Utility of Anti–Melanoma Differentiation–Associated Gene 5 Antibody Measurement in Identifying Patients With Dermatomyositis and a High Risk for Developing Rapidly Progressive Interstitial Lung Disease: A Review of the Literature and a Meta-Analysis

Authors

  • Zhiyong Chen,

    1. Affiliated Drum Tower Hospital, Nanjing University Medical School, Nanjing, China
    Search for more papers by this author
    • Drs. Chen and Cao contributed equally to this work.

  • Mengshu Cao,

    1. Affiliated Drum Tower Hospital, Nanjing University Medical School, Nanjing, China
    Search for more papers by this author
    • Drs. Chen and Cao contributed equally to this work.

  • Maria Nieves Plana,

    1. Hospital Ramon y Cajal (IRYCIS) and CIBER Epidemiology and Public Health (CIBERESP), Madrid, Spain
    Search for more papers by this author
  • Jun Liang,

    1. Affiliated Drum Tower Hospital, Nanjing University Medical School, Nanjing, China
    Search for more papers by this author
  • Hourong Cai,

    1. Affiliated Drum Tower Hospital, Nanjing University Medical School, Nanjing, China
    Search for more papers by this author
  • Masataka Kuwana,

    Corresponding author
    1. Keio University School of Medicine, Tokyo, Japan
    • Department of Rheumatology and Immunology, the Affiliated Drum Tower Hospital, Nanjing University Medical School, 321 Zhongshan Road, Nanjing, 210008, ChinaDivision of Rheumatology, Department of Internal Medicine, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo, 160-8582, Japan. E-mail: kuwanam@z5.keio.jp lingyunsun2001@yahoo.com.cn

    Search for more papers by this author
    • Dr. Kuwana holds a patent on an anti-MDA5 antibody measuring kit.

  • Lingyun Sun

    Corresponding author
    1. Affiliated Drum Tower Hospital, Nanjing University Medical School, Nanjing, China
    • Department of Rheumatology and Immunology, the Affiliated Drum Tower Hospital, Nanjing University Medical School, 321 Zhongshan Road, Nanjing, 210008, ChinaDivision of Rheumatology, Department of Internal Medicine, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo, 160-8582, Japan. E-mail: kuwanam@z5.keio.jp lingyunsun2001@yahoo.com.cn

    Search for more papers by this author

Abstract

Objective

To assess the utility of anti–melanoma differentiation–associated gene 5 (anti-MDA5) antibody measurement for predicting a risk for developing rapidly progressive interstitial lung disease (RP-ILD) in patients with polymyositis/dermatomyositis (PM/DM).

Methods

A single-center cohort of 64 consecutive Chinese patients with PM/DM was examined. Serum anti-MDA5 antibody was measured by enzyme-linked immunosorbent assay. For meta-analysis, we searched PubMed and the Institute for Scientific Information Web of Knowledge for original studies that measured anti-MDA5 antibodies in patients with PM/DM. We calculated pooled sensitivity, specificity, diagnostic odds ratio (DOR), and the summary receiver operating characteristic (sROC) curve.

Results

In Chinese patients, anti-MDA5 antibodies were detected in 26 patients with classic DM or clinically amyopathic DM (CADM). Compared with anti-MDA5–negative patients, anti-MDA5–positive patients showed a higher prevalence of RP-ILD (P = 0.001). In a total of 233 patients with anti-MDA5 antibody, derived from 16 studies, a higher frequency of CADM was found in Japanese than in non-Japanese patients (74.7% versus 39.2%; P = 1.2 × 10−7). Meta-analysis revealed that the pooled sensitivity and specificity of anti-MDA5 antibody for RP-ILD was 77% (95% confidence interval [95% CI] 64–87%) and 86% (95% CI 79–90%), respectively. The pooled DOR was 20.41 (95% CI 9.02–46.20) with a favorable area under the sROC curve of 0.89 (95% CI 0.63–0.98).

Conclusion

Detection of anti-MDA5 antibody is a valuable tool for identifying DM patients with a high risk for developing RP-ILD, but the distribution of classic DM and CADM in patients with this antibody varies among ethnic groups.

