Advancing rheumatoid arthritis synovial biopsy analysis: one B cell at a time

While biologic therapies have dramatically improved the management of rheumatoid arthritis (RA), not all patients respond well to treatment and the field continues to await individualized medicine approaches that promise to determine effective medications based on a patient’s distinct disease features. The cellular and molecular characterization of an individual’s synovium holds promise to understand pathology, predict outcomes and tailor treatment decisions, akin to the histologic and genetic assessment of tumor tissue that guides cancer care. Recently, several large consortia^1–5 have set out to examine whether histologic and functional genomic assays of synovial tissue samples can provide important insights into RA disease pathology and predict clinical outcomes. These large-scale efforts have also taken on the task of harmonizing methods for assessment of synovial tissues to facilitate standardized clinical use.

B cells are of particular interest in efforts looking to connect synovial tissue features with clinical measures. The notion that B cells impact RA pathogenesis is supported by the presence of B cells in RA synovium, the fact that autoantibodies are one of the few objective molecular markers of RA disease, and that B cell depleting therapies have been shown to be an effective treatment. However, there is considerable patient to patient variability in the extent of synovial B cell infiltration and while some groups have found B cell infiltrate scores useful for predicting RA treatment response, others have not found associations with clinical assessments of RA disease activity. One possible source of this discrepancy is the lack of standardized B cell scoring systems for synovial tissue. In this issue of Arthritis and Rheumatology, Rivellese et al (CITE) present steps to validate a semiquantitative CD20 immunohistochemistry-based B cell score. They applied their approach to synovial biopsies from two sizable cohorts; the Pathobiology of Early Arthritis Cohort (PEAC), which consisted of 165 consecutive treatment naïve patients with less than one year of disease, and a subset of TNF inhibitor inadequate responders (TNFiR) from the Response- Resistance to Rituximab versus Tocilizumab in Rheumatoid Arthritis (R4RA) trial, which consisted of 164 patients that had a mean disease duration of 12.8 years. All patient synovial biopsies were assessed by immunohistochemical stains of CD20 (B cell marker), CD68 (macrophage/fibroblast marker) and CD138 (plasma cell marker), while a substantial number were also analyzed by RNA sequencing. From the transcriptomic analysis, they generated B cell modules using subsets of 91 and 127 patients from the two respective cohorts using Functional Annotation of the Mouse/Mammalian Genome (FANTOM) data, an online resource of cDNA data established in part by the RIKEN institute in Japan. Their analysis found that the pathologist-interpreted semiquantitative CD20 immunohistochemistry-based B cell scores correlated with both a digital image analysis of CD20 immunohistochemistry and B cell gene expression module scores. With validation of the semiquantitative CD20 immunohistochemistry-based B cell scores using these two orthogonal approaches, they applied the B cell scores to the larger cohort of 329 patients and tested for clinical associations.

There were two major observations from this analysis. First, they found B cell rich synovitis was more common in the TNFiR cohort compared to the treatment naïve cohort. The TNFiR cohort was also enriched for anti-citrullinated peptide antibody (ACPA) positive patients, raising the possibility that increased B cell rich synovitis could be attributed to increased B cell rich synovitis in seropositive patients. It is also possible that B cell rich synovitis is more prevalent in patients with longer disease duration. Future studies evaluating the frequencies of B cell rich synovitis across a spectrum of disease durations and treatment exposures are needed to determine whether time or treatments impact synovial B cell accumulation.

A second notable finding related to B cell scores associating with numerous disease features such as Krenn synovitis scores (a global histology severity score for RA that includes assessments of synovial lining hyperplasia, infiltrates and stromal cell activation), DAS28 scores, swollen joint counts, ESR and CRP levels, and ACPA and RF positivity in the PEAC treatment-naïve cohort but not the TNFiR cohort (where B cell scores associated with only Krenn score and CRP). The finding that B cell scores associated with disease activity scores in the PEAC cohort confirms previous reports that demonstrated B cell rich lymphoid aggregates are associated with disease activity and which also predicted treatment response to DMARDs as well as radiographic progression³. Interestingly, the same B cell scores did not associate with measures of disease activity such as DAS28 in the TNF inadequate responder group. One possible explanation for this discordance is that the TNFiR cohort had overall more aggressive disease, thereby reducing the chances of detecting variance in disease activity. This discrepancy between histology scores of synovitis and DAS28 is in agreement with studies from our group that failed to find associations of inflammatory infiltrates with DAS28 in a cohort of patients with established disease undergoing arthroplasty⁶. We noted some patients in DAS28 remission or with low disease activity harbored subclinical synovitis ^6,7, suggesting that current clinical measures of annotating disease activity, such as DAS28, are not sufficiently sensitive for detecting synovitis in all patients. Indeed, multiple studies have noted ongoing radiographic synovitis as well as progression of erosions in patients with sustained low disease activity or remission^8,9,10,11. Further, over time, patients with longstanding disease may accrue joints with various levels and stages of synovial inflammation and damage. This increased joint to joint variability may exacerbate sampling error in patients with established disease.

The findings of Rivellese et al (CITE) with those of other recent studies are advancing our understanding of the transcriptional and cellular characteristics of the synovium in rheumatoid arthritis. Incorporation of synovial assessments into clinical management of patients is the next step in empowering clinicians to apply advances in molecular immunology to better tailor treatment decisions. Identification of subsets of RA that will specifically respond to therapeutics will enable targeting of their respective underlying disease mechanisms. Together, this work and complementary work by other groups will set the stage for “precision therapy for RA” and thereby advance the care for our patients.