Precision Medicine Today: Predicting Treatment Response in Patient Subgroups

LeoWolfert / shutterstock.com

SNOWMASS VILLAGE, COLO.—Choosing the right treatment at the right time is the brass ring all rheumatologists hope for. Precision medicine provides the ability to leverage clinical, biomarker and omics data to predict and personalize future treatment for rheumatoid arthritis (RA).

“New data and new methods to analyze the data are helping us better predict patterns of disease and treatments that work well for subgroups of patients,” noted Jeffrey Curtis, MD, MS, MPH, during his presentation on precision medicine in rheuma­tology at the 2020 ACR Winter Symposium in January. Dr. Curtis is the Harbert Ball Professor of Medicine in the Division of Clinical Immunology & Rheumatology, University of Alabama at Birmingham (UAB). He is the co-director of the UAB Pharmacoepidemiology Unit and has led the clinical trials unit at UAB for more than a decade, with a focus on RA and psoriatic arthritis and spondyloarthritis.

Predicting RA Outcomes

Early indications are that some prediction models may be accurate enough for clinical care and management. But Dr. Curtis asked, “How good a prediction is good enough for clinicians?”

Dr. Curtis previously conducted research to predict low disease activity and remission using early treatment response to anti-tumor necrosis factor therapy in patients with RA using data from an etanercept trial (TEMPO). He said the results of that trial indicated an accurate prediction could be made for 80–90% of patients within 12 weeks of starting therapy about whether they were likely to have low disease activity at week 52. However, researchers needed to see the remaining 10–20% of patients following additional time on therapy to determine whether treatment should continue.¹

Subsequent investigations have looked at predicting treatment response earlier (e.g., as early as four weeks after starting therapy). The ability to predict treatment response before a patient starts therapy based on blood or genetic-derived factors remains the holy grail for researchers.

He discussed balancing rigor (such as focusing on low disease activity, if not remission as the optimal target) and practicality (e.g., settling for significant improvement), and maximizing predicted response vs. non-response in assessing treatment efficacy.

In research to predict response to biologic drugs, one growing area of study is the identification of immunological biomarkers of treatment response through such methods as evaluating synovial tissue and fluid. Various biomarkers produced by a range of cell types have been suggested to be important. For example, macro­phages in the synovial lining of the joint are important elements of RA patho­genesis, and macrophage-derived markers have been correlated with future treatment response.

One study used single-cell RNA-sequencing to uncover four distinct subsets of mouse synovial macrophages and identified a dynamic role for synovial macrophages in inflammation.²

Dr. Curtis stressed the accuracy of prediction models should be judged by clini­cians to consider maximum benefit in real-world application.