INTRODUCTION

Polymyositis/dermatomyositis (PM/DM) belongs to the idiopathic inflammatory myopathies and its etiology remains unknown. In addition to muscle involvement (e.g., proximal muscle weakness, elevated serum levels of creatine phosphokinase [CK], myopathic changes of electromyography, and muscle biopsy evidence of inflammation), DM presents some unique cutaneous manifestations, such as heliotrope rashes, shawl (or V-like) sign, and Gottron's papules. The clinical spectrum of DM differs greatly from patient to patient, and the prognosis is largely dependent on complications, especially the underlying cancer or interstitial lung disease (ILD) ([1]).

Recently, a subtype of PM/DM patients having hallmark cutaneous manifestations of DM but without clinically significant muscle involvement has been defined as clinically amyopathic dermatomyositis (CADM) ([2]). A significant proportion of patients with CADM, mainly from eastern Asia, develop rapidly progressive ILD (RP-ILD), which represents a poor response to intensive therapy, such as high-dose corticosteroids and immunosuppressive agents, leading to fatal respiratory failure ([2]). A retrospective cohort study from China showed that the majority of CADM patients with ILD revealed RP-ILD courses and the 6-month survival rate was only 41% ([3]), demonstrating the poor prognosis of this clinical entity.

To date, several myositis-specific autoantibodies (MSAs) have been identified in PM/DM. Recent studies have revealed that PM/DM case stratification according to MSAs is useful because these antibodies are closely associated to the clinical phenotype ([4]). The anti-CADM140 antibody was first identified by Sato et al in Japanese patients with CADM and RP-ILD ([5]). The autoantigen recognized by anti-CADM140 antibody was later identified as RNA helicase, encoded by melanoma differentiation–associated gene 5 (MDA5) ([6, 7]). Following the original report by Sato et al ([5]), there have been an increasing number of reports describing the diagnostic and prognostic usefulness of anti-MDA5 antibody in Japanese patients with CADM and RP-ILD. In a single-center, cross-sectional study, Fujikawa et al found that all of 8 anti-MDA5 antibody–positive patients showed RP-ILD, and survival at 6 months from diagnosis was significantly lower in patients with anti-CADM140 antibody than in those without it ([8]). Gono et al also confirmed that the frequencies of RP-ILD and fatal outcome were significantly higher in the DM subset with the anti-MDA5 antibody ([9]).

In the present study, we examined the utility of anti-MDA5 antibody measurement for identifying RP-ILD in Chinese patients with PM/DM. We also performed a literature review and meta-analysis of published studies to assess the overall accuracy of anti-MDA5 antibodies for identifying RP-ILD in PM/DM patients.

Box 1. Significance & Innovations

  • We measured anti–melanoma differentiation–associated gene 5 (anti-MDA5) antibodies in a single-center cohort of Chinese patients with polymyositis/dermatomyositis (PM/DM) and examined clinical characteristics of patients with anti-MDA5 antibodies.
  • We also performed a literature review and a meta-analysis to assess the utility of the anti-MDA5 antibody measurement for identification of clinically amyopathic dermatomyositis (CADM) and rapidly progressive interstitial lung disease (RP-ILD) in patients with PM/DM.
  • A literature review indicates that anti-MDA5 antibody is primarily detected in patients with CADM in cohort studies conducted in Japan, but is frequently found in patients with classic DM in non-Japanese cohorts.
  • A meta-analysis reveals that anti-MDA5 antibody measurement would be a valuable tool for identifying PM/DM patients with a high risk for developing RP-ILD and facilitating early intensive treatment.

MATERIALS AND METHODS

Subjects.

