Background: Rheumatoid arthritis (RA) is a disease of chronic systemic inflammation (SI). SI is assessed either by the presence of acute phase proteins or by ratios of the leukocyte cell count (e.g. Neutrophil to Lymphocyte Ratio (NLR)) in blood. These measures of SI have been reliably used to assess development and progression of RA. However, cell count based SI measurements rely on the availability of freshly collected blood samples, which are not available for many studies. Recently, new algorithms have been developed to reliably estimate cell counts based on DNA methylation data and were used to estimate the methylation derived NLR (mdNLR).
Methods: In the present study, we used four datasets to explore whether mdNLR might be a marker of SI in new-onset, untreated and treated prevalent RA cases, and/or a marker of treatment response to the tumour necrosis factor inhibitor (TNFi) etanercept.
Results: mdNLR was associated with increased odds of being a RA case amongst untreated individuals (OR= 2.32, 95%CI = 1.95-2.80, P <2x10-16), and performed better in distinguishing new onset RA cases from controls compared to covariates: age, gender and smoking status (AUCmdNLR = 0.80, 95%CI = 0.768-0.833, AUCcov. = 0.56, 95% CI: 0.52-0.606, P <2x10-16). In untreated pre-clinical RA cases and controls, mdNLR at baseline was associated with diagnosis of RA in later life after adjusting for batch (ORadj= 4.30, 95%CI = 1.52-21.71, P = 0.029) although no association was observed before batch correction. When prevalent RA cases were treated, there was no association with mdNLR in samples before and after adjustment with batch (ORadj= 0.34, 95%CI = 0.05-1.82, P = 0.23), and mdNLR was not associated with treatment response to etanercept (OR= 1.10, 95%CI = 0.75-1.68, P = 0.64). Conclusions: Our results indicate that SI measured by DNA methylation data is indicative of recent onset of RA. Although we observed an association between pre-clinical RA and mdNLR, there was no difference in mean mdNLR between pre-clinical RA cases and controls. mdNLR is not associated with RA case status if treatment for RA has commenced, and it is not associated with treatment response. It would be interesting to test if mdNLR could be a useful marker for predicting and/or diagnosing RA in absence of cell count measurement.