Background/Purpose: Defining the precise CD4+ T cell subsets that are dysregulated in RA patients is critical to deciphering pathogenesis. Here we present Mixed effects modeling of Associations of Single Cells (MASC), a novel reverse single cell association strategy to determine if a cellular subpopulation is associated with case-control status while controlling for technical confounders and biological covariates. We applied MASC to identify T cell populations expanded in rheumatoid arthritis (RA).

Methods: Cryopreserved PBMCs from 26 RA and 26 osteoarthritis patients were analyzed by a 32-marker mass cytometry panel. All RA patients met ACR 2010 Rheumatoid Arthritis classification criteria. We identified 19 CD4+ T cell populations and tested for case-control associations using our novel association testing method, MASC. MASC uses a logistic regression mixed effects model where the cluster membership of each single cell is the dependent variable. The model estimates fixed effects for donor age, sex, and case-control status, and random effects for each donor and technical batch.

Results: MASC revealed a significantly expanded population of CD4+ T cells, identified as CD27- HLA-DR+ effector memory cells, in RA patients (OR = 1.7; p = 1.1×10^-3, Figure 1A). These cells comprised ~2% of CD4+ memory T cells in RA samples compared to 0.8% in OA controls. Compared to peripheral blood, synovial fluid and synovial tissue samples from RA patients were significantly enriched for CD27- HLA-DR+ cells, which comprised ~10-15% of synovial CD4+ T cells (p < 1×10^-3, Figure 1B). The abundance of CD4⁺ CD27^Ð HLA-DR⁺ cells decreased in RA patients who responded to immunosuppressive therapy (p = 0.006). Considered together, mass cytometry and flow cytometry analyses indicated that CD27- HLA-DR+ cells were significantly associated with RA (meta-analysis p = 4.8×10^-5). CD27- HLA-DR+ cells express a distinctive effector memory transcriptomic program with Th1- and cytotoxicity-associated features and produce abundant IFN-g and granzyme A upon stimulation (Figure 1C).

Conclusion: MASC is a sensitive and well-calibrated statistical method for analyzing high-dimensional mass cytometry data. We used it in an RA case-control study to identify a novel disease-associated CD4+ T cell population. The expansion of a unique CD27- HLA-DR+ effector memory population in RA periphery may represent disease-driving cells and could be ideal for defining the antigenic repertoire in RA. We propose that MASC is a broadly applicable method to perform association testing with single cell data and can help identify other cellular populations that are critical to rheumatic disease pathogenesis.