In total, 64 patients with PM or DM were recruited consecutively from March 2010 to August 2011 from the Affiliated Drum Tower Hospital, Nanjing University Medical School, China. Diagnoses were based on the Bohan and Peter criteria for probable or definite PM and classic DM ([10, 11]). Patients with CADM were defined as those with the characteristic rash of DM, including heliotrope rash and Gottron's papule/sign, for at least 6 months (if alive) without clinical weakness attributable to inflammatory myopathy, according to the criteria suggested by Sontheimer ([12]) and Sato and Kuwana ([2]). Baseline demographic data, clinical data, and laboratory data at presentation were taken from hospital records. The presence of ILD was determined by high-resolution computed tomography findings ([13]). RP-ILD was defined as a condition of worsening radiologic interstitial change with progressive dyspnea and hypoxemia within 1 month of the onset of respiratory symptoms ([6]). Routine antinuclear antibody (ANA) test was performed by indirect immunofluorescence on commercially available HEp-2 cell slides as a substrate (Euroimmun). Anti-MDA5 antibody titers were determined by enzyme-linked immunosorbent assay (ELISA) using recombinant MDA5 antigen as described previously ([6]). Anti–aminoacyl–transfer RNA synthetase (anti-aaRS) antibodies were determined by immunoprecipitation (IP) as described previously ([14]). Informed consent was obtained from each subject before blood sampling. The study protocol was approved by the Ethics Reviewing Committee of the Affiliated Drum Tower Hospital.

Retrieval of published studies and data extraction.

A comprehensive PubMed and Institute for Scientific Information Web of Knowledge (ISI WoK) search up to June 2012 was conducted. The following keywords and text words were used: “dermatomyositis” or “polymyositis” combined with “anti-CADM-140” or “anti-MDA5.” We applied no restrictions on language, study design, patient populations (adult or juvenile), or the methodology of anti-MDA5 antibody detection. The reference included had to report original data of anti-MDA5 antibodies, as well as PM/DM and ILD diagnoses. After an initial screening of titles and abstracts, only relevant articles were retained. The qualifying full-text publications and their reference lists were carefully read to determine whether information on the topic of interest was included. For convenience, both RP-ILD and acute/subacute interstitial pneumonia in the literature were referred to as RP-ILD for data summarization.

Statistical analyses.

The chi-square or Fisher's exact test was used for categorical data. Student's t-test or one-way analysis of variance, followed by Bonferroni post hoc (multiple comparisons) tests, were used for continuous data. Differences in survival were determined using the log rank Mantel-Cox test on Kaplan-Meier curves. Multivariate analysis was performed using the Cox proportional hazards model. The following variables were assessed as potential factors associated with poor survival: sex, age, presence of cutaneous manifestations, presence of muscle weakness, presence of anti-MDA5, diagnosis of CADM, and complication of RP-ILD. P values less than 0.05 were considered to indicate statistical significance. The utility of anti-MDA5 antibody measurement for identifying RP-ILD in PM/DM patients was assessed by diagnostic tests. For sensitivity and specificity, 95% confidence intervals (95% CIs) were calculated. Analyses were performed using the SPSS software (version 16.0) and Prism (GraphPad Software).

Meta-analysis.

Sensitivity and specificity, positive and negative likelihood ratios (LRs), and diagnostic odds ratio (DOR) were estimated for individual studies along with their 95% CIs and displayed in forest plots to explore for heterogeneity. A bivariate random-effect model ([15]) was adjusted to obtain a summary receiver operating characteristic (sROC) curve and the corresponding area under the curve (AUC) of anti-MDA5 antibody for detection of RP-ILD. The bivariate model assumes that logit transformations of sensitivity and specificity are negatively correlated and follow a bivariate normal distribution. Summary estimates of sensitivity and specificity with their 95% CIs were obtained from the fitted sROC curve. Publication bias was analyzed using a funnel plot, representing the DOR versus the inverse of the square root of the effective sample size. The degree of asymmetry of the funnel was assessed as recommended by Deeks et al ([16]). For the analyses, we used the Meta-DiSc program (version 1.4) ([17]) to produce the forest plots and the METANDI macro in Stata (version 11) ([18]) to estimate the bivariate model, as well as to obtain the pooled diagnostic accuracy index and the 95% CIs.

RESULTS

Clinical characteristics associated with anti-MDA5 antibody in Chinese PM/DM patients.

