Immunoprofiling of early, untreated rheumatoid arthritis using mass cytometry reveals an activated basophil subset inversely linked to ACPA status

Abstract

Background: Autoantibody production is a hallmark of rheumatoid arthritis (RA). Anti-citrullinated protein antibodies (ACPA) are highly disease-specific, and their presence is associated with more severe disease and poor prognosis compared to ACPA-negative patients. However, the immune cell composition associated with antibody-positive/negative disease is incompletely defined. Mass cytometry (MC) is a high-dimensional technique offering new possibilities in the determination of the immune cell composition in rheumatic diseases. Here, we set up a broad phenotyping panel to study the immune cell profile of early untreated RA to investigate if specific immune cell subsets are associated with ACPA+ versus ACPA- RA.

Methods: Freshly obtained PBMCs of early, untreated RA patients (8 ACPA+ and 7 ACPA-) were analysed using a 36-marker MC panel, including markers related to various immune lineages. Data were processed using Cytosplore for dimensional reduction (HSNE) and clustering. Groups were compared using Cytofast. A second validation cohort of cryopreserved PBMCs obtained from early RA patients (27 ACPA+ and 20 ACPA-) was used to confirm MC data by flow cytometry (FC). FC data were processed and analysed using both an unsupervised analysis pipeline and through manual gating.

Results: MC indicated no differences when comparing major immune lineages (i.e. monocytes, T and B cells), but highlighted two innate subsets: CD62L⁺ basophils (p = 0.33) and a subset of CD16^- NK cells (p = 0.063). Although the NK cell subset did not replicate by FC, FC replication confirmed the difference in CD62L⁺ basophil frequency when comparing ACPA+ to ACPA- patients (mean 0.32% vs. 0.13%; p = 0.01).

Conclusions: Although no differences in major lineages were found between early ACPA+ and ACPA- RA, this study identified the reduced presence of activated basophils in ACPA-negative disease as compared to ACPA-positive disease and thereby provides the first evidence for a connection between activated basophils and ACPA status.

Keywords: Basophils; CD62L; Immune-profiling; Mass cytometry; Rheumatoid arthritis.

Fig. 1

Analysis of major lineages revealed no differences between ACPA+ and ACPA− patients. A Five-level HSNE separated all major lineages as indicated by dashed circles, colour-coded for ACPA status (positive = red, negative = blue). The solid circle indicates innate landmarks. B Examples of expression patterns of the HSNE map shown in A. C Heatmap of clusters identified through default clustering in Cytosplore (Gaussian mean shift) showing the expression pattern. D Manual gating of major lineages showed no significant difference in frequency between ACPA+ and ACPA−

Fig. 2

HSNE analysis of all innate subsets indicated differences in basophil and NK cell subsets. A Heatmap representing all clusters within the innate population and their respective expression patterns, cluster 18/24 indicated with black boxes. B Cluster 18 (basophil subset, p = 0.033) was reduced in ACPA− patients and there was a strong trend for cluster 24 (NK cell subset p = 0.063) present only in ACPA− patients (Cytofast). C tSNE plotting samples using clusters 18 and 24 as input confirmed the separation shown in B. D HSNE map indicating cluster 18 and cluster 24 (dashed circles) and examples of the expressed markers. ACPA+ in red, ACPA− in blue

Fig. 3

Independent flow cytometry replication confirmed the reduced presence of CD62L⁺ basophils in ACPA− samples. A Cytosplore dimension reduction of manually transformed flow FCS files of panel 1 showing both ACPA+ (red) and ACPA− (blue). Smaller panels on the right represent the cells within the backbox. Expression patterns matched that of MC cluster 18 (dashed circle), including CD123, FcεRI and CD45RO. CD62L expression was linked to ACPA status. B Gates used in Boolean gating in Flowjo to manually gate cluster 18. C Frequency of manually gated cells representing MC cluster 18. Subset frequency was lower in ACPA− RA (mean 0.32% vs 0.13%; p = 0.01)