Abstract
Backgrounds
Anti-MDA5 antibody-positive dermatomyositis (DM) is characterized by the presence of anti-MDA5 antibodies. We aimed to assess longitudinal changes of anti-MDA5 antibody levels and long-term outcomes in 6-month survivors of anti-MDA5 + DM with interstitial lung disease (ILD).
Methods
Anti-MDA5 antibody titers were measured using enzyme-linked immunosorbent assay upon newly diagnosed anti-MDA5 + DM-ILD patients. Patients with follow-up beyond 6 months and at least two anti-MDA5 antibody measurements were subsequently subjected to group-based trajectory analysis by utilizing latent class mixture model. Seroconversion was defined as a change from positive to negative for anti-MDA5 antibodies. Outcome measurements include low disease activity (LDA) status and remission.
Findings
In 6-month survivors, two anti-MDA5 antibody trajectories were identified: “fast seroconvertors” and “slow seroconvertors”. The median time to anti-MDA5 seroconversion for fast seroconvertors was 9 (IQR: 5–15) months, whereas slow seroconvertors did not achieve 50% anti-MDA5 seroconversion during the follow-up period. Although with a lower baseline PaO2/FiO2 (P = 0.01), along with a lower baseline anti-MDA5 antibody level (P < 0.001), these anti-MDA5 fast seroconvertors had a much higher rate of LDA attainment compared to slow seroconvertors at 12 months (66.0% vs. 37.2%, P < 0.001); as well as a greater proportion of prednisone-free remission by 24 months (16.7% vs. 8.8%, P = 0.02), and drug-free remission by 36 months (10.5% vs. 2.7%, P = 0.006). The relative changes of anti-MDA5 antibody levels at the first three months (ΔMDA5@3m), with a cut-off point of 33.0% reduction, yielded a good predictive value for different seroconvertors with an area under curve of 0.86. Multivariable Cox regression model identified a reduction in ΔMDA5@3m of ≥ 33.0% as an independent favorable factor for achieving LDA (HR 1.96, 95% CI: 1.33–2.90, P < 0.001).
Conclusions
Two distinct trajectories of anti-MDA5 antibody levels among 6-month survivors with anti-MDA5 + DM-ILD have been identified. The long-term outcomes were favorable for those with fast seroconvertors.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12931-026-03572-y.
Keywords: Dermatomyositis, anti-MDA5 antibody, Interstitial lung disease, Trajectories, Outcomes.
Introduction
Anti-melanoma differentiation-associated protein 5 antibody-positive dermatomyositis (anti-MDA5 + DM) is a distinct subtype of DM characterized by a tendency to complicate rapidly progressive interstitial lung disease (RP-ILD) [1, 2]. RP-ILD is a critical factor contributing to the high mortality of this disease, with studies indicating that 33% to 75% of patients succumb within six months of diagnosis [3–6]. On the other hand, for patients who survive the first six month, the long-term outcome appears favorable [6]. However, in long-term survivors with anti-MDA5 + DM, failure to achieve remission and disease recurrence constitute a substantial clinical burden, particularly in the subset with persistently positive antibodies [7, 8].
As the signature serological biomarker, high anti-MDA5 antibody levels have been linked to increased mortality from RP-ILD [7, 9, 10]. Conversely, low titers of anti-MDA5 antibodies correlate with better outcomes [11]. Additionally, longitudinal data from small cohorts suggest that anti-MDA5 antibody titers are in parallel with disease activity status: remission often corresponds to a reduction or seroconversion of the anti-MDA5 antibody, while an increase or resurgence of the antibody might indicate a disease flare [8, 12]. Overall, anti-MDA5 antibodies should be considered key serological biomarkers, not only for establishing a diagnosis but also for monitoring disease status over time. However, a significant knowledge gap persists in understanding the longitudinal dynamics of anti-MDA5 antibody levels and the clinical significance of different trajectories.
Therefore, we conducted a real-world study using a large inception cohort of anti-MDA5 + DM-ILD patients, focusing on those who survived for more than six months post-diagnosis. We then employed trajectory analysis to identify distinct patterns of anti-MDA5 antibody changes over time. This approach aims to enhance our comprehensive understanding of the disease, ultimately informing more personalized and effective management strategies for this challenging condition.
Patients and methods
Study population
Between July 2017 and June 2024, adult patients aged 18 years or older admitted to the department of Rheumatology at Renji Hospital who met the following inclusion criteria: a diagnosis of DM based on the criteria proposed by Bohan and Peter [13, 14] or clinically amyopathic DM criteria proposed by Sontheimer [15], were included in this study. The diagnosis was reconfirmed according to the 239th European Neuromuscular Center DM criteria [14]. Only patients with an ILD disease duration of ≤ 3 months from the date of diagnosis were enrolled as the eligible inception cohort.
Anti-MDA5 antibodies were identified using line-blotting assay (Euroimmun) and/or enzyme-linked immunosorbent assays (ELISA). ILD was diagnosed based on respiratory symptoms, pulmonary function test results, and high-resolution CT (HRCT) imaging [16]. Patients who had a follow up exceeding 6 months and had a baseline anti-MDA5 antibody ELISA quantitative measurement along with at least one subsequent measurement were subjected to further trajectory analysis. All patients with follow-up beyond 6 months were included in both the antibody trajectory analysis and the long-term outcome analysis (including survival at 12, 24, and 36 months). Patients who died or underwent lung transplantation within 6 months of follow-up were excluded due to rapid disease progression and short disease duration. Patients were also excluded if they had overlap syndromes with other connective tissue diseases, or had underlying malignancies. The flow chart for the study is shown in Fig. 1. All patients provided written informed consent, and the study was approved by the Ethics Committee of Renji Hospital, Shanghai.
Fig. 1.