Our consecutive cohort of PM/DM patients consisted of 34 with classic DM, 9 with CADM, and 21 with PM. Anti-MDA5 antibody was detected in 26 patients (40.6%). Anti-aaRS antibodies, including anti-OJ, anti-EJ, anti–Jo-1, and anti-KS, were detected in 21 patients (32.8%), including 7 with classic DM and 14 with PM. None of the patients had anti-MDA5 and anti-aaRS antibodies together. Clinical characteristics at presentation were compared between 3 PM/DM patient groups, including anti-MDA5 antibody–positive patients, anti-aaRS antibody–positive patients, and patients negative for these antibodies (Table 1). Anti-MDA5 antibody was detected exclusively in patients with DM, but only 9 patients were diagnosed with CADM, and the remaining 17 patients had clinically apparent skeletal muscle involvement and were classified as having classic DM. Half of the classic DM patients had anti-MDA5 antibody, while all patients with CADM had this antibody. The overall prevalence of ILD in our patients was as high as 75%, and the frequencies in classic DM, CADM, and PM were 79.4%, 100%, and 57.1%, respectively. In particular, all patients with anti-MDA5 antibody had ILD, and this frequency was significantly different between these 3 groups (100%, 85.7%, and 23.5%, respectively; overall P < 0.001).

Table 1. Clinical characteristics at presentation in PM/DM patients positive for anti-MDA5 antibody, positive for anti-aaRS antibody, or negative for these antibodies*
CharacteristicsAnti-MDA5 antibody–positive group (n = 26)Anti-aaRS antibody–positive group (n = 21)Antibody-negative group (n = 17)Overall P
  1. Values are the mean ± SD unless indicated otherwise. PM/DM = polymyositis/dermatomyositis; anti-MDA5 = anti–melanoma differentiation–associated gene 5; anti-aaRS = anti–aminoacyl–transfer RNA synthetase; NS = not significant; CADM = clinically amyopathic DM; ILD = interstitial lung disease; WBC = white blood cell; ALT = alanine aminotransferase; CK = creatine phosphokinase; LDH = lactate dehydrogenase; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; ANA = antinuclear antibody.
  2. aAnti-MDA5 antibody–positive group vs. anti-aaRS antibody–positive group, P < 0.01.
  3. bAnti-MDA5 antibody–positive group vs. antibody-negative group, P < 0.001; anti-aaRS antibody–positive group vs. antibody-negative group, P < 0.05.
  4. cAnti-MDA5 antibody–positive group vs. antibody-negative group, P < 0.05.
  5. dAnti-MDA5 antibody–positive group vs. anti-aaRS antibody–positive group, P < 0.05.
  6. eData were obtained from 24 patients.
  7. fData were obtained from 18 patients.
  8. gData were obtained from 12 patients.
Age at diagnosis, years46.7 ± 13.1151.3 ± 12.143.8 ± 18.8NS
Female sex, no. (%)10 (38.5)15 (71.4)13 (76.5)0.02
Diagnosis, no. (%)   < 0.001
Classic DM17 (65.4)7 (33.3)10 (58.8) 
CADM9 (34.6)00 
PM014 (66.7)7 (41.2) 
ILD, no. (%)26 (100)18 (85.7)4 (23.5)< 0.001
Rapidly progressive ILD, no. (%)10 (38.5)1 (4.8)00.001
Fever, no. (%)9 (34.6)8 (38.1)3 (17.6)NS
Cutaneous manifestations, no. (%)26 (100)6 (28.6)9 (52.9)< 0.001
Cutaneous ulcer, no. (%)3 (11.5)00NS
Muscle weakness, no. (%)16 (61.5)12 (57.1)17 (100)0.003
WBC count, × 103/μl5.3 ± 0.68.8 ± 4.17.1 ± 3.80.009a
Hemoglobin, gm/liter129.0 ± 2.7130.5 ± 4.5125.9 ± 3.7NS
Serum ALT, units/liter58.6 ± 12.669.9 ± 14.6117.4 ± 34.6NS
Serum albumin, gm/liter34.5 ± 1.034.1 ± 1.237.9 ± 1.7NS
CK, units/liter77.4 ± 12.5973 ± 341.33,124 ± 9270.0002b
CK-MB, units/liter15.2 ± 1.463.9 ± 23.2184.2 ± 580.001b
LDH, units/liter556.6 ± 86.3784.7 ± 132.41,712 ± 637.10.04c
ESR, mm/hour38.0 ± 4.227.1 ± 3.224.4 ± 6.1NS
CRP level, mg/liter6.3 ± 2.738.7 ± 13.910.7 ± 7.80.02d
Nuclear staining on ANA test, no. (%)3 (12.5)e3 (16.7)f4 (33.3)gNS

