Rheumatoid arthritis and the ‘big bang’ at the ACR! Save

Is there ‘bang for the buck’ using big data to help predict who will develop RA in at-risk populations and similarly to predict response to csDMARDs, TNFi and JAKi in RA?

Top Abstracts

#0774

What: At risk RA patients were followed with longitudinal blood multi-omic profiling.
Why: To identify early and accurately people who will develop RA
How: Multi-omic profiling serially to find between groups differences of those who get RA vs others who don’t. Patients enrolled in the HCQ vs. placebo Stop RA study who were CCP3(+).
Results: For those who developed RA, PTPN22 locus had more chromatin accessibility in NK cells at baseline vs. those who didn’t develop RA. Other gene expression changes were seen.
Next steps: These data although not ready for clinical use yet will need to be verified/replicated and may help to predict who (ACPA+ people) will develop RA over the next few years more accurately.

#0795 Big data were used to predict therapeutic response in RA patients with multi-omics

What: Multi-omic data were used to predict response to Rx in RA patients starting new Rx (csDMARDs in ERA patients) or advanced therapy (TNFi or JAKi) in established RA.
Why: With several options and lack of home runs (rapid deep remission) for many patients with RA who cycle through various Rx, it would be optimal if we could predict a response to Rx early or even before starting a specific therapy
How: Peripheral blood samples of RA patients were used to integrate multi-omics to predict a response to Rx.
Results: Transcriptomes identified 3 groups with distinct gene module involving innate and adaptive immune responses, inflammation, and metabolism. Different results were found depending on Rx. In ERA using csDMARDs there were 5 inflammatory proteins, elevated monocytes, and high disease activity to predict response to Rx. For TNFi 9 proteins and high disease activity were predictive, whereas in JAKi 4 proteins, high lymphocytes, and TJC predicted therapeutic responses
Next steps: Individualizing Rx response is helpful, especially if making a choice between therapeutic classes.

However, the holy grail in prediction (preRA and RA response to Rx), will likely be a highly predictive accurate test that is accessible and low cost. So, we aren’t there yet but these models help to build on our current clinical knowledge. Big data will eventually provide useful prediction models for clinicians.