Poster 54 - Discovery of Novel Self-antigens in IgG4 Related Disease Using Computational Modeling and Human Proteome Screening

Saturday, October 26, 2024

10:00 AM - 12:00 PM EST

Location: Exhibit Hall E, F & G

Presenting Author(s)

Mahilan Guha (he/him/his)

Student
Brigham and Women's Hospital
Potomac, Maryland, United States

Co-Author(s)

Cory A. Perugino, DO

Assistant Professor of Medicine
Division of Rheumatology, Allergy and Immunology, Massachusetts General Hospital, Harvard Medical School, United States
SP

Shiv Pillai

Professor of Medicine
Ragon Institute, Massachusetts General Hospital, Harvard Medical School, United States
JS

John H. Stone

Professor and Chair of Medicine
Division of Rheumatology, Allergy, and Immunology, Massachusetts General Hospital, Harvard Medical School, United States

Last Author(s)

VM

Vinay S. Mahajan, MBBS PhD

Associate Bioengineer
Division of Engineering in Medicine, Brigham and Women’s Hospital, Harvard Medical School
Lexington, Massachusetts, United States

Introduction: IgG4-related disease (IgG4-RD) is an immune-mediated fibrotic disorder characterized clinically by tumor-like masses composed of tissue infiltrating lymphocytes, IgG4-expressing plasma cells, and fibrosis. The pathogenic importance of B cells has been established by the responsiveness of this condition to B cell-depleting therapies such as rituximab and inebilizumab. Plasmablasts, or short-lived antibody-secreting cells, are oligoclonally expanded in the blood and tissues of these patients and dominant clones recognize self-antigens such as galectin-3 and interleukin-1 receptor antagonists. However, no disease-specific auto-antibody response has been identified to date, nor has one been able to discriminate between clinical phenotypes of IgG4-RD. This study aimed to broadly screen the human proteome to identify novel self-antigens that could define or sub-cluster IgG4-RD.

Materials and

Methods: Plasma samples from 30 patients satisfying the 2019 ACR/EULAR Classification Criteria for IgG4-RD were used for these studies. Quantitative auto-antibody profiles were obtained using the HuProt microarray, which is a human protein microarray that displays over 21,000 unique human proteins. Plasma from 30 patients with systemic sclerosis (SSc), a distinct autoimmune disease with well-established auto-antibody specificities was used as a positive disease control cohort. Microarray data was analyzed using three different methods: Least Absolute Shrinkage and Selection Operator (LASSO) regression model, Generalized Linear Mixed Model (GLMM), and Significant Latent factor Interaction Discovery and Exploration (SLIDE). Analysis of differentially bound antigens was conducted using DESeq after applying each method, above, and 3 differentially bound antigen lists were compared as a means of cross-validation.

Results, Conclusions, and Discussions: First, to verify the efficacy of our computational approach, we validated the differential binding of several well-established auto-antibody specificities within the SSc cohort, including those associated with RNA polymerase III, centromeric proteins, and topoisomerase I (Figure 1). To identify self-antigen responses that characterize IgG4-RD, we created three models based on the three methods. All models performed with high prediction accuracy (area under the curve > 0.85). After applying the models to the data, we conducted the analysis of differentially bound antigens, identifying several potential unifying auto-antibody specificities among patients with IgG4-RD, including DDX3Y, BACH1, and LIMS1 (Figure 1). Unsurprisingly, this study also highlights the heterogeneous nature of IgG4-RD; even among the antigens of interest, reactivity levels varied greatly across the 30 IgG4-RD patients. Overall, this study identified multiple novel self-antigens that distinguish IgG4-RD and display the sharing of these antigenic targets across patients with IgG4-RD. In addition to identifying potential causal or disease-defining self-antigens in IgG4-RD, this study also illustrates the effectiveness of using generalized mixed models, LASSO, and SLIDE, as computational techniques that can be applied to extremely heterogeneous data.

Acknowledgements (Optional): Cory A. Perugino, D.O. – Division of Rheumatology, Allergy and Immunology, Massachusetts General Hospital, Harvard Medical School
Shiv Pillai, MBBS, PhD – Ragon Institute, Massachusetts General Hospital, Harvard Medical School
John H. Stone, MD, MPH – Division of Rheumatology, Allergy and Immunology, Massachusetts General Hospital, Harvard Medical School
Vinay S. Mahajan, MBBS, PhD – Division of Engineering in Medicine, Brigham and Women’s Hospital, Harvard Medical School

References (Optional):