WASHINGTON – Three gene expression markers appear to predict response to some of the most common and powerful rheumatoid arthritis drugs employed in clinical practice.
In the largest gene-expression study of these drugs to date, researchers from Glasgow found 23 genes that predicted response to TNF inhibitors, and 23 more that predicted response to rituximab. They also found eight genes that predicted positive response to both types of drugs.
Just as important, the team found that 10% of patients in the study had none of the markers, Duncan Porter, MD, said during a press briefing at the annual meeting of the American College of Rheumatology.
If confirmed in independent validation cohorts, the findings could revolutionize medical therapy for rheumatoid arthritis, Dr. Porter said in an interview.
“Right now, the choice of treatment is mostly a flip of the coin,” said Dr. Porter, a rheumatologist at Queen Elizabeth University Hospital in Glasgow. “We don’t have a lot of comparative data to support one drug over the other, and most head-to-head studies show noninferiority.
“So, it would be ideal to be able to identify patients who will respond to one class or the other, and give them the right medicine the first time,” he noted. “If we can do that, we have the beginnings of really meaningful personalized therapy in RA.”
Dr. Porter reported a transcriptome-wide association study of 250 patients with RA who were enrolled in the ORBIT trial (Lancet. 2016 Jul 16;388[10041]:239-47). Dr. Porter was the study’s primary investigator.
ORBIT comprised 295 patients with active, seropositive rheumatoid arthritis who had failed disease-modifying antirheumatic therapy. They were randomized to B-cell depletion with rituximab or to the TNF inhibitors adalimumab or etanercept. ORBIT concluded that clinical response was similar with all three drugs. However, while most patients did well on their assigned drug, no matter which class, 20% failed their initial assignment.
The genetic study examined RNA transcripts in whole blood samples among a subset of the ORBIT patients. Most of the samples (70%) were used to identify markers and created a prediction model; that was then verified in the remaining 30% of samples. Markers were correlated to clinical response associated with rituximab and TNF inhibition. Clinical response was defined as a 1.2-point decline in the Disease Activity Score from baseline to 3 months of therapy.
At least 1 of the 54 genetic response markers was present in 90% of the population; 50% would have responded well to either drug, or 40% to just one of them.
But 10% of the population lacked either of the response markers – a very important finding, Dr. Porter noted, because if identified early, these patients could potentially avoid ineffective medication trials.
The markers’ predictive values were uniformly high, with a sensitivity of 93% and specificity of 91%. The positive predictive value was 96%, and the negative predictive value was 86%. Patients predicted to respond at 3 months were also more likely to have a good response (43% vs. 23%) or remission (23% vs. 10%), as measured by the Disease Activity Score 28 at 12 months.
While Dr. Porter was excited about the findings, he eyed them cautiously.
“The landscape is littered with biomarkers that had great early promise but failed in later validation studies,” he said. “We must be very careful in interpreting this. However, the other response studies have been small, and have not compared two drugs. This study was large with robust results, so we are encouraged.”
Dr Porter is now designing the protocol for a large external validation cohort study, which he hopes to launch in 2017.
Roche funded aspects of the ORBIT trial and the genetic substudy. Dr. Porter has been a consultant for Roche, as well as for AbbVie and Pfizer.
msullivan@frontlinemedcom.comOn Twitter @alz_gal