PrismRA: a personalized medicine test that accurately predicts rheumatoid arthritis patients who will not respond adequately to tumor necrosis factor inhibitors


Keith Johnson, Nancy Schoenbrunner and Dina Ghiassian

2018 ACR/ARHP Annual Meeting


Background: Rheumatoid arthritis (RA) is a complex disease affecting about 1.3M adults in the US. The patient population is very heterogeneous; no single drug mechanism achieves treatment targets in the majority of patients. Tumor necrosis factor inhibitors (anti-TNFs) are the first line therapy in >90% of biologic naïve patients, although less than half of patients achieve treatment targets on them. This is despite the fact that in the past decade therapies with alternative mechanisms have been approved. The case for a test to predict which therapy a patient will respond to is overwhelming because of the disease heterogeneity, and because the approved therapies for RA are molecularly targeted. Scipher is developing PRISM-RA to accurately predict from a baseline blood sample, using a gene expression profile, which patients will not reach their treatment target on anti-TNFs. Identified non-responders will be offered alternative approved therapies much earlier in their treatment course than currently in clinical practice. This will lead to better patient outcomes and significant savings of healthcare costs.

Methods: The test user requirements were developed based on in depth discussions with payers and rheumatologists. The overwhelming consensus is that Prism-RA must identify non-responders (NRs) with >90% accuracy and with sufficient specificity to identify the majority of NRs. Response is defined as achieving ACR-defined treatment targets, equating to a minimum of ACR50 improvement in disease activity after 3 months of therapy. Independent patient cohorts treated with anti-TNFs with associated baseline gene expression data were analyzed to train a classifier to predict response.

Results: In a 50 subject training cohort Prism-RA achieved >90% accuracy (NPV) with the required specificity. Prism-RA passed proof of concept (POC) by exceeding the pre-specified user requirements in a blinded fashion across an independent cohort of 39 subjects. Scipher is validating Prism-RA in an ongoing observational study of >150 patients; the results of this confirmatory study will be presented in the near future. Thereafter, Scipher will conduct an appropriately powered, double-blinded prospective interventional trial to confirm both clinical performance and utility of Prism-RA. Prism-RA will be offered commercially as CLIA certified laboratory developed test (LDT). During the prospective trial, data will be collected to train classifiers to predict responders across all classes of approved targeted therapies for RA.

Conclusion: Prism-RA has achieved POC and will provide rheumatologists with a scientific rationale for selecting therapies based on a biologic naive patient’s molecular data before initiating treatment, meaning that more patients will achieve their treatment targets than is the case today. By de-selecting non-responders the overall rate of achieving treatment targets on anti-TNFs will increase significantly. Ultimately, Prism-RA will provide response predictions for all approved targeted therapies for RA, providing rheumatologists with a validated scientific rationale for the selection of the right therapeutic agent for an individual patient.

View Full Paper