The diagnostic sensitivity and specificity of anti-MDA5 antibody for ILD with DM in our cohort was 72.2% (95% CI 55–86%) and 100% (95% CI 80–100%), respectively. For identifying RP-ILD, the sensitivity and specificity of anti-MDA5 was calculated as 90.9% (95% CI 58.7–99.8%) and 69.8% (95% CI 55.7–81.7%), respectively. PM/DM patients positive for anti-MDA5 antibody showed the highest prevalence of RP-ILD (38.5%, 4.8%, and 0%, respectively; overall P = 0.001) and cutaneous manifestations (100%, 28.6%, and 52.9%, respectively; overall P < 0.001), but the lowest frequency of females (38.5%, 71.4%, and 76.5%, respectively; overall P = 0.02). All of 3 patients with cutaneous ulcers were positive for anti-MDA5 antibody. Significant differences of the frequency of muscle weakness (overall P = 0.003), white blood cell counts (overall P = 0.009), CK (overall P = 0.0002), CK-MB (overall P = 0.001), lactate dehydrogenase (LDH; overall P = 0.04), and C-reactive protein (overall P = 0.02) levels were also observed in these 3 groups. No significant difference of the frequencies for nuclear staining on ANA tests (12.5%, 16.7%, and 33.3%, respectively) was observed in these 3 groups. When we focused on 27 patients with classic DM and ILD, the levels of CK (P = 0.02), CK-MB (P = 0.009), and LDH (P = 0.002) were significantly lower in 17 patients with anti-MDA5 antibody than in 10 patients without anti-MDA5 antibody (see Supplementary Table 1, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21985/abstract).

In total, 11 patients, including 6 with CADM and 5 with classic DM, were complicated by RP-ILD. Of those, 6 patients died of respiratory failure within 9 months after disease onset. RP-ILD occurred more frequently in patients with CADM than in those with classic DM (66.7% versus 14.7%; P = 0.004). The cumulative survival rate of CADM patients at 24 months was significantly lower than that of classic DM patients (P = 0.009) (Figure 1A). We also found a strong association between RP-ILD and anti-MDA5 antibody, and 10 of 11 patients who experienced RP-ILD were positive for anti-MDA5 antibody. Six of 7 (85.7%) patients who died of RP-ILD had anti-MDA5 antibodies at presentation. When cumulative survival rates were compared between anti-MDA5 antibody–positive patients, anti-aaRS antibody–positive patients, and patients negative for anti-MDA5 or anti-aaRS antibody, anti-MDA5 antibody–positive patients showed the worst survival rate, which was lower than the survival rate in anti-aaRS antibody–positive patients or the antibody-negative patients (P = 0.03 and P = 0.04, respectively) (Figure 1B). Interestingly, all deaths in anti-MDA5 antibody–positive patients occurred within 12 months after disease onset, whereas anti-MDA5–negative patients died >18 months after onset because of pulmonary infection, heart failure, or acute exacerbation of chronic ILD. Multivariate analysis revealed that only RP-ILD was independently associated with poor survival (relative hazard 22.9 [95% CI 2.3–229.9]).

Figure 1.

Kaplan-Meier survival curves of clinically amyopathic dermatomyositis (CADM) and classic DM patients (A) and of polymyositis/DM patients positive for anti–melanoma differentiation–associated gene 5 (anti-MDA5) antibody, positive for anti–aminoacyl–transfer RNA synthetase antibody (anti-ARS), or negative for both antibodies (B). † = anti-MDA5 antibody–positive group versus anti-ARS antibody–positive group; †† = anti-MDA5 antibody–positive group versus antibody-negative group.

Comprehensive review of published studies on anti-MDA5 antibody–positive patients.

