitants. All myositis patients in this population are referred to our hospital for diagnosis and therapy, regardless of the severity of the disease. Serum samples from these patients are routinely collected at diagnosis and during follow-up in our outpatient clinic, and stored at −80 ∘ C. Patients and controls included in the study gave informed consent for the use of their serum for research purposes. The study was approved by the institutional review board of our hospital.
The diagnosis of DM and PM was based on the criteria of Bohan and Peter [14,15]. Only patients with definite or probable disease were included. The Sontheimer criteria were used to diagnose amyopathic DM [16]. Interstitial lung disease (ILD) was diagnosed according to the consensus classification of idiopathic interstitial pneumonias. The diagnosis of ILD was established by high-resolution CT findings, and rapidly progressive-interstitial lung disease (RP-ILD) was defined as a worsening of radiologic interstitial changes with progressive dyspnea and hypoxemia within 1 month after the onset of respiratory symptoms. [17] Cancer-associated myositis (CAM) was defined as cancer occurring within 3 years of the myositis diagnosis. Patients received treatment with corticosteroids and immunosuppressive drugs (methotrexate, azathioprine, calcineurin inhibitors [cyclosporine A or tacrolimus], or cyclophosphamide pulses) were added when needed; intravenous immunoglobulin was used as an adjuvant therapy, and biological therapy (rituximab) was also instituted when possible in refractory cases. A drug trial was defined as a single course from the beginning of the administration of a given drug to the time at which the drug was discontinued, or in the case of prednisone, as the time at which the dose was reduced to one quarter of the initial dose. Clinical data were obtained retrospectively by review of the patients' medical records.

Laboratory Tests and Serological
Assay. Serum samples from each patient were screened by indirect immunofluorescence for antinuclear antibodies (ANA) using HEp-2 cells, and by a commercial ELISA used in our routine laboratory setting for antibodies against extractable nuclear antigens (Ro, La, RNP, Sm) and anti-histidyl-tRNA synthetase (anti-Jo-1). Anti-TIF1 antibodies were detected by an in-house ELISA and confirmed by immunoblot [18]. In addition, all samples were tested by protein and RNA immunoprecipitation [19], which enabled detection of other synthetases and myositis-specific and myositis-associated antibodies (anti-Mi-2, anti-SRP, anti-Ro52, anti-Ro60, anti-La, anti-PM/Scl, anti-p155, and anti-U1RNP) that may have been overlooked by ELISA, and confirmed the ELISA results.

Anti-MDA5 ELISA.
Briefly, 96-well ELISA plates (NUNC, Kamstrup, Denmark) were coated with 100 ng of purified recombinant MDA5 (OriGene, Rockville, MD), diluted in phosphate buffered saline (PBS), and left to stand overnight at 4 ∘ C. Wells were incubated for 1 hour at room temperature (RT) with blocking buffer (10% nonfat dry milk in PBS). Plates were then washed (HRP Wash, INOVA Diagnostic Inc., San Diego, CA); human serum samples diluted 1 : 100 in blocking buffer were added in triplicate: two to MDA5-coated wells and one to a PBS-coated well (without antigen) to determine the background absorbance. Plates were incubated at RT for 1 hour. After washing, HRP-labeled goat anti-human IgG antibody (INOVA Diagnostic Inc., San Diego, CA) was added to each well, and plates were incubated for 1 h at RT and washed again. Color development was performed with peroxidase reagent TMB Chromogen (INOVA Diagnostic Inc., San Diego, CA) and absorbances at 450 nm were determined. For each sample, the background absorbance from the PBS-coated well was subtracted from that of the corresponding MDA5-coated wells (the average of the two results). Sample absorbance was expressed as optical density units. The same positive serum (from patient 11, confirmed by IP by Casciola Rosen, from Baltimore, USA) was used as the reference in each assay.

Anti-MDA5 Immunoblot.
Briefly, 5 g of purified recombinant MDA5 (OriGene, Rockville, MD) was run on 4% to 12% polyacrylamide-SDS minigels with MOPS running buffer, and western blot was performed on a nitrocellulose membrane using the Invitrogen NuPAGE (Carlsbad, CA) electrophoresis system. [20] MDA5-transferred nitrocellulose was vertically cut into several strips and incubated for 1 hour at RT in PBS with 0.05% Tween (PBS-T) containing 3% nonfat dry milk (blocking buffer). Each strip was then incubated with the corresponding human serum sample diluted 1 : 100 in blocking buffer for 1 hour at RT. After washing, phosphatase alkaline-labeled goat anti-human IgG antibody (Dako, Glostrup, Denmark 1 : 2000) was added to each strip and strips were incubated for 1 hour at RT.

Statistical Analysis.
Associations between anti-MDA5 antibodies and qualitative variables were evaluated with the chi-square and Fisher exact test. The strength of the associations between variables was measured using odds ratios (ORs) with 95% confidence intervals (CIs). The Mann-Whitney test was used for comparisons of median values. The corresponding area under the curve (AUC) of the ROC analysis of anti-MDA5 antibody for detection of RP-ILD, CADM, and total DM was analyzed with 95% of CIs. All tests were two-sided, and probability ( ) values of <0.05 were considered statistically significant. Cumulative survival rates were estimated by the Kaplan-Meier test. The log-rank test was also used to compare survival rates. All analyses were performed with SPSS, version 19.0 (SPSS, Chicago, IL).