Flow chart of the inception cohort of anti-MDA5 + DM-ILD
Data collection
At baseline, demographic data, comprehensive medical histories, physical examinations, laboratory tests and treatments, were collected for all patients. The arterial oxygen partial pressure to fractional inspired oxygen ratio (PaO2/FiO2, or oxygenation index) was calculated for each patient based on the results of arterial blood gas analysis. Infection was defined as any microbiologically confirmed infection present at baseline. Apart from the baseline data collection, after the intensive 6-month follow-up, pulmonary function tests, pulmonary HRCT and anti-MDA5 antibody titers were measured every 3 to 6 months in a scheduled manner, tailored to the patients’ condition and the treating physician’s judgement. All data were recorded in an EDC system.
Definitions
Seroconversion was defined as a change from positive to negative for anti-MDA5 antibodies. Low disease activity (LDA) was defined as the absence of active skin eruptions, the absence of new ground-glass opacity (GGO) or consolidation on HRCT, no need for home-based oxygen supplementation, and with a daily prednisone dose of ≤ 7.5 mg, in conjunction with stable immunosuppressive and/or anti-fibrotic therapy. Given that the majority of anti-MDA5 + DM-ILD cases presented as clinically amyopathic DM, we did not evaluate the muscle domain in defining LDA. Prednisone-free remission was defined the same as LDA except for being prednisone-free. A more stringent drug-free remission was defined as the ability to stop all immunosuppressive drugs and anti-fibrotic agents.
ELISA for anti-MDA5 antibody titers
Anti-MDA5 antibody titers were tracked from baseline to the last measurement (up to three years) for trajectory analysis. All serum samples were suitable for point-of-care testing of anti-MDA5 antibodies, including validated samples with high and low concentrations covering the assay’s detection range.
For the ELISA, plates were coated overnight at 4 °C with 10 µg per well of full-length recombinant MDA5 protein (Shanghai Free Zone Biotechnology Co., Ltd). Blocking was performed with 1% BSA. Patient serum was diluted at 1:100 and incubated, followed by incubation with HRP-conjugated anti-human IgG antibody (1:60,000, Jackson, USA) for 30 min. The reaction was stopped with TMB substrate (Invitrogen, USA), and absorbance was read at 450 nm (Euroimmun Sprinter XL, Germany).
Each run included at least two quality control wells (low and high concentration, respectively). A Levey-Jennings chart was used to monitor trends, and the coefficient of variation (CV) for quality controls was maintained within 15% of the target. The cut-off value was defined as the mean ± 3SD of results from 94 healthy donors [17]; values below 35 RU/mL were considered negative.
Latent class mixture models (LCMMs) establishment
The trajectories of anti-MDA5 antibody titers were investigated using LCMMs, via the ‘lcmm’ R package, following a validated eight-step methodology [18–20]. We tested mixed-effects models with random intercepts for latent classes ranging from K = 1–10. The optimal model was selected based on the Bayesian Information Criterion (BIC) and further refined in three parameters (normalized or unnormalized anti-MDA5 antibody titer, spline, time). Model adequacy was confirmed by the average posterior probability of assignment and the odds of correct classification. Further details are available in supplementary Methods, and Supplementary Tables S1-S3 and Figure S1.
Statistical analyses
Patients’ baseline characteristics were then compared across latent classes. Quantitative data was compared by non-parametric tests. Categorical data were compared by Pearson’s χ2 test or Fisher’s exact test, as appropriate. Kaplan-Meier curves at 12, 24 and 36 months were then constructed to compare the latent classes. The log-rank test was applied to analyze differences in Kaplan-Meier curves.
The percentage decrease in anti-MDA5 antibody levels over the initial 3 months (ΔMDA5@3m) was calculated. Missing 3-month antibody titers were imputed using patient-specific linear interpolation of longitudinal measurements. This method employed bounded interpolation, which avoids extrapolation, and included an optional log-transformation to account for exponential trends. The receiver operating characteristic (ROC) curve analysis was performed for the change of anti-MDA5 antibody titer compared to baseline and the areas under the curve (AUC) were calculated. A Cox regression model was applied to identify independent factors for LDA. All candidate parameters with P-values < 0.1 in the univariable analysis were incorporated into the multivariable Cox regression for LDA.
We conducted several sensitivity analyses to test the robustness of our results. First, we repeated the trajectory modeling analysis, Kaplan-Meier curves and ROC analysis among patients who did not receive anti-CD20 therapy, or immunoadsorption. Second, we conducted ROC curve analysis on the non-imputer data. A p-value of < 0.05 was considered statistically significant. Data analysis was performed using IBM SPSS (version 26.0) and R software (version 4.3.2).
Results
Patient characteristics
Between July 2017 and June 2024, a total of 525 patients with baseline anti-MDA5 antibody titers were enrolled in this study. Among them, 348 exhibited an oxygenation index (OI) of ≥ 300 mmHg, 92 had an index between 200 and 300 mmHg, and 85 presented with an index below 200 mmHg. Within this cohort, 131 patients died within the first six months, while the remaining 394 survived for more than 6 months (Fig. 1). The baseline clinical manifestations and laboratory results are presented in Supplementary Table S4. The deceased group was older and exhibited lower PaO2/FiO2 and lymphocyte levels, alongside higher levels of serum ferritin (SF), lactate dehydrogenase (LDH), and Krebs von den Lungen-6 (KL-6), and a higher incidence of spontaneous pneumomediastinum or pneumothorax, compared to the survivor group (all P < 0.001; Supplementary Table S4). The proportions of patients receiving Janus Kinase (JAK) inhibitors or anti-fibrosis drugs were comparable between the two groups. However, as the deceased group exhibited more severe conditions, they received a broader range of additional immunotherapies, such as tacrolimus, cyclophosphamide, and rituximab. Of note, the levels of anti-MDA5 antibody in the deceased group were significantly higher than those in the survivors within 6 months (174.47 ± 49.87 RU/mL vs. 152.36 ± 47.95 RU/mL, P < 0.001; Supplementary Figure S2A).