Our findings on Chinese patients with PM/DM differ somewhat from published reports in terms of the low prevalence of CADM in patients with anti-MDA5 antibody. To evaluate potential influences of ethnicity, primary specialty of the investigators (rheumatology, dermatology, pediatrics, and pulmonology), and anti-MDA5 antibody detection methods on clinical associations of anti-MDA5 antibodies, we conducted a comprehensive literature review. A total of 69 references were retrieved from PubMed and ISI WoK databases, but 50 references were excluded after screening titles and abstracts (see Supplementary Figure 1, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21985/abstract). The remaining 19 studies (15 from Japan, 1 from the US, 1 from Korea, and 2 from China), in addition to the present study, were initially selected ([5-9, 14, 19-31]). Table 2 lists characteristics of the studies reviewed. Three studies were excluded due to potential overlap of cases according to the private communications with the corresponding authors ([20, 24, 27]). One study was excluded because of incomplete clinical data ([26]). As a result, 16 studies eligible for the analysis enrolled 233 patients, including 7 juvenile patients. The distribution of classic DM and CADM in patients with anti-MDA5 antibodies in individual studies is shown in Table 2. Variability in disease distribution among the studies was apparent. These differences could not be explained by specialty or method of antibody detection, but the prevalence of CADM was significantly higher in Japanese patients than in non-Japanese patients (74.7% versus 39.2%; P = 1.2 × 10−7). In addition, in patients with anti-MDA5 antibody, frequency of females was significantly higher in Japanese than in non-Japanese patients (75.4% versus 53.2%; P = 0.0003) (Table 2). Seven juvenile patients were included, all from Japanese studies. Interestingly, all the juvenile patients had classic DM, and this frequency was significantly higher than the frequency in adult patients (100% versus 35.4%; P = 0.0009), indicating that anti-MDA5 antibody is associated with juvenile classic DM with skeletal muscle involvement, even in a Japanese population.

Table 2. Study characteristics and distribution of classic DM and CADM in patients with anti-MDA5 antibodies*
Author, year (ref.)No. of antibody-positive patients (male/female)Total no. patientsSpecialty of investigatorsMethod of antibody determinationCountryAdditional informationClassic DMCADM
  1. DM = dermatomyositis; CADM = clinically amyopathic DM; anti-MDA5 = anti–melanoma differentiation–associated gene 5; IP = immunoprecipitation; ELISA = enzyme-linked immunosorbent assay.
Sato et al, 2005 ([5])8 (2/6)42RheumatologyIP using radiolabeled K562 cell extractsJapan8 adults
Fujikawa et al, 2009 ([8])8 (1/7)30Rheumatology/ dermatologyIP using radiolabeled K562 cell extractsJapan2 adults6 adults
Sato et al, 2009 ([6])23 (NA)67RheumatologyIP using radiolabeled K562 cell extracts and ELISA using recombinant MDA5Japan1 adult22 adults
Nakashima et al, 2010 ([7])13 (4/9)37RheumatologyIP using radiolabeled K562 cell extractsJapan2 adults11 adults
Gono et al, 2010 ([9])14 (3/11)24RheumatologyIP using radiolabeled K562 cell extracts and ELISA using recombinant MDA5Japan6 adults8 adults
Hoshino et al, 2010 ([19])21 (4/17)82Rheumatology/dermatologyIP using radiolabeled HeLa cell extractsJapanIncludes 1 juvenile patient, excluded from meta-analysis due to lack of control data1 juvenile20 adults
Gono et al, 2011 ([20])2 (0/2)2RheumatologyELISA using recombinant MDA5JapanExcluded due to potential overlap of cases
Tanizawa et al, 2011 ([21])12 (4/8)25Pulmonology/rheumatologyIP using radiolabeled HeLa cell extractsJapan6 adults6 adults
Hamaguchi et al, 2011 ([22])43 (9/34)410DermatologyIP using radiolabeled HeLa cell extractsJapanExcluded from meta-analysis due to lack of control data10 adults33 adults
Kobayashi et al, 2011 ([23])5 (3/2)13PediatricsELISA using recombinant MDA5JapanIncludes 5 juvenile patients5 juveniles
Sato et al, 2011 ([24])1 (0/1)1RheumatologyELISA using recombinant MDA5JapanExcluded due to potential overlap of cases
Sakurai et al, 2011 ([25])1 (1/0)1PediatricsELISA using recombinant MDA5JapanIncludes 1 juvenile patient, excluded from meta-analysis due to lack of control data1 juvenile
Muro et al, 2011 ([26])26 (5/21)95DermatologyIP using biotinylated recombinant MDA5JapanExcluded due to lack of full clinical data
Ikeda et al, 2011 ([14])6 (3/3)55DermatologyIP using radiolabeled K562 cell extractsJapan5 adults1 adult
Gono et al, 2012 ([27])27 (7/20)27RheumatologyELISA using recombinant MDA5JapanExcluded due to potential overlap of cases
Fiorentino et al, 2011 ([28])10 (3/7)77DermatologyIP using radiolabeled recombinant MDA5US5 adults5 adults
Kang et al, 2010 ([29])9 (4/5)49RheumatologyIP using radiolabeled HeLa cell extractsKorea9 adults
Chen et al, 2011 ([30])19 (8/11)84RheumatologyELISA using recombinant MDA5China14 adults5 adults
Cao et al, 2012 ([31])15 (6/9)64DermatologyELISA using recombinant MDA5China3 adults12 adults
Current study26 (16/10)64Rheumatology/pulmonologyELISA using recombinant MDA5China17 adults9 adults