Results
One-hundred and seventeen adult DM patients, 15 of whom had CADM, were included in the study. Anti-MDA5 was determined by our in-house ELISA and immunoblot techniques using a commercially available recombinant MDA5. Fourteen patients, 8 with CADM, tested positive for anti-MDA5, which represents a prevalence of 12% of the DM patients from our cohort. Median (range) age at diagnosis of anti-MDA5 positive patients was 47 (28-60) years, which did not differ significantly from the remainder of the cohort. ANA was positive in 5 patients. Seven patients also tested positive to anti-Ro52, but none of them was positive for any antisynthetase antibody. Relevant clinical and immunological findings are summarized in Table 1.

Relationship between Anti-MDA5 and Cutaneous
Manifestations and Cancer. The CADM diagnosis was significantly associated with anti-MDA5 positive status (8 of 14 MDA5positive patients versus 7 of 103 MDA5-negative ( < 0.05; OR: 18.3, 95% CI 4.9-67.6)). No differences were observed in the frequency of Gottron's papules, heliotrope rash, photosensitivity, shawl or "V" sign, cuticular overgrowth, calcinosis, or the presence of mechanic's hands when the anti-MDA5-positive and anti-MDA5-negative groups were compared. Panniculitis was the only manifestation significantly associated with the presence of anti-MDA5 (5 out of 14 anti-MDA5-positive versus 13 out of 103 anti-MDA5negative; < 0.05; OR: 3.85, 95% CI 1.11-13.27); nevertheless, a multivariate analysis was not possible to be performed due to methodological reasons.
Cancer was diagnosed in 4 out of 14 (28.6%) patients with anti-MDA5 autoantibodies, and 3 (21%) of them fulfilled criteria of CAM. However, no association was found between anti-MDA5 autoantibodies and CAM ( > 0.05). A similar result was obtained when we repeated the analysis after excluding from the cohort the anti-TIF1 -positive patients (28 patients) in order to avoid a possible confounding effect of the presence of these patients in the control group.

Survival Rates of Anti-MDA5-Positive Patients.
The cumulative 70-month survival rate was significantly lower (38%) in the group of patients with anti-MDA5 than in the remainder of the cohort (62%) (log-rank test, < 0.05) (Figure 3(a)). Comparison of cumulative 70-month survival between anti-MDA5-associated ILD and antisynthetaseassociated ILD also showed a statistical difference (logrank test, < 0.05) (Figure 3(b)). No differences in 70month survival were observed between the CADM group and the classic DM group in anti-MDA5-positive patients. All groups were comparable in age, gender, and number of immunosuppressive agents added to the corticosteroid treatment. Differences in survival could not be attributed to a higher proportion of "deaths directly related to cancer" between groups.

Discussion
The results of this study prove the feasibility of detecting antibodies against MDA5 in adult patients with DM by in-house ELISA and immunoblot techniques using commercially available recombinant MDA5 as the antigen. In addition, the findings in our patients contribute to support the previously reported association of anti-MDA5 antibody with RP-ILD and CADM. This association has been mainly described in Asian patients. [1][2][3][4][5][6], and the only studies performed in a white population have come from the United States [7][8][9]. Our results in a large series of Mediterranean patients from a single reference center, together with those from other articles published on this topic, [1][2][3][4][5][6][7][8] indicate that anti-MDA5 antibodies may be a hallmark of adult CADM patients with RP-ILD regardless of their origin. Furthermore, some authors have suggested an association between this autoantibody and a specific, severe skin vasculopathy in adult DM, characterized by vascular fibrin deposition with variable perivascular inflammation [3,7,8]. We found an association only with panniculitis, one of the mucocutaneous findings previously described by Fiorentino et al. [8] in white adults with DM and anti-MDA5 antibody.
Although we routinely test all serum samples from patients with inflammatory myopathies by immunoprecipitation assays using 35 S-protein-labeled HeLa cells, until the development of our proposed method, we had not been able to clearly identify patients with anti-140 kDa antibodies (anti-CADM-140; i.e, anti-MDA5). In our analyses, this polypeptide migrated to an area in which almost 70% of sera, including those from normal subjects, immunoprecipitated a weak line around 140 kDa. Hence, until anti-MDA5 was described by Sato et al. [1,6], and we tested these patients with a commercially available recombinant MDA5 by our in-house ELISA and immunoblot techniques, we were unable to recognize this autoantibody. Our positive patients correspond to a period of 30 years of follow-up. Our first patient with CADM and RP-ILD died nearly 20 years ago, and anti-MDA5 was detected in a stored frozen serum sample by our in-house techniques. Thus, these methods could represent a significant advancement in identification of this autoantibody, even in laboratories with standard equipment.
Anti-Ro52 antibodies were present in most of our patients with RP-ILD and anti-MDA5, an association that has only recently been reported in anti-MDA5-positive ILD patients [9]. This fact confers relevant significance on anti-Ro52 as a costimulatory autoantibody, a concept reported in patients with antisynthetase syndrome [21,22].
One patient was positive to both anti-MDA5 and anti-TIF1 . To our knowledge, this is the first description of this association. Coexistence of two different myositis-specific antibodies has been rarely reported [19,23]. Hence, the situation of this patient is intriguing and warrants further investigation.
The clinical course of anti-MDA5-positive DM patients can be divided into three groups. First (and most important from the prognostic perspective), is the group of patients with CADM and RP-ILD, who usually have a poor prognosis and a mortality rate of nearly 50% despite aggressive immunosuppressive therapy and even lung transplantation. Second, the group with CADM and little lung involvement, who show skin manifestations, such as ulcerations, palmar pustules, and perhaps panniculitis, as was reported here. The prognosis does not seem to be unfavorable in this group. And finally, the third group of patients, who have ILD that is not rapidly progressive and shows a disease pattern similar to that of classic antisynthetase syndrome. [9] Differences in the prognosis between the first and the third group may be due to genetic background, and either early immunosuppressive therapy or use of certain drugs, such as calcineurin inhibitors [24].