Baseline characteristics of new-onset DM patients
Data from 142 new-onset DM patients with multiple variables was described. 66.2% were female, and the median age was 50 years old. For most patients, the initial course of the disease is about three months, and follow-up can be up to two years, with the longest follow-up being five years. 66.2% of the patients had DM-specific skin involvement (either heliotrope, Gottron’s sign, or papules), 75.4% had muscle weakness, 35.2% had elevated CK levels and 63.4% had ILD, besides, 62% of them were confirmed by muscle or skin biopsy. As for myositis auto-antibody profile, 80.3% of the patients were MSAs positive, and the proportion of anti-MDA5 positive patients was the highest (47.2%). Only one patient had no antibody information recorded, and this patient died of RP-ILD during hospitalization. We also focused on comorbidities in DM, with five cases of tuberculosis, two cases of pulmonary arterial hypertension, one case of subepidermal calcification. Of the nine newly diagnosed cases of tumor-associated dermatomyositis, three were three breast cancer, two were non-Hodgkin's lymphoma, two were lung cancer, one was nasopharyngeal cancer, and one was tonsil cancer. Of these, five patients were positive for anti-TIF1γ, three were positive for anti-MDA5, and one was positive for anti-NXP2. Notably, tumors were diagnosed within 6 months in all TIF1γ+ patients, while the time to develop tumors in MDA5+ and NXP2+ patients ranged from 1 to 3 years.
Baseline characteristics were compared among three subgroups, “Mild”, “Aggravation” and “Death”. Clearly, the overall disease course was shortest in the death group. The proportion of patients with fever, ILD/RP-ILD or positive auto-antibodies was higher in this group, and patients had lower lymphocytes levels or higher CRP levels. Although there was no significant difference in ferritin level between the three subgroups (probably due to the large difference in values), ferritin level was significantly elevated in MDA5+ patients (median 924 ng/mL in MDA5+ group vs. 462.1 ng/mL in MDA5- group, p<0.0001). In the aggravation group, muscle involvement was relatively more severe than in the other two subgroups, although not statistically significant. In the mild group, the patients were the youngest, had the highest proportion of females and were less severe in clinical presentation. The characteristics of these different subgroups were shown in Table 2.
We calculated possible exposures in the three subgroups, and we found that, in the aggravation or death group, a higher proportion of patients had a history of catching cold (10.7% or 21.4% vs. 5.6%) or tobacco exposure (17.9% or 21.4% vs. 16.7%) before onset. See Table 3.
Aggravated features in new-onset DM patients
In the aggravation group, the frequency and symptoms of aggravation were recorded. The median time point for new-onset DM patients to experience their first aggravation is 10 months (IQR, 2 ~ 18 months) after initial treatment. We also recorded the time point when ILD progressed, which was similar to the former (Figure 2A and C). 70% of the DM patients experienced one time of aggravation during follow-up period, rest of them experienced 2 ~ 4 aggravations (Figure 2D). Only one positive anti-PL-7 patient experienced five aggravations. This patient was followed for 39 months and experienced recurrent aggravations due to uncontrolled ILD. We calculated the interval period between each aggravation. The median interval period was 8 months, IQR 4 ~16 months (Figure 2B). As a result, in the aggravation group, most patients were able to achieve remission after medication adjustment, with the first aggravation occurring at 10 months after initial treatment.
We concluded 7 classic manifestations in the aggravation group (Figure 3A and B). For those who had only one aggravation during follow-up period (n = 39), 30.8% of them experienced a range of aggravating symptoms (developed skin rash, dyspnea and muscle weakness), 30.8% had developed skin rash, 26% had recurred muscle weakness or developed dyspnea, 7.7% combined with thrombocytopenia or dermatomyositis associated myocarditis, only 6% had developed arthritis / arthralgia or dysphagia. For those with recurrent disease activity for more than one time (n = 17), 47.1% of them experienced a range of aggravating symptoms, 47% had enlarged rash or developed dyspnea, 5.9% had recurred muscle weakness.
We further analyzed aggravation features in DM patients with different MSA (anti-MDA5 n = 27, anti-TIF1γ n = 5, anti-NXP2 n = 1, anti-SAE n = 3 and anti-Mi2 n = 5), as shown in Figure 4. Patients in the anti-MDA5 group had the earliest onset of aggravation, with median aggravation time in the fourth month. Symptoms of aggravation varied among patients with different MSA. In the anti-MDA5 group, aggravation was characterized by the development of a skin rash or dyspnea, either in combination or alone. In the anti-TIF1γ group, developed rash was the most common symptoms. More patients in the Mi2 group had recurrent aggravations of muscle weakness. In the anti-SAE and anti-NXP2 groups, the number of patients was too small to be representative of the overall population. In summary, an enlarged rash or progressive dyspnea remains the primary sign of aggravation in DM patients.
