Liver Involvement is Associated with Higher Risk of Rapidly Progressive Interstitial Lung Disease and Mortality in Anti-Melanoma Differentiation-Associated Gene 5 Antibody- Positive Dermatomyositis

Introduction

Idiopathic inflammatory myopathy (IIM) constitutes a large spectrum of autoimmune disorders, which can be classified into several groups: dermatomyositis (DM), anti-synthetase syndrome (ASyS), immune-mediated necrotizing myopathy, inclusion body myositis, polymyositis (PM) and overlap myositis¹.

A growing body of evidence indicated that myositis-specific antibodies have been associated with distinctive clinical phenotypes.² RNA helicase encoded by anti-melanoma differentiation-associated gene 5 (MDA5) was identified as the major autoantigen of anti-MDA5 antibody.³ MDA5 is a cytosolic protein, essential for antiviral host immune responses, which functions as a virus RNA sensor.⁴ Anti MDA5 antibody (anti-MDA5) positive DM is characterized by clinically amyopathic DM and rapidly progressive interstitial lung disease (RP-ILD), which is correlated with an aggressive course and poor prognosis. Skin ulceration and palmar papules are typical skin features of anti-MDA5 positive DM.^5,6 The prevalence of ILD reported in anti-MDA5 positive DM ranged from 50 to 100%.⁷

Anti-MDA5 positive DM has been frequently reported to be refractory to common combined immunosuppressive therapies, and the 6-month mortality of RP-ILD was reported to be as high as 50%.⁸ Consequently, exploring the characteristics and risk factors for RP-ILD and mortality is still urgently needed.

Liver injury and elevated liver enzymes in anti-MDA5 positive DM patients were reported in many studies.^4,9–11 Jiang et al reported that 36.4% patients with anti-MDA5 positive DM presented with liver involvement.⁹ However, studies focusing on the impact of liver involvement on RP-ILD and mortality are rare.

Analogous to anti-MDA5 positive DM, ASyS patients were also characterized by high frequency of ILD. About 50.67–100% ASyS patients have been shown to be complicated with ILD.¹² ASyS is characterized by the presence of autoantibodies directed against aminoacyl transfer RNA synthetase. Clinical features of ASyS consist of myositis, skin lesions, ILD, arthritis, Raynaud phenomenon, fever of unknown origin, and “mechanic’s hand”.^12,13

In the present study, we retrospectively analyzed 159 patients with anti-MDA5 positive DM or ASyS. We compared the clinical features and laboratory findings between anti-MDA5 positive DM patients and ASyS patients. Furthermore, we investigated the prevalence of liver involvement in patients with anti MDA5 positive DM, and explore whether liver involvement has an impact on RP-ILD and overall survival (OS).

Materials and Methods

Study Design

This was a retrospective single-center study. We analyzed the clinical data of 159 consecutive patients with anti-MDA5 positive DM or ASyS hospitalized in the Department of Rheumatology at the Second Affiliated Hospital of Xi’an Jiaotong University from October 2017 to June 2023. The patients were followed up until September 2023. Survival status was confirmed by hospital records or the follow-up calls. All patients were screened for a panel of myositis-specific antibodies and myositis-associated antibodies (anti-OJ, EJ, PL7, PL12, SRP, JO-1, MDA5, TIF1-γ, Mi-2α, Mi-2β, Ku, NXP2, PM-Scl75, PM-Scl100, SAE1 and Ro52 antibodies) using a commercial immunoblot assay. A diagnosis of anti-MDA5 positive DM was based on the Bohan and Peter criteria^14,15 or 239th European Neuromuscular Centre (ENMC) criteria.¹⁶ ASyS patients were clinically confirmed as having ASyS according to both the criteria of Connors et al and the criteria of Solomon et al and the presence of anti-tRNA synthetase antibodies.¹⁷

The presence of ILD was defined as the presence of interstitial changes on high-resolution computerized tomography (HRCT) scanning of the chest. RP-ILD was characterized by progressive dyspnea and hypoxemia, with a worsening of radiologic changes of interstitial lung inflammation within 3 months after the onset of respiratory symptoms.¹⁸ Clinical, demographic, and laboratory data were collected by reviewing the electronic medical record system. Fatty liver was defined by ultrasonographic detection of hepatic steatosis. Patients were screened for Hepatitis B virus (HBV) and hepatitis C virus (HCV) infection by testing HBV surface antigen and HCV antibody, respectively.