From the 392 patients with more than 6-month follow-up, 84 patients were excluded due to either the absence of additional anti-MDA5 antibody titer measurements or a lack of clinical data necessary for trajectory analysis. We compared baseline parameters between the 310 eligible subjects and excluded patients and found no statistically significant differences in gender, age, OI, or key laboratory indicators, including anti-MDA5 antibody levels, LDH, SF levels and lymphocyte counts (Supplementary Table S5 and Figure S2B). Most of the surviving patients received treatment with glucocorticoids at a dosage of ≥ 1 mg/kg combined with JAK inhibitors at baseline. Both groups had comparable proportions of patients receiving other immunotherapies or anti-fibrotic drugs. Therefore, the 310 eligible patients were considered representative of the anti-MDA5 + DM cohort with a chronic course and were included in the trajectory analysis. Cutaneous manifestations were observed in 97.4% of these eligible patients. Of the 310 patients, follow-up was completed by 310 (100%) at 12 months, 248 (80.0%) at 24 months, and 158 (54.8%) at 36 months. The mean follow-up time was 26.9 ± 11 months, and the median number of anti-MDA5 antibody titer measurements recorded was 4 (IQR: 3–6) among the eligible survivors.
Anti-MDA5 trajectory analyses for patients with chronic disease course
Using group-based trajectory modeling, two distinct trajectory patterns of anti-MDA5 antibody levels were identified, which were approximately equally distributed (Fig. 2A-D). Figure 2 shows the two divergent profiles: “slow seroconvertors” [Trajectory 1; n = 148 (47.7%)], and “fast seroconvertors” [Trajectory 2; n = 162 (52.3%)]. These trajectories showed variable declines in anti-MDA5 antibody. Trajectory 2 showed a rapid decrease within the first six months, while Trajectory 1 displayed a gradual decline in antibody levels over time. Fast seroconvertors of anti-MDA5 antibody showed lower baseline anti-MDA5 antibody levels (142.99 ± 52.90 vs. 167.60 ± 41.44 RU/mL, P < 0.001), comprised a higher proportion of women (73.9% vs. 62.9%, P = 0.04), exhibited a greater percentage of patients with OI < 300 (27.8% vs. 14.2%, P = 0.01), and had lower forced vital capacity (FVC) levels (67.7 ± 21.8 vs. 73.3 ± 19.2%, P = 0.02, Table 1). The analysis of baseline clinical features revealed no statistical differences in age, lymphocyte counts, LDH, SF, or other critical variables between the two trajectory groups (Table 1). In other words, fast seroconvertors was characterized by female predominance, lower baseline anti-MDA5 antibody level, and more severe baseline ILD. Most patients who underwent immunoadsorption experienced fast seroconversion (Table 1). Higher dosage exposure to glucocorticoids and higher frequent use of intravenous immunoglobulin (IVIG), and less frequent use of calcineurin inhibitors at baseline, may also enhance the probability of fast seroconversion. However, no significant differences in the rate of seroconversion were observed for other immunotherapies (Table 1).
Fig. 2.
Two trajectories of anti-MDA5 antibody levels in the 6-month survivors of anti-MDA5 + DM-ILD. A Trajectory plots with average and 95% predictive intervals of anti-MDA5 antibody levels for each cluster. The x axis represents time, and the y axis represents the level of normalized anti-MDA5 titers converted by ‘lcmm’ package. B Individual level ‘spaghetti plots’ showed changes of original anti-MDA5 antibody levels of each patient in all trajectories and in Trajectory 1 (C), and Trajectory 2 (D) defined as ‘MDA5 slow seroconvertors’, and ‘MDA5 fast seroconvertors’
Table 1.
Baseline characteristics of anti-MDA5 antibody-positive dermatomyositis patients between two groups
| Characteristics | Eligible patients (n = 310) | MDA5 fast seroconvertors (n = 162) | MDA5 slow seroconvertors (n = 148) | P value |
|---|---|---|---|---|
| Age, years | 49.26 ± 11.57 | 50.39 ± 11.35 | 48.12 ± 11.75 | 0.1 |
| Sex, female (n, %) | 215 (69.4) | 113 (73.9) | 90 (62.9) | 0.04 |
| PaO2/FiO2 (mmHg) | ||||
| ≥300 | 244 (78.7) | 117 (72.2) | 127 (85.8) | 0.01 |
| ≥ 200, and, < 300 | 50 (16.1) | 34 (21.0) | 16 (11.9) | |
| < 200 | 16 (5.2) | 11 (6.8) | 5 (3.4) | |
| Laboratory examination | ||||
| Forced vital capacity (%) | 70.46 ± 20.71 | 67.7 ± 21.8 | 73.3 ± 19.2 | 0.02 |
| CRP (mg/L) | 2(0.5, 5.56) | 2.0 (0.64, 6.01) | 2 (0.5, 4.69) | 0.25 |
| Lymphocytes (×109/L) | 1.0 ± 0.66 | 1.01 ± 0.67 | 0.99 ± 0.65 | 0.73 |
| ESR (mm/h) | 31.6 ± 21.62 | 33.29 ± 22.56 | 29.77 ± 20.47 | 0.15 |
| SF (ng/ml) | 743.7 (300.43, 1414.0) | 742.3 (309.8, 1275.23) | 751.35 (271.88, 1583.45) | 0.28 |
