There was no re lationship between log CXCL13 levels and the presence of erosions at baseline in seropositive patients, regardless of whether it was evaluated by Spearman correlation accor ding to number of erosions or by t test to compare the presence or absence of erosions. Relationships between serum CXCL13 and immunoglobulin A rheumatoid factor Despite a strong relationship of serum CXCL13 with IgM RF seropositivity, both established and recent onset RA cohorts exhibited weaker correlations between CXCL13 and serum IgG and IgG ACPA levels. These data suggest that elevated serum CXCL13 levels might correspond to a process in which non class switched B cells producing IgM RF were promoted independently of follicle and ger minal center formation that leads to immunoglobulin heavy chain class switching and IgG ACPA.

To begin to address this model, we analyzed the relationship between serum CXCL13 levels and IgA RF. Analysis of log transformed CXCL13 and log trans formed IgA RF in the seropositive patients of both co horts were strongly correlated. When evaluated by CXCL13 tertile analysis, the highest tertile had much higher IgA RF values than the first and second tertiles in the Dartmouth RA Cohort 45. 0 IU/ml, first and second tertile geometric mean 11. 2 IU/ml . P 0. 0001. A simi larly strong correlation was seen in recent onset, mostly untreated RA patients 74. 1 IU/ml, first and second tertile geometric mean 20. 4 IU/ml . P 0. 0001. Thus, serum CXCL13 levels exhibited strong correlations with both IgM and IgA RF titers.

Discussion We report a strong relationship between elevated serum CXCL13 levels and seropositive RA that was seen in an established disease cohort and confirmed in a mostly un treated early RA cohort. Modest Drug_discovery associations were seen with disease activity measures in established RA, but no associations were present in patients with early disease. We did not find any evidence to support a relationship between CXCL13 and HLA DRB1 alleles containing the shared epitope or complement C4B deficiency. The most striking finding was the strength of the relationship bet ween serum CXCL13 with IgM and IgA RF relative to that seen with ACPA titers, an ob servation not due to the heterophilic effects of RF. Although an association between CXCL13 and ACPA was observed, it was not as strong a relationship as that seen with RF.

These observations sug gest interesting and potentially specific associations of CXCL13 with both RF autoantibody formation and the pathogenesis of RA. We evaluated patients with very high CXCL13 values and did not observe any significant variation in RF or ACPA values compared with the remaining pa tients in the highest tertile. Further, we did not identify any competing diagnosis or therapy to account for the very high levels of CXCL13.