Background
Polymyositis (PM) and dermatomyositis (DM) are idiopathic inflammatory myopathies (IIM) of unknown causes, which are often associated with extramuscular manifestations such as interstitial lung disease (ILD), arthropathy, cardiomyopathy, and malignancies. PM/DM-associated ILD (PM/DM-ILD) is a major cause of death [
1,
2], with an estimated excess mortality rate of around 40%, and has a widely varied clinical course.
ILD is categorized according to histopathological findings into usual interstitial pneumonia, non-specific interstitial pneumonia (NSIP), organizing pneumonia, diffuse alveolar damage, respiratory bronchiolitis, desquamative interstitial pneumonia, and lymphoid interstitial pneumonia. In clinical practice, a multidisciplinary approach based on pathology findings, clinical symptoms, laboratory tests, imaging examinations, and respiratory function tests, is the gold standard for clinical diagnosis and classification of ILD [
3]. High-resolution computed tomography (HRCT) scan is one of the most informative examinations for ILD. However, in comparison with idiopathic ILD, extrapulmonary findings including serological data are particularly important for making the diagnosis and determining the prognosis of connective tissue disease (CTD)-ILD including PM/DM-ILD. Indeed, previous studies have identified extrapulmonary findings as prognostic factors for PM/DM-ILD, such as old age [
4], skin ulcer, ILD with low serum creatine kinase (CK) [
5], non-Caucasian race, male sex [
6], and seropositivity for anti-melanoma differentiation-associated protein 5 (MDA5) antibody (Ab) [
7,
8]. It is important to optimize disease management based on the prognostic factors to improve clinical outcome.
In this study, we retrospectively assessed clinical data, treatments, and clinical outcomes in patients with PM/DM using the multicenter database to identify predictive prognostic factors for PM/DM-ILD.
Methods
Patient selection
One hundred sixteen patients with PM/DM-ILD, who had received initial treatment in six hospitals including Yokohama City University and the affiliated hospitals, from 2003 to 2016, were enrolled. The diagnosis of PM/DM was based on the Bohan and Peter criteria [
9]. The modified Sontheimer criteria were adopted for the classification of clinically amyopathic dermatomyositis (CADM) [
10]. Briefly, patients was categorized as having CADM if they presented with DM-specific skin disease but subclinical or no clinical evidence of proximal muscle weakness and myositis on laboratory, electrophysiologic, and/or radiologic evaluation more than 6 months after rash onset. This study was conducted in accordance with the Declaration of Helsinki, and informed consent was obtained from the patients and healthy controls. The study design was approved by the ethics committee of Yokohama City University.
Collection of clinical data
We retrospectively obtained the data from the clinical charts of the individual patients. According to PM/DM-related manifestations such as presence of muscle weakness, typical dermatologic manifestations such as Gottron’s signs, Gottron’s papules, heliotrope rashes, and mechanic hands, fever, dysphasia, ILD, and malignancies, the patients were categorized into three groups according to the Bohan and Peter criteria and the modified Sontheimer criteria [
9,
10]: PM; clinically amyopathic dermatomyositis (CADM); and classical dermatomyositis (DM). In addition, the diagnosis of cancer-associated myositis (CAM) was made in patients who had the complication of malignancy within 3 years before or after the onset of PM/DM.
We reviewed the disease onset, besides demographic findings including age, sex, habitual history such as smoking, and clinical findings such as laboratory data and imaging findings. The laboratory data included the routine biochemical and haematological data, KL-6, ferritin, those on blood gas analysis, and myositis-associated autoantibodies including anti-nuclear antibodies (ANA), anti-histidyl-tRNA synthetase (Jo-1), anti-aminoacyl transfer RNA synthetase (ARS), anti-MDA5, anti-transcriptional intermediary factor 1-γ (TIF1-γ), and anti- Sjögren’s syndrome A (SS-A) antibodies (Abs).