| LDH (U/L) | 332.6 ± 297.6 | 345.04 ± 390.34 | 318.65 ± 130.34 | 0.44 |
| CK (U/L) | 51 (28.0, 81) | 48 (24.25, 72.75) | 57.0 (32, 88.0) | 0.05 |
| ANA positive | 79 (25.48) | 38 (23.46) | 41 (27.7) | 0.39 |
| Anti-Ro52 antibody positive | 151 (48.71) | 78 (48.15) | 73 (49.32) | 0.84 |
| Anti-MDA5 antibody titers (RU/mL) | 154.8 ± 49.2 | 142.99 ± 52.9 | 167.60 ± 41.44 | < 0.001 |
| Treatment (n, %) | ||||
| Maximum dosage of MP (mg/d) | 80 (60, 120) | 80 (80, 120) | 80 (60,80) | < 0.001 |
| Maximum dosage of MP ≥ 200 mg | 24 (7.7) | 14 (8.64) | 10 (6.76) | 0.54 |
| JAK inhibitors | 306 (98.7) | 160 (98.8) | 146 (98.6) | 0.93 |
| CNI | 77 (24.8) | 31 (19.1) | 46 (31.1) | 0.02 |
| Anti-CD20 monoclonal antibody | 16 (5.2) | 8 (4.9) | 8 (5.4) | 0.85 |
| Immunoglobulin | 126 (40.6) | 75 (46.3) | 51 (34.5) | 0.02 |
| Immunoadsorption | 14 (4.5) | 10 (6.17) | 4 (2.70) | 0.18 |
| Plasma exchange | 2 (0.6) | 1 (0.6) | 1 (0.7) | 1.0 |
| Anti-fibrotic drugs | 143 (46.1) | 80 (49.4) | 63 (42.6) | 0.23 |
| Others* | 55 (17.7) | 28 (17.3) | 32 (21.6) | 0.32 |
| Comorbidity (n, %) | ||||
| Spontaneous pneumomediastinum and/or pneumothorax | 19 (6.1) | 11 (6.9) | 8 (5.4) | 0.6 |
| Infections | 117 (37.7) | 69 (42.6) | 48 (32.4) | 0.06 |
*Other treatments, including MTX Methotrexate, CTX Cyclophosphamide, abatacept, IL-6 inhibitors, and zanubrutinib, either alone or in combination. PaO2/FiO2, oxygen partial pressure to fractional inspired oxygen ratio, CRP C-reactive protein, LDH Lactate dehydrogenase, CK Creatine kinase, ANA Anti-nuclear antibody, MP Methylprednisolone, JAK Inhibitors, JAK Janus Kinase inhibitors, CNI Calcineurin inhibitor
Long-term outcomes
During the 36-month follow-up period, 7 (4.7%) patients died in the slow seroconversion group, whereas only 1 (0.6%) death was observed in the fast seroconversion group (P = 0.03; Table 2). The daily prednisone dosages were comparable between fast seroconvertors and slow seroconvertors at 6 months; however, the difference became significant with a much lower daily dosage of prednisone for fast seroconvertors at 12 months (8.13 ± 4.61 mg vs. 12.36 ± 9.17 mg, P < 0.001; Table 2).
Table 2.
Long-term outcomes of anti-MDA5 antibody-positive dermatomyositis patients between two groups
| Outcomes | Eligible patients (n = 310) | MDA5 fast seroconvertors (n = 162) | MDA5 slow seroconvertors (n = 148) | P value |
|---|---|---|---|---|
| Follow-up time (months) | 26.9 ± 11.0 | 28.2 ± 9.3 | 28.9 ± 9.2 | 0.15 |
| Death | 8 (2.6) | 1 (0.6) | 7 (4.7) | 0.03 |
| Seroconversion within 36 months | 151 (48.7) | 126 (77.78) | 25 (16.89) | < 0.001 |
| Time of seroconversion (months) | 9 (5.0, 18) | 9 (5, 15) | / | |
| Prednisone dosage at 6 months (mg/d) | 14.59 ± 10.44 | 13.98 ± 6.39 | 14.56 ± 5.99 | 0.41 |
| Prednisone dosage at 12 months (mg/d) | 9.98 ± 7.48 | 8.13 ± 4.61 | 12.36 ± 9.17 | < 0.001 |
| Low disease activity at 12 months (n, %) | 162 (52.3) | 107 (66.0) | 55 (37.2) | < 0.001 |
| Prednisone-free remission at 24 months (n, %) | 40 (12.9) | 27 (16.7) | 13 (8.8) | 0.02 |
| Prednisone-free remission at 36 months (n, %) | 64 (20.6) | 45 (27.8) | 19 (12.8) | 0.001 |
| Drugs-free remission at 36 months (n, %) | 21 (6.8) | 17 (10.5) | 4 (2.7) | 0.006 |
Furthermore, to better capture disease activity status, we developed a clinically intuitive LDA or remission definition, i.e., quiescent disease in both cutaneous and ILD domain, along with a daily prednisone dosage ≤ 7.5 mg or zero prednisone, which signified LDA or remission, respectively (Fig. 3A). Table 2; Fig. 3B showed the proportions achieving LDA or remission for each group. Interestingly, a greater percentage of patients in the fast seroconvertors group achieved LDA by 12 months [66.0% vs. 37.2%, P < 0.001, odds ratio (OR): 3.29, 95%CI: 2.07–5.24; Table 2]. Furthermore, the proportions of patients achieving prednisone-free remission were significantly higher among fast seroconvertors than among slow seroconvertors at both 24 months (16.7% vs. 8.8%, OR: 2.17, 95%CI: 1.08–4.37, P = 0.02) and 36 months (27.8% vs. 12.8%, OR: 2.61, 95%CI: 1.45–4.72, P = 0.001). Additionally, 17 patients (10.5%) in the fast seroconversion group and 4 patients (2.7%) in the slow seroconversion group achieved drug-free remission at 36 months (OR: 4.22, 95%CI: 1.39–12.85, P = 0.006). In the survival analysis, more patients in the fast seroconversion group achieved LDA by 12 months (P < 0.0001), prednisone-free remission by 24 months (P = 0.02) and both prednisone-free remission (P = 0.001) and drug-free remission (P = 0.005) by 36 months, respectively, compared with those in the slow seroconvertors group (Fig. 3C -F).