The normal range for alanine transaminase (ALT) is ≤30 U/l. We defined liver involvement as ALT level >60 U/l (more than 2 times the upper limit of the normal range).

The study was approved by the medical ethics committee of The Second Affiliated Hospital of Xi’an Jiaotong University (ID: 2023304) and performed in accordance with the Declaration of Helsinki. Because of its retrospective design, patient informed consent was waived by the above-mentioned ethics committee. The data was anonymized and maintained with confidentiality.

Statistical Analyses

Continuous variables were described as median (first and third quartiles). Comparison of continuous variables was performed by Mann–Whitney U-test. Categorical variables were presented as numbers and percentages. Comparison of categorical variables were assessed by the Chi-squared test or Fisher’s exact test. The “Surv_cutpoint” function of the R package “survminer” was used to determine the optimal cutoff points for continuous parameter, and then each continuous parameter was divided into a categorical variable. Kaplan–Meier method was performed to analyze the OS, and Log rank tests were used to compare survival curves. Univariate and multivariate analyses of the risk factors for mortality and RP-ILD were performed using Cox regression analysis. Two-tailed P-values of <0.05 were considered statistically significant. All statistical analyses were performed with R statistical package version 4.2.3.

Results

Comparison of Clinical Characteristics and Laboratory Features Between Patients with Anti-MDA5 Positive DM and ASyS

A total of 159 patients diagnosed with anti-MDA5 positive DM or ASyS were enrolled between 2017.10.1 and 2023.6.30. There were 93 patients diagnosed with anti-MDA5 positive DM and 66 patients diagnosed with ASyS, with a median follow-up time of 380 days and 598 days, respectively. Among the anti-MDA5 positive DM patients, all patients met the 239th ENMC criteria. Overall, 31, 30 and 19 patients met the possible DM criteria, the probable DM criteria and the definite DM criteria by the Bohan/Peter criteria, respectively.

Comparisons of the demographic features, clinical features and laboratory findings on admission between patients with anti-MDA5 positive DM and patients with ASyS are summarized in Table 1. Anti-MDA5 positive DM group was composed of 61 women and 32 men, with a median age of 52 years. Among them, 77.4% (n=72) patients were treatment-naive (no prior treatment with glucocorticoids (GCs) or immunosuppressants). As expected, anti-MDA5 positive DM group had a greater frequency of heliotrope rash (43.0% vs 10.6%, p<0.001), Gottron’s sign (63.4% vs 25.7%, p<0.001) and skin ulceration (27.9% vs 1.5%, p<0.001) than ASyS group.

Table 1 Comparison of Clinical Features Between Anti-MDA5 Positive DM and ASyS

Consistent with previous studies, a significant increase in serum ferritin level was also identified in anti-MDA5 positive DM patients compared to those in ASyS patients (835.5 (485.8, 1297.3) vs 165.0 (70.3, 429.9) ng/mL, p<0.001). As expected, significantly lower serum creatine kinase (CK) level in anti-MDA5 positive DM patients was documented than in patients with ASyS (83.0 (47.0, 150.0) vs 152.5(65.0, 475.3) ng/mL, p=0.002). White blood cell (WBC) counts were also lower in anti-MDA5 positive DM patients when compared with ASyS patients (4.2(3.3,5.7) vs 7.5(5.7, 9.3) ×10^9/L, p<0.001).

Since liver injury was frequently reported in anti-MDA5 positive DM patients in previous study,¹⁰ we comprehensively evaluated liver enzymes in the current study. As shown in Table 1, serum levels of AST (60.0(39.3, 94.8) vs 28.0 (21.0, 55.0) U/l, p<0.001), ALT (41.5(26.0,86.8) vs 26.0(14.0, 42.0) U/l, p<0.001), γGT (48.0(28.0, 94.0) vs 19.0(12.0, 31.0) U/l, p<0.001) and ALP(81.0(65.0, 104.0) vs 67.0 (54.0, 85.0) U/l, p<0.001) were all remarkably higher in patients with anti-MDA5 positive DM patients than those in patients with ASyS, indicating common liver involvement in anti-MDA5 positive DM patients.