There were some differences in baseline clinical data between new-onset DM patients with single aggravation and those with multiple aggravations after regular treatment (Table 4). Patients with recurrent aggravations had significant higher CRP (median 8.3 mg/L vs. 4.8 mg/L, p<0.001) and CK levels (median 521 U/L vs. 210 U/L, p<0.05) compared with those with single aggravation. Intriguingly, the proportion of positive anti-MDA5 in the multiple aggravations group was actually lower than that in the single aggravation group, a departure from our previous understanding that anti-MDA5 was generally associated with a worse prognosis. We speculated that this may be related to the promotion of new therapies and the aggravation characteristics of MDA5+ DM themselves, which will be elaborated in the discussion section.
Treatment regimens for different clinical outcome groups
The detailed treatment regimens for this cohort were summarized in Table 5. Regimens included the use of glucocorticoid (GC), immunosuppressants (IS), plasma exchange (PE), and janus kinase inhibitors (JAKinibs). The combined use of GC, IS and PE or JAKinibs were more prevalent in the aggravation group than in the mild group, this may be because patients in the aggravation group had heavier baseline conditions. We further classified GC + IS into three cases: GC combined with a single, double or triple IS. We found the highest fraction of GC + single IS and the lowest fraction of GC + double IS in the multiple aggravations group. In addition, the proportion of GC+ triple IS was highest in the mild group. Specific IS profiles were also recorded, including hydroxychloroquine (HCQ), methotrexate (MTX), mycophenolate mofetil (MMF), cyclophosphamide (CYC), tacrolimus (TAC), ciclosporin A (CsA), and thalidomide (Thd). In addition to HCQ, the most commonly used IS in the 4 groups were TAC and Thd in mild and single aggravation group, CYC in multiple aggravation group, TAC and CYC in death group, respectively. The above results suggested different regimens in DM patients with various clinical outcomes. Besides, GC in combination with more than two IS was more likely to improve clinical outcomes in DM patients.
Pathological and imaging characteristics in DM patients with different clinical outcomes
As mentioned above, more than 2/3 of the patients had dermatomyositis-specific rash, ILD, and biopsy-confirmed muscle lesions, therefore, we further recorded the imaging features of pulmonary HRCT and the pathological features of muscle biopsy in DM patients, and compared among subgroups to further clarify the differences in patients with different clinical outcomes.
In this cohort, HRCT imaging features include patch shadow, linear opacities, ground-glass opacities, thread net and honeycomb sign. In the mild group, ground-glass opacities, thread net and honeycomb sign were rarely seen, especially honeycomb sign, which appeared only in the aggravation or death group. Linear opacities were more common in the aggravation group while patch shadows were more common in the death group, as shown in Figure 5C.
Muscular pathological changes in DM include connective tissue hyperplasia, inflammatory cell infiltration, muscle fiber necrosis, perifascicular atrophy and peripheral vasculitis, as shown in Figure 5A. In the mild group, 63.6% had at most two types of myopathy, which occurred primarily as connective tissue hyperplasia and inflammatory cell infiltration. In contrast, 81.3% of the aggravation or death group had 2 to 5 types of myopathic features with predominantly inflammatory cell infiltration, thus, we further analyzed the types of infiltrating inflammatory cells, as shown in Figure 5B. Relatively speaking, the rate of CD4 was higher in the aggravation group, while CD68 was higher in the mild group. In summary, pulmonary HRCT findings and muscle pathology features differ in DM with different clinical outcomes.
Prediction of clinical outcomes with decision tree algorithm
We constructed a decision tree model to predict the clinical outcomes for patients. To identify the variables that best classified patients into subgroups, the following defining variables from Table 2 were selected as candidates: disease course, fever, lymphocytes, CRP levels, and ILD. We obtained a pruned decision tree from cost-complexity analysis, with a minimum cross-validation error (n = 142, x-error = 0.9714286 for 5 splits and complexity parameter = 0.02). Finally, the following variables were included in the decision tree: disease course, lymphocytes, CRP levels. We set 70% of the DM cohort as the training set and 30% as the testing set. Decision tree accuracy was 79.1% (Figure 6).