Fig. 3.
Outcomes between ‘MDA5 slow seroconvertors’ and ‘MDA5 fast seroconvertors’. (A) The definition of low disease activity (LDA) and prednisone-free (pred-free) remission. (B) The proportion of each trajectory group achieving LDA and prednisone-free or drug-free remission. (C-F) Comparison of LDA at 12 months (C), pred-free remission at 24 months (D), pred-free remission at 36 months (E) and drug-free remission at 36 months (F) in Kaplan Meier curves
ΔMDA5@3m associated with disease trajectory and LDA
The fast seroconversion group showed a significantly higher percentage of anti-MDA5 antibody seroconversion compared to the slow seroconversion group (77.8% vs. 16.9%, P < 0.001), with a median time-to-seroconversion of 9 months (IQR: 5–15) (Table 2; Fig. 4A). In contrast, the time-to-seroconversion in the slow seroconversion group was not applicable due to the sustained anti-MDA5 antibody positivity throughout the follow-up period (Fig. 4A).
Fig. 4.
Comparison of anti-MDA5 antibody seroconversion between different seroconvertors and predictive value of ΔMDA5@3m. A Overall survival curve of seroconversion between ‘MDA5 slow seroconvertors’, and ‘MDA5 fast seroconvertors’. B Receiver operating characteristic (ROC) curves of ΔMDA5@3m and anti-MDA5 antibody level at baseline in the eligible survivors for trajectory classification. ΔMDA5@3m, the relative changes of anti-MDA5 antibody levels at the first three months.
Although baseline levels of anti-MDA5 antibodies exhibited differences, its performance in predicting the two trajectories was suboptimal (AUC = 0.65, Fig. 4B). Based on the distinct characteristics of the two trajectories, the percentage decline in anti-MDA5 antibody levels during the first three months (ΔMDA5@3m) was calculated to enhance predictive accuracy. The AUC for ΔMDA5@3m in terms of trajectory class prediction was 0.86, with a cut-off point of 33.0% (Fig. 4B). Patients exhibiting more than a 33.0% reduction in anti-MDA5 titer at 3 months from baseline had a high likelihood of being classified as fast seroconvertors, with sensitivity and specificity rates of 70.7% and 83.2%, respectively.
Furthermore, a multivariable Cox regression model identified a reduction in ΔMDA5@3m of ≥ 33.0% as an independent favorable factor for achieving LDA (HR 1.96, 95% CI: 1.33–2.90, P < 0.001) (Supplementary Table S6).
Sensitivity analyses
To minimize interference from treatments that directly clear MDA5 antibodies, we excluded 8 patients who received anti-CD20 therapy, and 14 who underwent immunoadsorption. LCMM trajectory modeling also identified two distinct trajectories as the best model in the sensitivity analysis. As shown in Supplementary Figure S3, the shapes of the two trajectories closely resembled those shown in Fig. 2. As shown in Supplementary Figure S4, the proportions achieving LDA by 12 months and prednisone-free remission by 36 months were higher in fast-seroconvertors (64.8% and 29.7%, respectively), and lower in slow-seroconvertors (37.6% and 12.8%, respectively). These results were consistent with the primary analysis presented in Table 2; Fig. 3. Moreover, the performance of ΔMDA5@3m for trajectory discrimination was also excellent (AUC = 0.88) (Supplementary Figure S5). Considering the study timeline, it is probable that a large subset of the cohort admitted during the COVID-19 pandemic. The trajectory modeling analysis additionally demonstrates a comparable pattern in anti-MDA5 + DM patients enrolled between January 2020 and June 2024 (Supplementary Figure S6).
Another sensitivity analysis using non-imputer data of ΔMDA5@3m yielded similar results. The AUCs of ΔMDA5@3m for trajectory classification were 0.86 in the overall eligible cohort and 0.89 in patients who did not receive anti-CD20 therapy, or immunoadsorption (Supplementary Figure S7).
Discussion
This study analyzes the relationship between anti-MDA5 antibody levels and long-term outcomes in a relatively large inception cohort of anti-MDA5 + DM patients, revealing two trajectories of anti-MDA5 antibody changes in 6-month survivors over time. This is the first study to employ a semi-supervised machine learning approach, to delineate the trajectories of the disease in a real-world setting. These trajectories provide a framework for developing, testing, and implementing strategies aimed at improving the outcomes of this challenging disease.
Our research has revealed the significant clinical value of quantitatively detecting anti-MDA5 antibodies. First, baseline anti-MDA5 titers were slightly higher in patients who died within 6 months compared to those who survived beyond 6 months, consistent with previous studies in several smaller cohorts [7, 21, 22]. Second, for 6-month survivors, the target population for this study, two distinct trajectories were identified. Again, the ‘fast seroconvertors’, who exhibited better outcomes, had relatively lower baseline anti-MDA5 antibody titers as opposed to ‘slow seroconvertors’. This evidence, in line with other studies, suggests that lower baseline anti-MDA5 antibody levels are linked to more favorable prognosis, such as long-term survival, LDA or remission. However, the discriminative value based on baseline anti-MDA5 antibody levels is minimal with an AUC around 0.65. On the other hand, the performance of the slope of the anti-MDA5 antibody decline is more helpful in terms of differentiating the two trajectories by definition.
