Disease-associated brain activation predicts clinical response to TNF inhibition in rheumatoid arthritis (PreCePra): a randomised, multicentre, double-blind, placebo-controlled phase 3 study.

Background: Rheumatoid arthritis is an inflammatory disease frequently treated with TNF inhibitors. Little is known about predictors of response to TNF inhibitors. Because clinical response in rheumatoid arthritis is measured by composite scores containing subjective patient-orientated domains (eg, pain and global disease perception), we hypothesised that patients with high disease representation in the CNS might respond better to TNF inhibitors than patients with less CNS disease representation.

Methods: We did a phase 3, multicentre, double-blind, placebo-controlled, parallel-group randomised trial in patients with active rheumatoid arthritis at six rheumatology centres across Germany, Portugal, and Serbia. All patients had a functional MRI (fMRI) brain scan at baseline to measure CNS pain activation. Patients (aged ≥18 years) with active rheumatoid arthritis who have active disease despite the use of at least one conventional synthetic disease-modifying antirheumatic drug were stratified according to fMRI (high volume or low volume) and were randomly assigned 2:1 using a randomisation list generated by a study statistician to treatment with the TNF inhibitor certolizumab pegol (400 mg subcutaneously on weeks 0, 2, and 4, and 200 mg once every 2 weeks for maximum 24 weeks) or placebo. Patients and clinicians were masked to allocation. The primary outcome was the proportion of patients reaching low disease activity (Disease Activity Score in 28 joints ≤3·2) at week 12, analysed in the intention-to-treat population. There was no lived experience involvement in study design. The study was registered with EudraCT (2013-000337-13) and ClinicalTrials.gov (NCT01864265).

Findings: Between Sept 3, 2013, and Jan 10, 2020, 148 patients with rheumatoid arthritis were screened and 139 (99 [71%] women and 40 [29%] men) were randomly assigned to the high-volume certolizumab pegol group (n=49), the low-volume certolizumab pegol group (n=43), or the placebo group (n=47). Low disease activity was reached by 28 (57%) in the high-volume certolizumab pegol group, 19 (44%) in the low-volume certolizumab pegol group, and 12 (26%) in the placebo group at week 12. Response in the high-volume certolizumab pegol group was significantly different (p=0·0017) to the placebo group, but not the low-volume certolizumab pegol group (p=0·063). There were 25 treatment-related adverse events: 22 in the certolizumab pegol groups and three in the placebo group.

Interpretation: High disease-associated fMRI CNS pain activation might predict clinical response of patients with rheumatoid arthritis to TNF inhibitor treatment.

Funding: UCB Biopharma.