Meta-analysis of utility of anti-MDA5 antibody measurement for identifying RP-ILD in patients with PM/DM.

We further examined the potential utility of anti-MDA5 antibody measurement for identifying RP-ILD in patients with PM/DM using a meta-analysis. Three studies were further excluded due to lack of control patients with PM/DM but without anti-MDA5 antibody ([18, 21, 24]) (see Supplementary Figure 1, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21985/abstract). In total, the meta-analysis included 631 patients with PM/DM, consisting of 168 with anti-MDA5 antibody and 463 without it. As shown in Figure 2A, sensitivity values for the individual studies ranged from 40% to 100%, and the pooled estimated sensitivity was 77% (95% CI 64–87%), with some heterogeneity between the studies. Specificity values for the individual studies ranged from 63% to 100%, and the pooled estimated specificity was excellent (86%, 95% CI 79–90%) (Figure 2B). The LRs for a positive result of anti-MDA5 antibody ranged from 1.37 to 32.45 across the studies, with a pooled estimate of 5.37 (95% CI 3.51–8.21), while the LRs for a negative result were between 0.11 and 0.84, with a pooled estimate of 0.26 (95% CI 0.16–0.44), although there was considerable heterogeneity between the studies. These estimates are not affected by the presence of publication bias as demonstrated by the symmetry of the funnel plot (see Supplementary Figure 2, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21985/abstract). Finally, the DORs for the individual studies had the widest range, from 1.63 to 139.4, with an overall value of 20.41 (95% CI 9.02–46.20) and substantial heterogeneity (Figure 2C).

Figure 2.

Forest plots of estimated sensitivity (A) and specificity (B) of anti–melanoma differentiation–associated gene 5 antibody testing for identifying rapidly progressing interstitial lung disease with polymyositis/dermatomyositis, as well as diagnostic odds ratios (ORs) (C). Circles represent the point estimates from each study, and horizontal lines show the 95% confidence intervals (95% CIs). Black circles and red circles represent the data from Japanese studies and from non-Japanese studies, respectively.

The value of anti-MDA5 antibody measurement for identifying RP-ILD was confirmed by ROC analysis involving all studies included in meta-analysis (AUC 0.89, 95% CI 0.63–0.98) (Figure 3). The utility of anti-MDA5 antibody for identifying RP-ILD in patients with PM/DM was observed consistently in both Japanese and non-Japanese patients.

Figure 3.

Summary receiver operating characteristics curve (HSROC) for all studies included in the meta-analysis of anti–melanoma differentiation–associated gene 5 antibody utility for identifying rapidly progressing interstitial lung disease with polymyositis/dermatomyositis. Open circles and red circles represent the data from Japanese studies and from non-Japanese studies, respectively. AUC = area under the curve; 95% CI = 95% confidence interval. Color figure can be viewed in the online issue which is available at http://onlinelibrary.wiley.com/doi/10.1002/acr.21985/abstract.

DISCUSSION

Our investigation of Chinese patients with PM/DM from a single medical institute revealed that anti-MDA5 antibody was exclusively detected in DM patients with ILD, including both classic DM and CADM. Anti-MDA5 antibody was strongly associated with RP-ILD–related early death and poor prognosis. From a literature review, we found a significant difference in disease distribution in anti-MDA5–positive patients between Japanese and non-Japanese populations. However, our meta-analysis revealed that anti-MDA5 antibodies can be considered a valuable tool for identifying high risk of developing RP-ILD in PM/DM patients with high sensitivity (77%) and high specificity (86%), regardless of ethnic origin. PM/DM patients with anti-MDA5 antibody had 20-fold higher odds of having RP-ILD than patients negative for anti-MDA5 antibody.

