Research redefines rheumatoid arthritis: Over 314,000 cells analyzed for precise treatment strategies

By Dr. Liji Thomas, MDReviewed by Danielle Ellis, B.Sc.Nov 10 2023

Rheumatoid arthritis (RA) is among the earliest autoimmune joint diseases to have been identified and remains incurable. Despite the discovery of several disease-modifying treatments, the response to each remains unpredictable. This indicates a difference in the pathophysiology of RA between patients.

A new paper recently published in Nature, reported the examination of synovial tissue from the joints of nearly 80 people with RA, in combination with RNA sequencing and surface protein analyses. This allowed the researchers to build an atlas of RA synovial changes from over 314,000 single cells. This could help develop targeted therapies that recognize the diversity of RA disease processes.

Background

RA affects about 1 in a hundred people worldwide. Its main characteristic is the painful swelling of synovial joints that eventually culminates in joint damage and disability. Recognition of the immunologic origin of RA has led to the deployment of therapies targeting inflammatory cytokines and pathways, including the tumor necrosis factor (TNF), IL-6, stimulation of T and B cells together, and the pro-inflammatory JAK-STAT transcriptional regulatory pathway.

Genetic differences have been identified, as well as varying clinical characteristics, but neither fully predict or explain why treatment response varies between patients or help identify therapeutic targets. The need for a more detailed picture of RA synovial disease activity motivated the current study.  

Multiple effector cells participate in RA activity at the synovial level. Prior research suggests that the synovial cellular profile could predict response to treatment. Moreover, the presence of common cell state compositions could extend the utility of this study to other autoimmune or inflammatory conditions.

What does the study show?

The study was based on 82 synovial tissue samples taken from patients with a spectrum of RA activity from moderate to high. This is measured by the CDAI (clinical disease activity index), which was ten or higher for all participants. The samples came from those who had not yet initiated treatment, some with poor response to methotrexate (which arrests inflammatory cell proliferation), those who responded poorly to anti-TNF agents (to arrest pro-inflammatory signaling), and some who had osteoarthritis.

The scientists were able to classify the RA synovium into six groups by the types of cells selectively enriched in each. Each group is accordingly termed a cell type abundance phenotype (CTAP) and is defined by specific cell states.  

While some samples showed very low levels of lymphocytes, others were abundant in T and B cells, indicating marked synovial differences. Each cell state reflects different disease stages and types, as well as varying cytokine profiles, and the risk genes were expressed differentially between groups.

The investigators created an atlas of RA synovial cell states comprising 77 cell states, including 24 T cell clusters, 9 B cell clusters, 14 clusters of natural killer (NK) cells, and 15 myeloid clusters. There were also ten stromal cells and five endothelial clusters. These corroborated RA-associated cell states identified in a previous study from over 5,000 synovial cells.

For instance, the CTAP-TB was enriched in T_PH and T_FH cells, perhaps because these promote B cell differentiation into plasmablasts and ABC cells, unlike non-T_FH/T_PH memory CD4+ T cells that do only the latter. Both T_FH and T_PH cells are enriched in the synovial tissue of all CTAPs, but extra-follicular activation pathways also seem to be present in CTAP-TB.

Conversely, the CTAP-TF includes primarily cytotoxic along with naïve CD4 and CD8 T cells, with selective NK cells that may share their transcription profile promoted by the tissue microenvironment. Fibroblast subsets were differentially enriched in this CTAP vs CTAP-M. The latter also showed myeloid cell enrichment, perhaps because inflammatory monocytes were being recruited to transform into macrophages as a result of exposure to the specific cell types and soluble factors present in each CTAP.

These cell neighborhoods did not show consistent associations with RA aggregate scores from histology, which are based on the extent and type of inflammatory cell infiltration. This is probably because the former are so diverse. However, the CTAPs contribute a fifth each of the variance of histologic density and aggregate scores and are associated with inflammation scores.

Interestingly, the CTAPs showed a close relationship with clinical parameters like the commonly used cyclic citrullinated peptide (CCP) autoantibodies, reflecting increased lymphocyte infiltration in CCP-positive synovial tissue. CTAP-M was associated with CCP-negative synovial tissue. There was no distinct association with the strongest genetic risk predictor, HLA-DRB1.

The CTAPs did show distinct cytokine profiles. For instance, the T cell neighborhood of CTAP-TB expressed the TFH/TPH marker gene CXCL13 as expected, while for CTAP-TF, the T and NK cell neighborhood was associated with the expression of the genes IFNG and TNF.

As expected, there was little correlation between disease activity and either the CTAP or treatment response. This lends support to the theory that inflammatory phenotypes in different types of RA are reflected in the CTAPs and not clinical disease activity as shown by CDAI and other clinical scores.

CTAPs do change over time, however, mostly to CTAP-F, following anti-inflammatory therapies like rituximab and the anti-IL-6 agent tocilizumab. CTAP-F is a predictor of poor treatment response.

What are the implications?

“The CTAP paradigm has the potential to serve as a powerful prototype to classify other types of tissue inflammation.” The subtypes of enriched inflammatory cells in different CTAPs also uncover new research questions as to how these interact to produce a range of inflammatory phenotypes in such illnesses.

“CTAPs are dynamic and can predict treatment response, highlighting the clinical utility of classifying rheumatoid arthritis synovial phenotypes.” It was possible to predict the CTAP using RNA sequencing by various methods. This offers potential therapeutic targets for the future.

Meanwhile, the spectrum of inflammatory changes in RA explains why treatment responses vary so markedly among patients treated with anti-TNF agents. This may imply that specific therapies targeting the cells and pathways enriched in each CTAP may induce better responses, as well as promote drug development and precision medicine.