The overall mortality of patients with anti-MDA5 positive DM was 29(31.2%), significantly higher than that of patients with ASyS [1(1.5%)]. Survival analysis showed that OS of anti-MDA5 positive DM group was remarkably lower than that of ASyS group (p<0.001, Supplementary Figure 1).

Comparison of Characteristics Between Anti-MDA5 Positive DM Patients with and without Liver Involvement

We defined liver involvement as ALT level >60 U/l (more than 2 times the upper limit of the normal range). After excluding 3 patients whose initial serum ALT was not available, we compared the clinical features of anti-MDA5 positive DM patients with and without liver involvement (Table 2). There were 31 patients in liver involvement group and 59 patients in non-liver involvement group.

Table 2 Comparison of Characteristics Between Anti MDA5 Positive DM Patients with and without Liver Involvement

We further analyzed R factor to determine the pattern of liver injury in each patient.¹⁹ Among patients with liver involvement, 13.3% had a cholestatic injury pattern, 56.7% had a mixed pattern, 30.0% had a hepatocellular pattern. Serum ferritin level was strikingly higher in patients with liver involvement than in patients without liver involvement (1281.0 (748.0, 1700.5) vs 642.0 (326.9, 1016.0) ng/mL, p=0.005). The mortality of the liver involvement group was significantly higher than that of the non-liver involvement group (48.9% vs 18.6%, p = 0.007). Moreover, the incidence of RP-ILD was higher in patients with liver involvement when compared with patients who did not have liver involvement (58.1% vs 22.0%, p=0.001). Thirty-six patents performed electromyogram (EMG) on admission. MRI data were available in 24 patients. There was no difference between liver involvement group and non-liver involvement group regarding the presence of myogenic pattern at EMG (62.5% vs 46.2%, p = 0.689) and magnetic resonance imaging (MRI) testing indicative of myositis (87.5% vs 75.0%, p = 0.859). No patient was complicated with HBV infection in anti-MDA5 positive DM. Only one patient was complicated with chronic HCV infection. No differences were observed between patients with liver involvement and patients without liver involvement in terms of incidences of HBV infection, HCV infection and fatty liver (Table 2).

Comparison of Baseline Characteristics of Anti MDA5 Positive DM Patients Between Survivors and Non‑survivors

The baseline characteristics between deceased and surviving patients are shown in Table 3. Ferritin level (1385.0.5(925,0, 2189.5) vs 627.5(359.6, 1007.0) ng/mL, p<0.001) were higher in the non-survivor group than in the survivor group. The non-survivors also had higher levels of CK (131.0 (69.8, 279.0) vs 72.0(45.5, 127.0) U/l, p=0.012) and lactate dehydrogenase (LDH) (431.0(316.0, 509.5) vs 307.0 (270.5, 418.5) U/l, p=0.003).

Table 3 Comparison of Baseline Characteristics of Anti MDA5 Positive DM Patients Between Survivors and Non‑survivors

To evaluate liver involvement in anti-MDA5 positive DM patients, we compared liver injury markers between survivors and non-survivors. AST level (77.0 (62.1, 130.5) vs 48.5 (31.8, 83.3) U/l, p <0.001), ALT level (97.0(43.3, 145.8) vs 34.0 (18.0, 57.8) U/l, p<0.001), γGT level (89.0 (41.0, 239.0) vs 38.5(25.0, 88.3) U/l, p=0.007) and direct bilirubin (DBIL) level (3.3 (2.7, 4.0) vs 2.6(2, 3.3) μmol/l, p=0.004) were all increased in non-survivors (Table 3). Also, the level of plasma albumin (ALB) was significantly lower in non-survivors (29.83 (27.1, 32.9) vs 34.7 (30.0, 37.7) g/l, p=0.006).

Correlation Between Liver Enzymes and Serum Ferritin of Anti MDA5 Positive DM Patients

We compared the correlation between liver injury markers and serum ferritin. As shown in Figure 1, There was a significant association between ferritin level and level of ALT (p=0.002), AST (p<0.001), γGT (p<0.001), ALP (p=0.005), DBIL (p=0.001) and LDH (p<0.001), respectively.

