Background: Anti-melanoma differentiation-associated gene 5 (MDA5) antibody-positive dermatomyositis is a distinct type of idiopathic inflammatory myopathy, typically characterized by rapidly progressive interstitial lung disease (RP-ILD) responding poorly to treatment and conveying a high risk of mortality. However, the disease’s presentation and prognosis are very heterogeneous when comparing individual patients.

Objectives: We compared the phenotypic clusters of anti-MDA5 antibody-positive dermatomyositis that have been described in multiple previous studies, which attempted to use such clusters for characterizing patients’ different presentations and predicting patients’ prognosis at the time of diagnosis.

Methods: We conducted a literature search for studies that described phenotypic clusters of anti-MDA5 antibody-positive dermatomyositis. We critically summarized their methods and results. We compared the characteristics and prognosis of these clusters, both within each study and between the studies.

Results: We found four studies that used different methods of unsupervised clustering analysis to describe phenotypic clusters of anti-MDA5 antibody-positive dermatomyositis at the time of diagnosis. All studies defined three clusters. All studies defined one cluster that included patients with predominantly RP-ILD and high risk of mortality. Patients in this cluster were predominantly older, female, and Asian. The remaining two clusters were defined variably in these studies, including patients with predominantly myopathy, rashes, arthritis, or vasculopathy, respectively, and having a variable prognosis. The clusters’ characteristics differed between studies. In the cluster including patients with predominantly RP-ILD, 67% to 93% of patients in the different studies had RP-ILD. In the remaining two clusters, 7% to 26% of patients in the different studies had RP-ILD. The remaining clusters included widely varying numbers of patients with myopathy or arthritis in the different studies. The studies proposed very different decision trees based on clinical characteristics to predict a patient’s cluster at the time of diagnosis. The decision trees had a predictive accuracy of 70% to 83%. It was unclear when RP-ILD was diagnosed relative to the time of diagnosis.

Conclusion: Anti-MDA5 antibody-positive dermatomyositis can be divided in phenotypic clusters, of which one includes patients with predominantly RP-ILD with high risk of mortality. The remaining clusters are heterogeneous. All clusters vary between studies, possibly due to differences in the source population and/or methodology. The variability within and between clusters, the differences between the studies, and the weak predictive accuracy of the decision trees indicate that future studies are needed before the clusters and decision trees can be used in clinical practice.

REFERENCES: [1] Allenbach Y, et al. Neurology 2020;95:e70.

[2] Yang Q, et al. Clin. Exp. Rheumatol. 2022;40:304.

[3] Jin Q, et al. J. Intern. Med. 2023;293:494.

[4] Xu L, et al. Arthritis Rheumatol. 2023;75:609.

Acknowledgements: NIL.

Disclosure of Interests: Jacob J.E. Koopman: None declared, May Y. Choi AstraZeneca, Werfen, Mitogendx, Celltrion, Organon, Mallinckrodt Pharmaceuticals.