The strong correlation between the trajectories of anti-MDA5 antibodies and disease prognosis suggests that anti-MDA5 antibodies can serve not only as diagnostic markers for dermatomyositis, but also as dynamic indicators of achieving LDA at 12 months, prednisone-free remission at 24 months and drug-free remission at 36 months. Previous studies have indicated that the titers of anti-MDA5 antibodies may decrease following effective treatment [7, 8]; however, the limited number of cases examined was insufficient to fully describe the population-level trajectories of anti-MDA5 antibodies over the long-term course of dermatomyositis. This study, with a sufficiently large sample size, effectively demonstrates the variations in the trajectories of anti-MDA5 antibodies within a large population and underscores the close association of these variations with disease prognosis. The evidence presented further emphasizes the importance of closely monitoring anti-MDA5 antibodies during the treatment follow-up process, serving as both a predictor of disease prognosis and a suggestion that the rapid clearance of these antibodies may benefit the acute phase prognosis of patients [23].
The fast and slow seroconvertors exhibited distinctly different patterns in the progression of chronic disease. In our cohort, patients with severe pulmonary involvement at presentation received more intensive immunotherapy, which in turn facilitated rapid antibody clearance. These findings suggest a plausible treatment paradigm: a more effective treatment protocol may steer the disease course toward “fast seroconversion” and lead to favorable outcomes. This hypothesis warrants further investigation; however, our recent ProJAK real-world study supports this notion [24]. In the ProJAK study, patients who received protein A immunoadsorption plus a JAK inhibitor (ProJAK group) were compared to those treated with JAK inhibitors alone (JAK group). Adding immunoadsorption can rapidly reduce anti-MDA5 antibody titers by 40% within two weeks, and the combination protocol (ProJAK) is associated with a 20% improvement in survival compared to a JAK inhibitor-based regimen among anti-MDA5 + DM-RPILD patients [24]. Notably, among the eight patients who underwent emergency bilateral lung transplantation (only one of whom received immunoadsorption prior to the procedure), all exhibited a decline in anti-MDA5 antibody titers of more than 50% within 48 h post-surgery, with most achieving seroconversion within one to three months. These observations further highlight the potential pathogenic role of anti-MDA5 antibodies, which may reflect the burden of autoimmune disease; as this burden is alleviated, the disease improves.
Supported by our previous research [1, 8, 25], anti-MDA5 antibody titers are a validated marker of disease progression. The proposed mechanism begins with viral infections damaging alveolar cells, which exposes MDA5 antigens and triggers anti-MDA5 antibody production via plasmacytoid dendritic cells (pDCs) [26]. These antibodies then accumulate in the lungs, forming immune complexes that perpetuate a vicious cycle of cellular damage and further antibody production [27, 28], correlating with ILD worsening [25]. Consequently, effective treatment reduces antigen exposure and antibody titers, while disease relapse, by increasing inflammation and damage, slows the decline or even raises antibody levels. This dynamic underpins the utility of anti-MDA5 titers as an early indicator of disease severity, treatment response and long-term prognosis.
The stark contrast in seroconversion rates and timelines between the fast and slow seroconversion groups underscores the existence of two distinct phenotypic patterns with profound clinical implications in patients with long-term disease courses. Our analysis indicated ΔMDA5@3m to be a potential and clinically practical predictor for long-term disease prognosis in 6-month survivors with anti-MDA5 antibodies. Serial measurement, particularly the calculation of ΔMDA5@3m, provides a critical early window into predicting long-term disease behavior, enabling more tailored and proactive management strategies. Once “slow seroconversion” is identified, it should prompt intensified vigilance in both monitoring and treatment. For these patients, enhancing the comprehensiveness and frequency of disease surveillance is necessary, including clinical symptoms, serum biomarkers, pulmonary function tests, and pulmonary HRCT. If persistent non-response is observed in “slow seroconvertors”, immunotherapy should be evaluated for adjustment as early as possible.
Our study has several limitations. First, this retrospective, single-center study was limited by incomplete clinical data and different treatment intensities, necessitating validation in larger, multi-center, prospective cohorts. Second, the trajectory analysis was limited to 6-month survivors and did not account for patients with more severe lung involvement, who likely contributed to early mortality. This survivorship bias compromises the comprehensive analysis of disease behavior in all patients with anti-MDA5 + DM. Third, the average follow-up time was insufficient to address very long-term outcomes, and it did not adequately capture mortality patterns. Finally, COVID-19 was a confounder associated with treatment and outcomes in anti-MDA5 + DM patients, but our retrospective study lacked information related to COVID-19.
In summary, we identified two distinct patterns of change in anti-MDA5 antibody titers in 6-month survivors with anti-MDA5 + DM. The association of slow seroconversion during follow-up with unfavorable clinical outcomes (e.g., non-achievement of LDA or remission) underscores the importance of personalized management and the monitoring of anti-MDA5 antibody titers for patients with anti-MDA5 + DM. Further investigation into tailored strategies for rapidly reducing antibody levels and maintaining low titers is warranted. Our findings provide a framework for future research aimed at improving long-term disease outcomes, including LDA and remission, for 6-month survivors.
Contributions
All people have drafted the work or substantively revised it, and have approved the submitted version. Z.C., and H.W. wrote the main manuscript text, designed the work and analyzed data. L.W., Y.F., W.X. and X.W. curated the data; D.Z. and K.W. designed methodology and created the models; S.Y., and L.L. wrote the main manuscript text, designed and visualized the work, and supervised the data.
Supplementary Information
Authors’ contributions
All people have drafted the work or substantively revised it, and have approved the submitted version. Z.C., and H.W. wrote the main manuscript text, designed the work and analyzed data. L.W., Y.F., W.X. and X.W. curated the data; D.Z. and K.W. designed methodology and created the models; S.Y., and L.L. wrote the main manuscript text, designed and visualized the work, and supervised the data.
Funding
The study was sponsored by Shanghai Science and Technology Committee Rising-Star Program (24YF2725100) and Shanghai Pujiang Young Rheumatologists Training Program (SPROG202401) by Huijing Wang.
