Two Distinct Immune Cell Signatures Predict the Clinical Outcomes in Patients With Amyopathic Dermatomyositis With Interstitial Lung Disease

• Published in: Arthritis & rheumatology (Hoboken, N.J.) N.J.), 2022-11, Vol.74 (11), p.1822-1832
• Main Authors: Ye, Yan, Zhang, Xueliang, Li, Teng, Ma, Jiaqiang, Wang, Ran, Wu, Chunmei, Wang, Runci, Bao, Chunde, Ye, Shuang, Shen, Nan, Guo, Qiang, Fu, Qiong, Zhang, Xiaoming
• Format: Article
• Language: English
• Subjects: Antibodies; Autoimmune diseases; Biomarkers; CD45RA antigen; CD8 antigen; Cell survival; Clinical outcomes; Cluster analysis; Clustering; Dermatomyositis; Flow cytometry; Histocompatibility antigen HLA; Immune system; Learning algorithms; Lung diseases; Lymphocytes; Lymphocytes T; Machine learning; Medical prognosis; Melanoma; Mortality; Natural killer cells; Pathogenesis; Patients; Peripheral blood; Phenotypes; Rank tests; Risk analysis; Risk factors; Subgroups; Survival; Survival analysis
• ISSN: 2326-5191; 2326-5205
• DOI: 10.1002/art.42264

Objective Amyopathic dermatomyositis (ADM) is a heterogeneous and life‐threatening autoimmune disease with a high mortality rate. In particular, anti–melanoma differentiation–associated protein 5 antibody–positive patients are at a high risk of developing rapidly progressive interstitial lung disease (RPILD). This study was undertaken to identify immunologic signatures among patients who have ADM with ILD (ADM‐ILD) and to discover the biomarkers predicting prognosis. Methods The landscape of 42 immune cell phenotypes in the peripheral blood of 82 ADM‐ILD patients and 82 age‐ and sex‐matched healthy donors was assessed by multicolor flow cytometry. Patients were stratified using an unsupervised machine learning method (hierarchical clustering analysis) by immune cell subsets. Multiple Wilcoxon's signed rank tests and supervised machine learning methods were performed to identify important immune cell subsets. Kaplan‐Meier survival analysis with log rank tests was used to create survival curves. Results We identified 2 distinct clusters correlating with different disease activities and clinical outcomes in ADM‐ILD. Cluster 1 was enriched in the activated CD45RA+HLA–DR+CD8+ T cells with decreased CD56dim natural killer cell proportions and showed a higher prevalence of RPILD and higher mortality. In contrast, the other subgroup, cluster 2 (the nonactivated T cell–dominant cluster), displayed favorable clinical outcomes with high survival rates. Our data also revealed that immunophenotype was an independent risk factor associated with 1‐year survival. Conclusion Peripheral immunologic features may have the potential to stratify patients with ADM‐ILD according to different disease severity and clinical outcomes, which may have implications for outcome prediction, pathogenesis study, and therapy selection.