Figure 1 Correlations between serum ferritin and liver enzymes. (A) Serum ferritin and ALT; (B) Serum ferritin and AST; (C) Serum ferritin and γGT; (D) Serum ferritin and ALP; (E) Serum ferritin and DBIL; (F) Serum ferritin and LDH. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; γGT, γ-glutamyl transferase; ALP, alkaline phosphatase; DBIL, direct bilirubin; LDH, lactate dehydrogenase.

Elevated Liver Enzyme Levels Predict RP-ILD and Mortality of Anti MDA5 Positive DM Patients

The standard treatment regimen for anti-MDA5 positive DM has not been established. In our clinical practice, all anti-MDA5 positive DM patients were treated with GCs. Immunosuppressants were simultaneously administered, including hydroxychloroquine, tacrolimus, intravenous cyclophosphamide, or Janus kinase inhibitor.

The OS rate was remarkably higher in anti-MDA5 positive DM patients treated with tacrolimus than in those treated without tacrolimus (p<0.001, Supplementary Figure 2A). We also analyzed the efficacy of tofacitinib and found that patients treated with tofacitinib had a substantially better prognosis than those not treated with it (p<0.001, Supplementary Figure 2B).

To further validate whether the aforementioned differentially expressed markers are prognostic risk factors, the “surv_cutpoint” function of the R package “survminer” was used to evaluate the optimal cut-off values of ALT, AST, γGT, ALP, LDH, ALB, ferritin and DBIL at diagnosis for the risk of mortality. Patients were stratified into high-level and low-level groups according to the optimal cut-off value of the markers above. As shown Supplementary Figure 3A-F, Kaplan–Meier survival analysis showed that serum ALT >103.0U/l (p<0.001), AST >49.0U/l (p<0.001), γGT >82.0U/l (p<0.001), ALP >133.0U/l (p<0.001), DBIL >2.80 μmol/l (p=0.002) and ALB <35.7 g/l (p<0.001) predicted poor OS. Moreover, patients with ferritin >1030.0 ng/mL (p<0.001) and LDH >474.0U/l (p<0.001) had lower OS (Supplementary Figure 3G and H).

Due to the relatively small sample size of deceased patient (n=29), 2 multivariable Cox proportional hazards models (model 1: adjusting for age and ferritin levels; model 2: adjusting for age, ferritin levels and treatment with/without tacrolimus) were performed to examine the association of ALT levels and mortality. Multivariate Cox regression analysis revealed that both high ALT level (Model 1: HR 3.9, 95% CI 1.6, 9.5, p<0.001; Model 2: HR 3.7, 95% CI 1.4, 9.5, p=0.008) and high serum ferritin (Model 1: HR 6.0, 95% CI 2.3, 16.2, p<0.001; Model 2: HR 3.3, 95% CI 1.2, 9.5, p=0.024) were independent risk factors for mortality of the anti-MDA5 positive DM patients (Table 4).

Table 4 Initial Parameters Associated with Death Using Cox Regression Model

In the current study, the incidence of RP-ILD was notably higher in anti MDA5 positive DM patients with liver involvement than in patients without liver involvement. We investigated whether initial liver injury markers could predict RP-ILD in anti MDA5 positive DM patients. Univariate regression analysis revealed that ALT >103.0 U/l (HR 6.8, CI 2.9, 16, p<0.001), AST >49.0 U/l (HR 3.0, CI 1.3–6.9, p=0.008), γGT >82.0U/l (HR 2.1, CI 1–4.5, p=0.045), ALP >133.0 U/l (HR 4, CI 1.8–9, p<0.001), LDH >474.0U/l (HR 2.2, CI 0.9–4.9, p=0.007), ALB <35.7g/l (HR 6.4, CI 2.2–19, p<0.001) and ferritin >1030.0 ng/mL (HR 3.8, CI 1.7–8.5, p=0.001) predicted RP-ILD (Table 5). Levels of serum ferritin, ALT and ALB were then included for multivariate regression analysis. As expected, multivariate COX regression analysis for risk factors of RP-ILD also showed a marked impact of elevated ALT level on RP-ILD (HR 5.9, 95% CI 2.2, 15.9, p<0.001).