We also analyzed therapy and major clinical outcomes including survival, comorbidities, and complications. Common remission induction therapies for PM/DM-ILD were as follows; oral prednisolone (PSL) at a dose of 0.2–1.5 mg/kg of body weight per day, calcineurin inhibitors including tacrolimus or cyclosporine A, and administration of intravenous cyclophosphamide (IVCY), intravenous methyl-prednisolone (mPSL) pulse, and intravenous immunoglobulin (IVIg). All of the therapeutic strategies were determined by the attending physicians on the basis of comprehensive assessment of clinical manifestations and laboratory data. Serious infections were defined as conditions requiring hospitalization or extension of hospital stay for additional therapy with antimicrobial agents except for prophylactic purposes.
HRCT
Presence of PM/DM-ILD was determined by chest HRCT findings such as ground glass opacity, reticular opacity, areas of consolidation, honeycombing, traction bronchiectasia, and linear opacity [
11,
12]. Patients who needed initiation or intensification of immunosuppressive therapy for the lung lesions were included in the study. Distribution of ILD lesions was assessed in four lung zones, zone A (above the aortic arch), zone B (between the aortic arch and the level of the carina), zone C (between the level of the carina and the level of inferior pulmonary veins) and zone D (below the inferior pulmonary veins), according to a previous report [
13]. In each zone, the extent of ILD lesions was semiquantitatively scored based on the percentage of the lung parenchyma involved: no involvement (0 points), 1–4% (1 point), 5–14% (2 points), 15–29% (3 points), 30–49% (4 points), and ≥ 50% involved (5 points). A total score was calculated by summing the individual area scores in each zone.
The HRCT images were independently reviewed by two rheumatologists well-trained in interpretation of chest CT images and blinded to the patients’ baseline data and treatment regimen. In the reliability assessment, the total score of the inter-evaluator and intra-evaluator coefficients of variation were good (κ = 0.71) and excellent (κ = 0.83), respectively. The inter-evaluator coefficient of variation in zone A to zone C was excellent (κ = 0.85, 0.84 and 0.92, respectively) and was good in zone D (κ = 0.72). The intra-evaluator coefficients of variation was excellent in every zone (κ = 0.84, 0.93, 0.93, and 0.80, respectively).
Statistical analysis
To identify risk factors for death or complications from serious infections, statistical analysis was performed using SPSS software (IBM) and GraphPad Prism (GraphPad Software). We used the chi-square test or Fisher’s exact test for categorical variables, Student’s t test or analysis of variance for parametric analysis of continuous variables, or the two-tailed Mann-Whitney U test, Wilcoxon rank sum test, or Kruskal-Wallis test for non-parametric analysis. Receiver operating characteristic (ROC) analysis was performed to identify the appropriate cutoff values for variables that were significantly different on univariate analysis. We divided all the patients into two groups by the cutoff values and evaluated the survival curves statistically using the log-rank test. Cox proportional hazards regression and binomial logistic regression were performed as multivariate analysis to identify independent risk factors for death and serious infections. A p value < 0.05 was considered significant in all analyses.
Discussion
Consistent with some previous reports [
1,
2], we had identified the existence of ILD at baseline and the complication of serious infection as risk factors for death (OR 6.10,
p = 0.010, and OR 2.46,
p = 0.038, respectively) in patients with PM/DM (n = 188) in our multicenter database. In this study, we focused on the patients with PM/DM who had ILD and investigated their independent prognostic factors.
This study showed that presence of the ILD lesion in the upper lung fields and hypocapnia were independently associated with unfavorable clinical outcomes, especially early death in patients with PM/DM-ILD. In general, mortality in PM/DM-ILD has two waves [
14‐
16]. Exacerbation of ILD and/or complication of serious infections are mainly responsible for the first peak [
14,
17]. The present study revealed that hypocapnia and extent of ILD lesions up to the upper lung fields are the independent prognostic factors for PM/DM-ILD, indicating that progression and ILD severity was the most critical factor in the early phase. In concordance with the present study, we have previously shown that positron emission tomography (PET)-CT scan visualizes active ILD lesions and the extent is closely associated with progression of ILD in connective tissue diseases including PM/DM [
18]. The second wave of mortality is flatter than the first one and is mainly due to malignancy in patients with CAM [
14,
19].
