Toward new targeted treatments for rheumatoid arthritis

Published today in the journal Nature, their findings reflect the work of dozens of researchers working together as members of the Accelerating Medicines Partnership: Rheumatoid Arthritis and Systemic Lupus Erythematosus (AMP: RA/SLE) Network, including Michael Holers, M.D., professor of medicine and the site principal investigator at the CU School of Medicine.

The AMP: RA/SLE Network collected inflamed tissue from 70 patients with RA from across the country and the United Kingdom. Jonsson supervised the team of scientists who processed these samples for analysis, and Zhang led the computation analysis of the data. These efforts yielded a cell atlas encompassing more than 300,000 cells from synovial tissue. Further analysis revealed that there are six different subgroups of RA based on their cellular makeup.

“We hope the data will help us discover new treatment targets,” says Jonsson, assistant professor of rheumatology. “We wanted to make it public so that researchers across the country and across the world can continue working on new treatment ideas for rheumatoid arthritis going forward.”

No more guess-and-check

Jonsson, who is a practicing rheumatologist as well as a researcher, knows that RA patients respond differently to different treatments. Until now, she says, rheumatologists used a “guess and check” method to find a treatment that works for an individual patient.

With the new data and powerful computational classification methods developed by Zhang and the computational analysis team, the researchers were able to quantitatively classify RA types into what they call “cell-type abundance phenotypes,” or CTAPs. Developed methods, together with the new cell atlas, can start to identify which patients will respond to which treatments.

“Even when you classify rheumatoid arthritis inflammation using these simple markers — T cell markers, B cells, macrophages and other myeloid cells, fibroblasts, endothelial cells — what we found is that each of those categories is associated with very specific kinds of pathogenic cell types we’ve already discovered,” Jonsson says. “Previous rheumatoid arthritis research found that T cell populations called peripheral helper T cells are relevant in rheumatoid arthritis, as are B cells called antibody-producing B cells, and other specific cell types. What we found is that they’re usually not found all together.

“For example, the peripheral helper cells are found with the B cells in only one category of RA, and the pathogenic macrophage populations tend to exist in a different category. Because of this, we can start asking questions about how these specific partners work together.”

Interdisciplinary research and cross-institution effort are key

“I view this as interdisciplinary and big data-driven research. Many new findings were generated through our novel computational methods and systems immunology approaches,” says Zhang, assistant professor of rheumatology and faculty member in the Department of Biomedical Informatics. “We used the cutting edge of single-cell multimodal technology to develop this reproducible classification schema. It’s a big step toward precision medicine for rheumatologic diseases. With the robust computational AI methods, we are able to make sense of integrating large single-cell omics, imaging, and clinical data to stratify patient heterogeneity in a generalized manner.”

The CTAP research is part of a multicenter consortium effort that began in 2018. It is funded by the Accelerating Medicines Partnership Rheumatoid Arthritis and Systemic Lupus Erythematosus Network, an initiative coordinated by the National Institutes of Health and the Foundation for the National Institutes of Health. The project relies on a nationwide network of research teams that work collaboratively to deepen understanding of autoimmune diseases through a focus on RA and systemic lupus erythematosus.

“The research on inflammation subgroups might also be used to study other autoimmune diseases or immune responses to cancer or infection. From there, it might be leveraged for increased understanding across multiple different kinds of diseases,” Jonsson says.

“Building disease-driven computational AI methods will be the next step to generate testable hypotheses across multiple immune-mediated diseases,” Zhang adds.

Fertile ground for research

For Jonsson and Zhang, the Nature publication is the culmination of years of work that began when both worked at Brigham and Women’s Hospital, the teaching hospital of Harvard University. They brought the project with them when they came to the CU School of Medicine, and they hope to build new collaborations with faculty members across the CU Anschutz Medical Campus.

“One of the things that drew us to come here to start our research groups is that human translational research at the University of Colorado School of Medicine is so strong,” Jonsson says.

Similarly, “I felt like this would be the most fertile ground for us to continue our computational-experimental productive pattern for translational medicine,” Zhang says.