Data availability
Data are available upon reasonable request.
Declarations
Ethics approval and consent to participate
The retrospective study was approved by the ethics committees of Renji Hospital, Shanghai Jiao Tong University School of Medicine in accordance with the Declaration of Helsinki. Participants gave informed consent to participate in the study before.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Clinical trial number
Not applicable.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Zhiwei Chen and Huijing Wang contributed equally to this work.
Contributor Information
Liangjing Lu, Email: lu_liangjing@163.com.
Shuang Ye, Email: ye_shuang2000@163.com.
References
- 1.Lu X, Peng Q, Wang G. Anti-MDA5 antibody-positive dermatomyositis: pathogenesis and clinical progress. Nat Rev Rheumatol. 2024;20:48–62. [DOI] [PubMed] [Google Scholar]
- 2.Wu W, Guo L, Fu Y, Wang K, Zhang D, Xu W, Chen Z, Ye S. Interstitial lung disease in Anti-MDA5 positive dermatomyositis. Clin Rev Allergy Immunol. 2021;60:293–304. [DOI] [PubMed] [Google Scholar]
- 3.Ye S, Chen XX, Lu XY, Wu MF, Deng Y, Huang WQ, Guo Q, Yang CD, Gu YY, Bao CD, Chen SL. Adult clinically amyopathic dermatomyositis with rapid progressive interstitial lung disease: a retrospective cohort study. Clin Rheumatol. 2007;26:1647–54. [DOI] [PubMed] [Google Scholar]
- 4.Hamaguchi Y, Kuwana M, Hoshino K, Hasegawa M, Kaji K, Matsushita T, Komura K, Nakamura M, Kodera M, Suga N, Higashi A, Ogusu K, Tsutsui K, Furusaki A, Tanabe H, Sasaoka S, Muro Y, Yoshikawa M, Ishiguro N, Ayano M, Muroi E, Fujikawa K, Umeda Y, Kawase M, Mabuchi E, Asano Y, Sodemoto K, Seishima M, Yamada H, Sato S, Takehara K, Fujimoto M. Clinical correlations with dermatomyositis-specific autoantibodies in adult Japanese patients with dermatomyositis: a multicenter cross-sectional study. Arch Dermatol. 2011;147:391–8. [DOI] [PubMed] [Google Scholar]
- 5.Jin Q, Fu L, Yang H, Chen X, Lin S, Huang Z, Gao B, Tian X, Jiang W, Shu X, Lu X, Wang G, Peng Q. Peripheral lymphocyte count defines the clinical phenotypes and prognosis in patients with anti-MDA5-positive dermatomyositis. J Intern Med. 2023;293:494–507. [DOI] [PubMed] [Google Scholar]
- 6.Lian X, Ye Y, Zou J, Wu C, Ye S, Guo Q, Chen S, Lu L, Wang R, Fu Q, Bao C. Longitudinal study of patients with antimelanoma differentiation-associated gene 5 antibody-positive dermatomyositis-associated interstitial lung disease. Rheumatology (Oxford). 2023;62:1910–9. [DOI] [PubMed] [Google Scholar]
- 7.Sato S, Kuwana M, Fujita T, Suzuki Y. Anti-CADM-140/MDA5 autoantibody titer correlates with disease activity and predicts disease outcome in patients with dermatomyositis and rapidly progressive interstitial lung disease. Mod Rheumatol. 2013;23:496–502. [DOI] [PubMed] [Google Scholar]
- 8.Matsushita T, Mizumaki K, Kano M, Yagi N, Tennichi M, Takeuchi A, Okamoto Y, Hamaguchi Y, Murakami A, Hasegawa M, Kuwana M, Fujimoto M, Takehara K. Antimelanoma differentiation-associated protein 5 antibody level is a novel tool for monitoring disease activity in rapidly progressive interstitial lung disease with dermatomyositis. Br J Dermatol. 2017;176:395–402. [DOI] [PubMed] [Google Scholar]
- 9.Gono T, Kawaguchi Y, Satoh T, Kuwana M, Katsumata Y, Takagi K, Masuda I, Tochimoto A, Baba S, Okamoto Y, Ota Y, Yamanaka H. Clinical manifestation and prognostic factor in anti-melanoma differentiation-associated gene 5 antibody-associated interstitial lung disease as a complication of dermatomyositis. Rheumatology (Oxford). 2010;49:1713–9. [DOI] [PubMed] [Google Scholar]
- 10.Lian X, Zou J, Guo Q, Chen S, Lu L, Wang R, Zhou M, Fu Q, Ye Y, Bao C. Mortality risk prediction in amyopathic dermatomyositis associated with interstitial lung disease: the FLAIR model. Chest. 2020;158:1535–45. [DOI] [PubMed] [Google Scholar]
- 11.Sakamoto S, Okamoto M, Kaieda S, Fujimoto K, Nagata S, Tominaga M, Nakamura M, Zaizen Y, Nouno T, Koga T, Kawayama T, Kuwana M, Ida H, Hoshino T. Low positive titer of anti-melanoma differentiation-associated gene 5 antibody is not associated with a poor long-term outcome of interstitial lung disease in patients with dermatomyositis. Respir Investig. 2018;56:464–72. [DOI] [PubMed] [Google Scholar]
- 12.Cao H, Pan M, Kang Y, Xia Q, Li X, Zhao X, Shi R, Lou J, Zhou M, Kuwana M, Ding X, Zheng J. Clinical manifestations of dermatomyositis and clinically amyopathic dermatomyositis patients with positive expression of anti-melanoma differentiation-associated gene 5 antibody. Arthritis Care Res (Hoboken). 2012;64:1602–10. [DOI] [PubMed] [Google Scholar]
