GM-CSF Expression and Macrophage Polarization in Joints of Undifferentiated Arthritis Patients Evolving to Rheumatoid Arthritis or Psoriatic Arthritis

BACKGROUND AND AIMS: GM-CSF-dependent macrophage polarization has been demonstrated in rheumatoid arthritis (RA). Our aim was to seek diagnostic/prognostic biomarkers for undifferentiated arthritis (UA) by analyzing GM-CSF expression and source, macrophage polarization and density in joints of patients with UA evolving to RA or PsA compared with established RA or PsA, respectively. METHODS: Synovial tissue (ST) from patients with UA evolving to RA (UA>RA, n=8), PsA (UA>PsA, n=9), persistent UA (UA, n=16), established RA (n=12) and PsA (n=10), and healthy controls (n=6), were analyzed. Cell source and quantitative expression of GM-CSF and proteins associated with pro-inflammatory (GM-CSF-driven) and anti-inflammatory (M-CSF-driven) macrophage polarization (activin A, TNFα, MMP12, and CD209, respectively) were assessed in ST CD163(+) macrophages by multicolor immunofluorescence. GM-CSF and activin A levels were also quantified in paired synovial fluid samples. CD163(+) macrophage density was determined in all groups by immunofluorescence. RESULTS: Synovial stromal cells (FAP(+) CD90(+) fibroblast, CD90(+) endothelial cells) and CD163(+) sublining macrophages were the sources of GM-CSF. ST CD163(+) macrophages from all groups expressed pro-inflammatory polarization markers (activin A, TNFα, and MMP12). Expression of the M-CSF-dependent anti-inflammatory marker CD209 identified two macrophage subsets (CD163(+) CD209(high) and CD163(+) CD209(low/-)). CD209(+) macrophages were more abundant in ST from healthy controls and PsA patients, although both macrophage subtypes showed similar levels of pro-inflammatory markers in all groups. In paired synovial fluid samples, activin A was detected in all patients, with higher levels in UA>RA and RA, while GM-CSF was infrequently detected. ST CD163(+) macrophage density was comparable between UA>RA and UA>PsA patients, but significantly higher than in persistent UA. CONCLUSIONS: GM-CSF is highly expressed by sublining CD90(+) FAP(+) synovial fibroblasts, CD90(+) activated endothelium and CD163(+) macrophages in different types of arthritis. The polarization state of ST macrophages was similar in all UA and established arthritis groups, with a predominance of pro-inflammatory GM-CSF-associated markers. CD163(+) macrophage density was significantly higher in the UA phases of RA and PsA compared with persistent UA. Taken together, our findings support the idea that GM-CSF is a strong driver of macrophage polarization and a potential therapeutic target not only in RA but also in PsA and all types of UA.