Research Spotlight: Biomarkers for Anti-TNF Therapy Response in RA

August 4, 2016

With the hope of wasting less time on medications doomed to fail, researchers are seeking biomarkers to predict a patient’s response to particular RA treatments.

In rheumatoid arthritis, swift treatment is key - but valuable time can slip away when patients don't respond to the first drug treatments tried.

As a result, researchers are seeking some way to predict who will respond to which treatment, with the hope of wasting less time on medications doomed to fail. The subtext of this research is that rheumatoid arthritis isn't just one disease, but a condition caused by diverse molecular pathways.

Fewer than half of rheumatoid arthritis patients experience an improvement in their arthritis of at least 50 percent when treated with any one biologic, according to a 2010 review on biomarkers in the Scandinavian Journal of Clinical and Laboratory Investigation. Investigators have searched for biomarkers to predict drug response in the gene expression of white blood cells and in cytokine profiles in whole blood. The task is complicated by ethnic differences among cohorts, though some complex biomarker profiles show promise across groups. A 2009 study published in Arthritis Research & Therapy, for example, developed a 24-biomarker assay that could predict response to etanercept with accuracy ranging from 58 percent in a Japanese cohort to 72 percent in a U.S. Caucasian cohort.

Ultimately, no one molecule is likely to unveil the secrets of drug response in one fell swoop. Instead, the cumulative approaches of multiple laboratories investigating multiple biomarkers will likely converge on patterns that can help illuminate which patients will respond and which won't, said Timothy Niewold, M.D., who researches autoimmune disease at Mayo Clinic in Minnesota.

Dr. Niewold and his colleagues have been investigating a particular biomarker, the ratio of IFN-β to IFN-α in pretreatment serum, as a predictor of response to anti-TNF therapy. In a paper published online November 6 in the journal Annals of the Rheumatic Diseases, researchers found that patients with an IFN-β to IFN-α ration of greater than 1.3 were significantly less likely to respond to TNF inhibitors (p=0.018, OR 6.67, 95 percent CI, 1.37-32.55).

Rheumatology Network spoke with Dr. Niewold to learn more about this study and how it fits into the puzzle of drug response in rheumatoid arthritis.  [[{"type":"media","view_mode":"media_crop","fid":"50707","attributes":{"alt":"Timothy B. Niewold, M.D.","class":"media-image media-image-right","id":"media_crop_759950591673","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"6210","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"font-size: 13.008px; line-height: 1.538em; float: right;","title":"Timothy B. Niewold, M.D.","typeof":"foaf:Image"}}]]

RN: Why were you interested in pursuing IFN as a biomarker?

Dr. Niewold: The concept is that we really don't know who will respond to TNF-α blockers before we give them. lt's a common problem throughout a lot of rheumatology, and a lot of the medications that we use. We try one of the things that we think is most likely to work best, but we can't be really sure, in a given person, whether it is going to work or not. The hope is that we can find some sort of a marker in the blood that you could check before you gave the medicine that would give you an idea of whether they would respond or not.

We had an angle that we thought would be somewhat unique to our group and the work that we do. We do a lot of work with the type 1 interferon system. This is something that interacts with the TNF- α, system so we thought that this might be used to predict response. It seemed that these two cytokines, interferon- α and TNF- α cross-regulate each other to some degree.

There was some preliminary studies that supported this, but we had an opportunity to do a much larger study, and the assay we had for type 1 interferon, we had a lot of confidence in, because we used this for lupus, but this was the first time we applied this to rheumatoid arthritis.

RN:  How did you conduct the study?

Dr. Niewold: We had samples available from RA patients before they started a TNF blocker. There were two registries that we used, we used the ABCoN [Auto-immune Biomarkers Collaborative Network Consortium] registry as well as the TETRAD [Toxicity in Rheumatoid Arthritis Database and Repository] registry. They're both really great prospective sample collection efforts where a blood sample was drawn right before they would start a new medication, in this case a TNF blocker, and then there was prospective follow-up. That was really the benefit. We started off with 32 patients from the ABCoN registry, and we were able to follow up with 92 patients from the TETRAD registry.

