Supporting this notion is our observation that the mean CXCL13 levels in the two cohorts were nearly identical. By extension, this result would predict the absence of a correlation between serum CXCL13 and disease activity measures. Indeed, only a modest relationship with DAS28 CRP, and no relationship with hsCRP or CDAI, was seen in the Dartmouth RA Cohort. In addition, no re lationship was observed with baseline hsCRP, DAS28 CRP or radiographic erosions in the recent onset RA patients from the Sherbrooke EUPA Cohort. Similar findings were recently reported in a French cross sectional cohort. Thus, elevated serum CXCL13 levels did not simply reflect quantitative differences in synovial or systemic inflamma tion between patients.

rather, these data suggest the pre sence of a qualitatively distinct subset of seropositive RA manifested by specific increases in both IgM and IgA RF. This interpretation is further supported by the findings reported in a small study demonstrating that serum CXCL13 levels did not correlate with DAS28 measures. Perhaps more important, the same study identified a strong relationship between serum CXCL13 protein and Batimastat synovial CXCL13 mRNA expression. Thus, serum CXCL13 levels appear to derive from a syn ovial inflammation process characterized by the pro duction of CXCL13. Although this interpretation differs somewhat from prior reports, these latter studies may have been confounded by inclusion of seronegative pa tients or the inappropriate use of RF level as a criterion for disease activity. Clearly, additional studies are needed to clarify this issue.

Murine models indicate that the function of CXCL13 CXCR5 interactions promotes recruitment of B cells and follicular T helper cells to the follicle and germinal cen ters in secondary lymphoid organs. In humans, the precise role of CXCL13, let alone rheumatoid synovium, is less clear. Synovial expression of CXCL13 has been as sociated with diffuse lymphoid infiltration as well as the presence of lymphoid aggregates that resemble germinal centers. However, the relationship of synovial histology to either inflammation or autoantibody pro duction remains controversial. The surprisingly strong correlation between serum CXCL13 levels and RF titers, relative to that seen with either serum IgG or IgG ACPA levels, in both an established RA cohort and a recent onset, mostly untreated RA cohort may clarify the role of CXCL13 in autoantibody production. A similar but lesser relationship was observed in a recent cross sectional analysis. The most straightforward interpretation is that a greater pro portion of circulating RF derives from synovial production relative to IgG ACPA and IgG, which are presumably pro duced at other sites, including the bone marrow.