There is accumulating evidence that patients with CADM and anti-MDA5 Ab frequently have acute progressive ILD requiring potent immunosuppressive therapy [
20,
21]. However, neither CADM nor anti-MDA5 Ab was identified as prognostic factors in this study. The discrepancy from the previous studies was caused by the incomplete study of anti-MDA5 Ab, which was examined in some of the patients, mainly in those diagnosed more recently. Our study did not show any difference in survival prognosis between CADM and classical DM. The data suggest that serological anti-MDA5 Ab is more closely associated with acute progressive ILD than CADM as a clinical phenotype, as previous studies have shown that patients with classical DM anti-MDA5 Ab present with rapidly progressive ILD as do patients with seropositive CADM [
22‐
24]. Serum ferritin level has been shown to be a biomarker for the activity of ILD in patients with PM/DM [
25]. Unlike the previous studies, however, we did not identify a significant association between serum ferritin and ILD-related death. In this study, we only analyzed the baseline data but not those in those with the most active disease, presumably leading to the discrepancy.
There is no evidence-based treatment for PM/DM-ILD, because of lack of prospective randomized controlled trials. In the real world, the first-line therapy for PM/DM-ILD is high-dose glucocorticoid therapy with or without immunosuppressants [
26‐
29]. As several lines of study suggest, potent immunosuppressive combination therapy using high-dose glucocorticoid, calcineurin inhibitors, and cyclophosphamide should be considered as induction therapy for acute progressive ILD, particularly in patients with CADM who are positive for anti-MDA5 Ab, because of unfavorable prognosis, as mentioned above [
30].
On the other hand, it has been shown that infection is one of the major complications during treatment for PM/DM. Some recent studies showed that the serious infection occurred in 26–37% of patients with PM/DM during treatment [
17,
31‐
33]. The incidence rate of serious infection in our database cohort was quite similar to those in these studies. The rate of serious infection as a cause of death was also similar to those of the previous studies (2–22%) [
5,
17,
31,
33,
34].
Intensive immunosuppressive therapies are associated with high risk of infection [
17,
31,
35]. Nzeusseu et al. also demonstrated that high dose of initial corticosteroids is associated with increased adverse events including infection but not with improving clinical outcomes of myopathies [
36]. Another group identified different sets of risk factors including esophageal involvement, ventilator insufficiency, malignancy, myalgia, and use of methotrexate, but not a cumulative dose of glucocorticoids or use of immunosuppressants in patients with PM/DM [
32]. However, these reports might be too old to assess recent styles of combination immunosuppressive therapies including calcineurin inhibitors. Unlike previous studies, besides therapeutic factors including the initial PSL dose, and combination immunosuppressive therapy, the present study showed high serum KL-6 to be a risk factor for infection, suggesting that progressive ILD predisposes to the local complication of infection.
We would like to emphasize that our study identified serious infections, especially with respiratory infection, as another prognostic factor for PM/DM-ILD patients. It is likely that complication with respiratory infection further accelerates the progression of respiratory insufficiency in a patient with PM/DM-ILD requiring intensive immunosuppressive therapies, resulting in a fatal event. Indeed, such clinical courses were observed in 10 of 14 patients with PM/DM-ILD who died in this study.
Our data suggest the reconsideration of the appropriate therapeutic regimen for PM/DM-ILD. The present retrospective observational analyses suggests that the clinical outcome is favorable in the regimen using 0.5 mg/kg body weight per day of PSL with a calcineurin inhibitor compared with that using 1.0 mg/kg/day of PSL or IVCY, although the choice of either regimen has not been controlled. This issue should be analyzed in a prospective study in the future. At this moment, we would like to emphasize that infection should be controlled by prophylaxis based on screening, and by early detection using monitoring and subsequent treatment, in the management of patients with PM/DM-ILD receiving intensive immunosuppressive therapies.
There are several limitations in this study. First, the previously reported prognostic factors for PM/DM-ILD, such as myositis-specific autoantibodies (MSA) and ferritin, were not collected comprehensively because of the retrospective study design and unavailability of stocked serum as discussed above. It can be attributed to the fact that measurement of MSA, except for anti-Jo-1 Ab, had not been commercially available and the relationship between serum ferritin level and mortality in PM/DM-ILD had not yet been identified in the significant part of the study period. Second, therapeutic decision-making relied on the individual attending physicians without use of consistent protocols.