We used the ABCON registry to try to do discovery and see if there was an optimal cutoff in the type 1 interferon. And then we applied that in the replication cohort in TETRAD.

RN:  What were your findings?

Dr. Niewold: We did find predictive capacity. One of the interesting findings is that it was interferon-β that was the strongest predictor, and it was actually the ratio of interferon-β to alpha that was most predictive. Those are the two major type 1 interferons and they signal to the same receptor but they have some differences in how they signal and what they do. It seems that accounting for both interferons was important.

In the replication cohort, the specificity was 77 percent and the sensitivity was 45 percent. There was no person that had a really high ratio that had a response to the drug. That's reflected in this high sensitivity - you can actually predict nonresponses fairly accurately. The test performed fairly well for predicting nonresponse. It's a little harder to know responders and there may be people that you may not be able to show that they're going to respond.

We used the EULAR response criteria and it performed like you'd expect. The biggest difference was from the good response versus the nonresponse, but there was still a difference between moderate and nonresponse. It wasn't quite as strong. The greatest difference in responses also showed the greatest difference in the interferon ratio.

RN:  From a physiological standpoint, why might this ratio matter?

Dr. Niewold: It seems in some cases that the TNF-α can cross-regulate the interferon and vice versa [so that high levels of one inhibit production of another]. We were wondering if, say, there was a group of patients who had interferon as part of their pathogeneses, so removing the TNF-α may help is some ways but also remove a break on the interferon - by taking away one cytokine you give a different cytokine more room to work. You sort of remove that natural break. We feel that rheumatoid arthritis is not one uniform condition. There are different people with different molecular pathogenesis and that the variable responses that we when you block different pathways or use specific drugs is the result. There are different toads to rheumatoid arthritis.

RN:  Does the ratio itself cause the disease or affect its pathogenesis?

Dr. Niewold: I don't think so. That would be a stretch. More likely this is indicating a subgroup of patients that has a different pathogenesis. Whether it's causal in these patients I'm not sure, but probably not.

RN:  What is the next step for developing this interferon biomarker?

Dr. Niewold: We're going in two directions. We are accessing a larger group of patients, which I think will be helpful to try to move this into practice. We're likely to see if we can take an even larger cohort and test it in a real-world group of patients. The other direction is going back toward the biology and immunology of it. We think that this marker is telling us something important about the disease. It may not be the best marker in the end. It may lead us back to something that may be an even better marker. We're trying to work backwards to see if we can find some more primary immune system process that's given rise to this ratio. That's still somewhat mysterious, why you would get different ratio of beta to alpha-interferon.

RN:  Is testing for this biomarker expensive or difficult?

Dr. Niewold: It's somewhat tricky. We do it a lot and we developed it for lupus. Applying it as a widespread, validating clinical test would take a lot of work because it is a cell-based assay and there are a lot of moving parts to it. That's one of the reasons we want to move backward to see where this is coming from because there might be a simpler thing that could be tested.

In fairness to the large amount of research that's going on in this area, I think it most likely it will be a multi-parameter kind of a test. I think it's complex enough that we probably won't solve it with a single marker. This is just one thing and you could easily envision kind of a multi-parameter score. There are number of other groups that are chasing down their favorite molecules, and we may all be touching a different part of the same elephant, in the old metaphor.

 

 

References:

Lindstrom TM, Robinson WH. Biomarkers for rheumatoid arthritis: Making it personal. Scandinavian Journal of Clinical and Laboratory Investigation. 2010;70(sup242):79-84. doi:10.3109/00365513.2010.493406.

 

Hueber W, Tomooka BH, Batliwalla F, et al. Blood autoantibody and cytokine profiles predict response to anti-tumor necrosis factor therapy in rheumatoid arthritis. Arthritis Res Ther Arthritis Research & Therapy. 2009;11(3). doi:10.1186/ar2706.

 

Muskardin TW, Vashisht P, Dorschner JM, et al. Increased pretreatment serum IFN-β/α ratio predicts non-response to tumour necrosis factor α inhibition in rheumatoid arthritis. Annals of the Rheumatic Diseases Ann Rheum Dis. June 2015. doi:10.1136/annrheumdis-2015-208001.