RP-ILD is an intractable and life-threatening complication of PM/DM, even if treated with aggressive immunosuppressive therapy ([31, 32]). In 2 RP-ILD with CADM patients' serial data, the 6-month survival rate was estimated as 40.8–45% ([3, 32]). Thus, early diagnosis and treatment will be of critical importance to improving this prognosis. Recently, Horai et al reported that in 2 cases of RP-ILD with CADM, anti-MDA5 antibody measurements resulted in early diagnosis and successful treatment with early immunosuppressive therapy ([33]). Anti-MDA5 antibody titer decreased with effective treatment ([31, 34, 35]), suggesting that the antibody titer may be a marker for the evaluation of disease severity and treatment effectiveness. In this regard, as a quantitative and convenient method for anti-MDA5 antibody level measurement, ELISA will be more practical than an IP assay in clinical practice.

To date, anti-MDA5 antibodies have primarily been found in Japanese RP-ILD–complicated CADM patients ([2]). From a review of the literature, the overall CADM/classic DM ratio in Japanese patients was as high as 2.95, while this ratio in Chinese patients was 0.53. This observation in Chinese patients was compatible with that in a Korean study ([29]), indicating significant DM subclassification difference in anti-MDA5–positive patients between Japanese and other eastern Asian populations. In addition, female predominance seen in Japanese patients with anti-MDA5 antibodies was not detected in non-Japanese patients. These differences may be due to both environmental and genetic factors between these populations, although a population-based study is necessary to confirm these ethnic differences. Recently, Muro et al reported a detailed epidemiologic analysis of anti-MDA5 antibody–positive patients in central Japan ([26]). The increasing prevalence of anti-MDA5 antibodies in small rural towns and the geographical clustering of such patients in 2 areas along the Kiso River indicated a role of environmental factors. It is noteworthy that all Chinese patients in the present study were from the Nanjing area of Jiangsu province, through which the Yangtze River, the biggest river in China, flows. A detailed epidemiologic study should be conducted in the future to identify the environmental difference(s) between central Japan and the Jiangsu area of China. Further, considering the widely accepted association between myositis-specific autoantibodies and genetic backgrounds, these differences may also be due to genetic differences among these populations, especially in the HLA region ([36, 37]). In this regard, production of anti-MDA5 antibodies is reported to be associated with HLA–DRB1*0405 and DRB1*0101 in the Japanese population ([38]).

This study has several limitations. First, case selection bias may exist because all data for review and the meta-analysis were collected from retrospective studies. Given that anti-MDA5 antibody may appear before onset of ILD ([23]), the association between anti-MDA5 antibody and a DM subset with RP-ILD needs to be further explored in a prospective study. Second, several different methods were used for the detection of anti-MDA5 antibodies in previous studies. Even in the ELISA system used in this study, sensitivity and specificity in comparison with the gold standard IP were shown to be 85% and 100%, respectively ([6]). A standardized method for detection of anti-MDA5 antibodies is necessary for future clinical studies, as well as clinical practice. Finally, current literature reports are primarily clustered on eastern Asian countries, including Japan and China. It would be interesting to evaluate the prevalence and clinical associations of anti-MDA5 antibodies in patients with PM/DM from other ethnicities.

In conclusion, anti-MDA5 antibody measurements are valuable for identifying DM patients with a high risk for developing RP-ILD. Making the test for anti-MDA5 antibodies routinely available is an urgent matter because early identification of patients with the high RP-ILD risk enables early aggressive therapy that may potentially improve the prognosis of patients with this devastating complication.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Drs. Kuwana and Sun had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Chen, Cao, Kuwana, Sun.

Acquisition of data. Chen, Cao, Plana, Liang, Cai, Kuwana.

Analysis and interpretation of data. Chen, Cao, Plana, Kuwana, Sun.

ACKNOWLEDGMENTS

We thank all the subjects included in this study and their families for sample collection. We thank Yuka Okazaki and Shigeaki Suzuki for excellent technical assistance. We would also like to acknowledge the advice on statistics by Professor Javier Zamora.

Ancillary