RNA sequencing and machine learning as molecular scalpels

The DNA methylation Profile of Undifferentiated Arthritis Patients Anticipates their Subsequent Differentiation to Rheumatoid Arthritis

OBJECTIVE

Undifferentiated arthritis (UA) is the term used to cover all the cases of arthritis that do not fit a specific diagnosis. A significant percentage of UA patients progress to rheumatoid arthritis (RA), others to a different definite rheumatic disease, and the rest undergo spontaneous remission. Therapeutic intervention in patients with UA can delay or halt disease progression and its long-term consequences. It is therefore of inherent interest to identify those UA patients with a high probability of progressing to RA who would benefit from early appropriate therapy. We hypothesized that alterations in the DNA methylation profiles of immune cells may inform on the genetically- or environmentally-determined status of patients and potentially discriminate between disease subtypes.

METHODS

In this study, we performed DNA methylation profiling of a UA patient cohort, in which progression into RA occurs for a significant proportion of the patients.

RESULTS

We find differential DNA methylation in UA patients compared to healthy controls. Most importantly, our analysis identifies a DNA methylation signature characteristic of those UA cases that differentiate to RA. We demonstrate that the methylome of peripheral mononuclear cells can be used to anticipate the evolution of UA to RA, and that this methylome is associated with a number of inflammatory pathways and transcription factors. Finally, we design a machine-learning strategy for DNA methylation-based classification that predicts the differentiation of UA patients towards RA.

CONCLUSION

DNA methylation profiling provides a good predictor of UA-to-RA progression to anticipate targeted treatments and improve clinical management.