- 13.Bohan A, Peter JB. Polymyositis and dermatomyositis (first of two parts). N Engl J Med. 1975;292:344–7. [DOI] [PubMed] [Google Scholar]
- 14.Akira M, Hamada H, Sakatani M, Kobayashi C, Nishioka M, Yamamoto S. CT findings during phase of accelerated deterioration in patients with idiopathic pulmonary fibrosis. AJR Am J Roentgenol. 1997;168:79–83. [DOI] [PubMed] [Google Scholar]
- 15.Ichikado K, Suga M, Müller NL, Taniguchi H, Kondoh Y, Akira M, Johkoh T, Mihara N, Nakamura H, Takahashi M, Ando M. Acute interstitial pneumonia: comparison of high-resolution computed tomography findings between survivors and nonsurvivors. Am J Respir Crit Care Med. 2002;165:1551–6. [DOI] [PubMed] [Google Scholar]
- 16.Raghu G, Remy-Jardin M, Myers JL, Richeldi L, Ryerson CJ, Lederer DJ, Behr J, Cottin V, Danoff SK, Morell F, Flaherty KR, Wells A, Martinez FJ, Azuma A, Bice TJ, Bouros D, Brown KK, Collard HR, Duggal A, Galvin L, Inoue Y, Jenkins RG, Johkoh T, Kazerooni EA, Kitaichi M, Knight SL, Mansour G, Nicholson AG, Pipavath SNJ, Buendía-Roldán I, Selman M, Travis WD, Walsh S, Wilson KC. Diagnosis of idiopathic pulmonary Fibrosis. An official ATS/ERS/JRS/ALAT clinical practice guideline. Am J Respir Crit Care Med. 2018;198:e44–68. [DOI] [PubMed] [Google Scholar]
- 17.Wang K, Zhao J, Wu W, Xu W, Sun S, Chen Z, Fu Y, Guo L, Du H, Ye S. RNA-Containing immune complexes formed by Anti-Melanoma differentiation associated gene 5 autoantibody are potent inducers of IFN-alpha. Front Immunol. 2021;12:743704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lennon H, Kelly S, Sperrin M, Buchan I, Cross AJ, Leitzmann M, Cook MB, Renehan AG. Framework to construct and interpret latent class trajectory modelling. BMJ Open. 2018;8:e020683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Proust-Lima C, Philipps V, Liquet B. Estimation of extended mixed models using latent classes and latent processes: the R package Lcmm. J Stat Softw. 2017;78:1–56. [Google Scholar]
- 20.Zhang D, Sun F, Chen J, Ding H, Wang X, Shen N, Li T, Ye S. Four trajectories of 24-hour urine protein levels in real-world lupus nephritis cohorts. RMD Open. 2023;9(2):e002930. [DOI] [PMC free article] [PubMed]
- 21.Nishioka A, Tsunoda S, Abe T, Yoshikawa T, Takata M, Kitano M, Matsui K, Nakashima R, Hosono Y, Ohmura K, Mimori T, Sano H. Serum neopterin as well as ferritin, soluble interleukin-2 receptor, KL-6 and anti-MDA5 antibody titer provide markers of the response to therapy in patients with interstitial lung disease complicating anti-MDA5 antibody-positive dermatomyositis. Mod Rheumatol. 2019;29:814–20. [DOI] [PubMed] [Google Scholar]
- 22.Hayashi S, Nawata Y, Takabayashi K, Iwamoto I, Yoshida S. [Hemophagocytic syndrome observed in a patient with systemic lupus erythematosus]. Ryumachi. 1994;34:779–85. [PubMed] [Google Scholar]
- 23.Wu W, Guo B, Sun W, Chen D, Xu W, Chen Z, Fu Y, Ye Y, Lyu X, Xue Z, Wang K, Zhao J, Xie C, Chen Y, Ye C, Dai M, Fan W, Li J, Wang X, Xue Y, Wan W, Sun L, Wu H, Luo Q, Han Q, Fu Q, Ye S. Effectiveness and safety of Tofacitinib versus calcineurin inhibitor in interstitial lung disease secondary to anti-MDA5-positive dermatomyositis: a multi-centre cohort study. Eur Respir J. 2025;65(5):240148. [DOI] [PMC free article] [PubMed]
- 24.Fu Y, Chen Z, Chen J, Wu W, Gu L, Ma Y, Wang R, Wang K, Xie C, Chen Y, Cai J, Fu Q, Wang X, Ye S. Efficacy of JAK inhibitor combined with protein A immunoadsorption in MDA5+ dermatomyositis with RPILD: A real-world study. Med. 2025;6(10):100802. [DOI] [PubMed]
- 25.Zaizen Y, Okamoto M, Azuma K, Fukuoka J, Hozumi H, Sakamoto N, Suda T, Mukae H, Hoshino T. Enhanced immune complex formation in the lungs of patients with dermatomyositis. Respir Res. 2023;24:86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Coutant F, Bachet R, Pin JJ, Alonzo M, Miossec P. Monoclonal antibodies from B cells of patients with anti-MDA5 antibody-positive dermatomyositis directly stimulate interferon gamma production. J Autoimmun. 2022;130:102831. [DOI] [PubMed] [Google Scholar]
- 27.Gono T, Sato S, Kawaguchi Y, Kuwana M, Hanaoka M, Katsumata Y, Takagi K, Baba S, Okamoto Y, Ota Y, Yamanaka H. Anti-MDA5 antibody, ferritin and IL-18 are useful for the evaluation of response to treatment in interstitial lung disease with anti-MDA5 antibody-positive dermatomyositis. Rheumatology (Oxford). 2012;51:1563–70. [DOI] [PubMed] [Google Scholar]
- 28.Chen Z, Wang X, Ye S. Tofacitinib in amyopathic Dermatomyositis-Associated interstitial lung disease. N Engl J Med. 2019;381:291–3. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available upon reasonable request.