Abstract

Abstract Background/Aims Diagnosing rheumatoid arthritis (RA) early is difficult despite the use of anti-citrullinated protein antibodies in the new EULAR-2010 classification criteria, particularly for antibody-negative patients. Alterations in epigenetic patterns have been associated with RA CD4+T-cells at the TNF gene locus (1). The second aim of my PhD project was to select a CpG site to develop a quantitative methylation sensitive PCR (qMSP) assay as a diagnostic biomarker test for RA classification. Methods Patients attending an early arthritis clinic were used (with full ethical approval). An 450K Illumina methylation genome-wide dataset analysing DNA from CD4+T-cells identified CpG on TNF promoter region as a differentially methylated CpG between controls and RA patients (1). A qMSP assay was designed using a TaqMan approach (primers and probe), using a reference gene (GAPHD) for normalisation. The assay was validated in DNA extracted from PBMC. The performance of the assay for RA classification was accessed using binary logistic regression. Results We noted differences in levels of methylation in CD4+T-cells in RA, notably affecting the TNF gene promoter in about 15% of cells. We validated this using bisulphite sequencing showing about 30% of cells with differential methylation in early RA compared to controls. This allowed us to select cg11484872, cg21370522, and cg01569083 as candidates. We optimised a qMSP assay using fully methylated/un-methylated DNA. PBMC samples were retrieved from our tissue bank for 65 early RA, and 64 non-RA patients (including 11 reactive arthritis, 37 undifferentiated arthritis, and 16 psoriatic arthritis). The methylation levels (%) detected by the TNF-qMSP assay were significantly lower in RA (Δ-DM =-10.28%, p = 7x10-7). The loss of TNF promoter gene methylation was associated with RA classification (p < 0.0001, unadjusted OR = 0.900 (95% CI: 0.85-0.94)) with a good classification performance (AUROC = 0.76 (95% CI: 0.67, 0.84)). An adjusted regression model using the demographic and clinical parameters associated with RA classification in this group of patients (age, smoking, counts for tender and swollen joints) showed that adding TNF to the clinical model improved the model’s performance (increasing AUROC from 0.86 , to 0.91). Comparing classification accuracy, adding the TNF-qMSP resulted in an increased correct classification of 5.7 % compared to the clinical model. A similar result was obtained in the antibody negative patients, with TNF-qMSP assay improving the model’s classification from AUROC 0.91 to 0.96 and 3.3% more patients correctly classified. Conclusion The loss of DNA methylation on the TNF gene promoter identified in early RA patients provided a valuable target for a biomarker assay development. The results will need to be confirmed in a larger cohort but the TNF-qMSP assay offers good opportunities for RA versus non-RA classification particular in antibody negative disease. Disclosure R. Pitaksalee: Grants/research support; ROYAL THAI GOVERNMENT SCHOLARSHIP. P. Emery: None. R. Hodgett: None. F. Ponchel: None.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call