Table 5 Initial Parameters Associated with RP-ILD Using Cox Regression Model

Discussion

In the present study, we included 159 patients diagnosed with anti-MDA5 positive DM or ASyS. We comprehensively compared the clinical characteristics and liver injury markers between the two groups. We found that the levels of ALT, AST, γGT and ALP in anti-MDA5 positive DM patients were all notably higher than those in ASyS group. Liver involvement was more common in anti-MDA5 positive DM patients than ASyS patients (34.4% vs 12.3%). Consistent with previous studies, the ferritin level was strikingly elevated and significantly higher in anti-MDA5 positive DM group than that in ASyS group. Moreover, in patients with anti-MDA5 positive DM, levels of ALT and AST were significantly higher in non-survivors compared with survivors. Survival analysis identified high levels of ALT, AST, γGT and ALP as poor prognostic factors in anti-MDA5 positive DM group. Additionally, increased ALT level was also defined as an independent risk factor for RP-ILD in anti-MDA5 positive DM patients.

Liver involvement was previously reported as an extra-muscular manifestation in anti-MDA5 positive DM.^10,11 Surprisingly, in Nagashima et al’s study, patients with liver dysfunction were all positive for the anti- MDA5 antibody. Liver damage may be associated with autoimmune-mediated inflammation or other causes, including viral hepatitis, fatty liver and drug-induced hepatitis. In the current anti-MDA5 positive DM cohort, no patients were complicated with HBV infection. Only one patient was complicated with chronic HCV infection. Twenty-five percent patients were complicated with fatty liver on admission; however, there was a similar percentage of patients complicated with fatty liver in anti-MDA5 positive DM patients with and without liver involvement (32.1% vs 21.4%). Additionally, as high as 77.4% patients were treatment-naive. Therefore, we can largely exclude the effect of viral hepatitis, fatty liver and medication on the elevation of liver injury markers. ALT and AST levels can be elevated in patients with muscle involvement. However, in our study, there was no difference between liver involvement group and non-liver involvement group regarding the presence of myogenic pattern at EMG, MRI testing indicative of myositis and serum CK level. Therefore, the effect of muscle involvement on ALT and AST levels can be largely ruled out. We proposed that the elevation of the liver injury markers was mainly due to anti-MDA5 positive DM itself.

The mechanism of liver injury in anti-MDA5 positive DM patients is still unclear. In Nagashima et al’s study, liver biopsies from anti-MDA5 positive DM patients showed hepatocyte ballooning, pigmented macrophages, and glycogenated nuclei and mild steatosis.¹⁰ Our study revealed significant positive correlations between serum liver enzymes and ferritin. High serum ferritin level, a key marker of systemic macrophage activation, has been reported as a hallmark inflammatory marker in anti-MDA5 positive DM patients. Accumulating evidence suggests that hyperferritinemia may play a pathogenetic role in increasing the inflammatory process.²⁰ Thus, we hypothesized that macrophage activation may be essential in the pathogenesis of liver injury in anti-MDA5 positive DM patients. In the autopsy of an anti-MDA5 positive DM patient with RP-ILD and hyperferritinemia, systemic ferritin producing macrophages were detected in the alveoli, bone marrow, liver, and spleen.²¹ Moreover, soluble CD206, a marker of activated macrophages, was shown to significantly increase in fatal cases of anti-MDA5 positive DM-ILD patients. In that study, dense accumulation of CD206+ macrophages into the airspace of a fatal DM-ILD case was also demonstrated.²² Levels of other macrophage activation markers, including serum chitotriosidase, neopterin and serum-soluble CD163, were significantly elevated in anti-MDA5 positive DM patients.^23–25 Peng et al showed that serum levels of neopterin were significantly associated with RP-ILD and poor prognosis in DM patients.²⁴ Immunohistochemical analysis of the lung specimens showed patients with RP-ILD had significantly higher amounts of CD163-positive macrophages at the alveolar spaces when compared with the patients with chronic ILD.²⁶

In a recent study by Zhao et al, high HScore was identified as a risk factor for poor survival among anti-MDA5-positive DM-ILD patients, suggesting the involvement of macrophage activation in the pathogenesis.²⁷ In Zhao et al's study, serum AST concentration was also notably higher in survivors than in non-survivors.

MDA5 recognizes intracellular viral nucleic acids and triggers type I interferon (IFN) production to suppress the replication of viruses.²⁸ Integrated miRNA-mRNA association analysis using circulating monocytes from patients with anti-MDA5 positive DM-ILD patients has shown an anti-viral inflammatory response.²⁹ Great similarities of clinical and pathogenic features have been reported between Coronavirus disease 2019 (COVID-19) and anti-MDA5 positive DM. They presented with similar radiologic appearance on chest CT and hyperferritinemia.^4,30 Besides, anti-MDA5 antibodies have been identified in the serum of 48.2% Chinese patients with COVID-19.³¹ Cytokines and chemokines have been evaluated in anti-MDA5 positive DM patients by many studies. Increased levels of Interferon-a (INF-a), INF-β, IFNλ3, macrophagecolony stimulating factor (M-CSF), interleukin-2 (IL-2), IL-6, IL-8, IL-10, IL15, IL-18, C-C motif ligand 2 (CCL2), C-X-C motif ligand 10 (CXCL10), CXCL11 and tumor necrosis factor α were reported recently, indicating that anti-MDA5 positive DM is characterized by “cytokine storm” conditions.^32–36 Taken together, we may hypothesize that viral infection may induce the activation of macrophage and further triggers the “cytokine storm”, leading to injuries of different target organs in anti-MDA5 positive DM patients. In our study, the occurrence of RP-ILD and mortality was significantly higher in liver involvement than in non-liver involvement group. Multivariate COX regression analysis confirmed that higher ALT level was independently associated with higher risk of RP-ILD and mortality. Thus, we speculate that elevated liver enzymes may indicate uncontrolled hyperinflammation and macrophage activation, leading to increased risk of RP-ILD and mortality.

So far, the optimal treatment therapy for anti-MDA5 positive DM has not been established.⁸ It is generally agreed that GCs should remain the first-line drug for initial treatment. High-dose GCs in combination with immunosuppressive agents have been commonly used for the treatment of anti-MDA5 positive DM-ILD.^37,38 A recent prospective study supported the benefit of triple combination therapy (GCs, tacrolimus and cyclophosphamide) in the management of anti-MDA5 positive DM-ILD patients.³⁹ Tacrolimus has been increasingly used in the treatment of DM-ILD and may lead to a better prognosis.⁴⁰ A retrospective study from Japan demonstrated that the addition of tacrolimus significantly improves the event-free and disease-free survival of patients with PM-/DM-related ILD.⁴¹ In a recent prospective study, the 52-week survival rate in the prospective group initially treated with tacrolimus and GCs was 88.0%, suggesting that combination treatment with tacrolimus and GCs may improve the short-term mortality of PM/DM-ILD patients.⁴² In our cohort, the combination of tacrolimus with conventional treatment (GCs alone or GCs plus other immunosuppressants except tacrolimus) significantly improved OS of patients with anti-MDA5 positive DM.

Tofacitinib has also been used as a promising immunosuppressant in treating patients with anti-MDA5 positive DM.⁴³ In a single-center, open-label clinical study, early-stage anti-MDA5 positive DM-ILD patients treated with tofacitinib exhibited significantly higher OS compared with historical controls.⁴⁴ As for the treatment of refractory anti-MDA5 positive DM-ILD, Kurasawa et al reported patients who had poor prognostic factors and who were predicted to have a poor prognosis and failed to respond to triple therapy received a combination therapy with tofacitinib. The survival rate of the patients treated with the combination therapy was significantly better than that of the historical controls before tofacitinib.⁴⁵ Consistent with previous studies, the efficacy of tofacitinib was also proved in our study.

It should be noted that our study had several limitations. First, this study was a single-center, retrospective study. Biases (selection bias and information bias) were inevitable with such studies. Numerous baseline data on admission were missing due to the retrospective design. Second, the relatively small sample size forbad comprehensive multivariate regression analysis to remove all the confounding effects. Thirdly, cytokine data, liver biopsies and disease activity measurements were not performed in the present study. Prospective randomized controlled trial studies are needed to validate our findings.

In conclusion, the current study indicated that liver involvement is common in Chinese patients with anti-MDA5 positive DM. Elevated serum ALT level is an independent risk factor for RP-ILD and mortality of patients with anti-MDA5 positive DM.

Conclusion

The current study indicated that liver involvement is common in Chinese patients with anti-MDA5 positive DM. Elevated serum ALT level is an independent risk factor for RP-ILD and mortality of patients with anti-MDA5 positive DM.