The search for biomarkers in rheumatoid arthritis (RA), ones that will predict which patients will respond to certain therapies, is often described as the search for the holy grail—long, frustrating, intense and elusive.
You Might Also Like
Explore This IssueFebruary 2015
Also By This Author
“Biomarkers in RA are hard to find. It’s as simple as that. And the real reason is that there is no simple biomarker,” says Eric Ruderman, MD, professor of medicine at Northwestern University Feinberg School of Medicine. “A lot of people think that what we call RA is really not one disease, but is a syndrome that involves a number of people who are going to have something different both phenotypically and genotypically.”
The field of biomarker research, therefore, moves slowly “because it’s very complicated. If it was easy, we would have had it by now,” he says.
Joel Kremer, MD, Pfaff Family Professor of Medicine at Albany Medical College, says that researchers remain unclear about what questions to ask in the search for predictive biomarkers, given that there are potentially tens of thousands of them. “The ideal biomarker would have an evaluable baseline, which predicts which drug the patient would do well on. But we are not close to that.
“We are really dipping our toes in a very large lake. The reason it is taking so long is that people are on a lot of fishing expeditions, and it’s very difficult to predict in advance a biomarker that is associated with a particular intervention,” Dr. Kremer says.
Outcomes of research cannot be predicted, he says. Even the early promise of individualized medicine that started with the human genome project “has really not paid the kinds of dividends that we anticipated. I think that science has embraced the idea that these areas are complex and multigenomic and, except for some rare, single-gene mutations that cause rare diseases, many of these diseases are multigenomic,” Dr. Kremer says.
Predicting Response to Therapy
The concept of biomarkers arose from single-cell mutational events, as seen in cancer, according to William F.C. Rigby, MD, professor of medicine at Geisel School of Medicine at Dartmouth. That prompted interest in finding surrogates of disease activity that are reflected in the blood stream or in DNA that can work in polygenic disorders, such as RA or psoriatic arthritis. “The challenge in these disorders is identifying biomarkers that are not epiphenomena,” he says.
Biomarkers can cover a multitude of indications, Dr. Rigby says. One kind of biomarker can make an early diagnosis of a disease; another could be used to predict an outcome of a disease, such as discriminating between RA that will be difficult to control and cause a lot of joint damage and a more indolent form of the disease; others could be used to predict responses to different types of therapy.
Although a combination of medications in a treat-to-target approach will work for most patients, some people are completely refractory to any therapy. “Rather than cycling through three or four treatments, a biomarker could help you pick the ideal treatment,” Dr. Rigby says. Although RA is a polygenic disorder and can be divided between seropositive and seronegative disease, “those distinctions have really not had that much power in terms of guiding therapy.”
Targets with Potential
William Robinson, MD, associate professor of immunology and rheumatology at Stanford School of Medicine (Calif.), says the objective is to “create biomarkers that are informative in different dimensions than DAS scores. DAS scores simply measure current disease activity, and we need biomarkers that can predict response to therapy.”
There are no good candidates currently for biomarkers that could be used to identify subsets of patients likely to respond to a particular therapy, but the “overarching objective is to develop such in the future.” Mechanistic biomarkers offer the greatest potential, Dr. Robinson says. In a recent article in Nature Reviews Rheumatology, Dr. Robinson and colleagues explained that a mechanistic biomarker is directly involved in the pathogenesis of a disease, enables differentiation of distinct subtypes of the same disease, can be used to stratify disease and target treatment, and could indicate whether a therapy is targeting the cause of a disease.1
In rheumatology, defining a new molecular taxonomy of disease and the subsequent identification of diagnostic, predictive and prognostic mechanistic biomarkers begins with first stratifying current clinical classifications and identifying molecular pathways that mediate the pathogenesis of disease, they wrote.
“We anticipate that involvement of certain molecular pathways will be shared across subsets of multiple different rheumatic diseases, whereas other pathways will be disease specific. Molecular classification of disease could enable the identification of disease subtypes that are responsive to specific therapeutics and eventually the use of patient-derived biomarkers for guiding target therapy,” they said.
Incremental research is underway to identify mechanistic biomarkers that may one day guide therapy. A study by Dennis and colleagues in Arthritis Research & Therapy used genome-wide expression analysis of synovial tissues from a large RA cohort to look for potential biomarkers.2 The researchers used a combination of global gene expression, histologic and cellular analyses, and analysis of gene expression data. That research defined molecular and cellular phenotypes that reflect the wide heterogeneity present in RA synovium.
The researchers found evidence of four major phenotypes of RA synovium: lymphoid, myeloid, low inflammatory and fibroid, and each type has a distinct underlying gene expression signature. Baseline synovial myeloid, but not lymphoid, gene signature expression was higher in patients with good compared with poor EULAR clinical response to anti-TNFα therapy at Week 16. High baseline serum soluble intercellular adhesion molecule 1 (sICAM1), associated with the myeloid phenotype, and high serum C-X-C motif chemokine 13 (CXCL13), associated with the lymphoid phenotype, had different responses to anti-TNFα therapy (adalimumab) compared with responses to anti-IL6R therapy (tocilizumab). Patients with high sICAM1 and low CXCL13 had the highest ACR50 response rate at Week 24 to anti-TNFα monotherapy compared with patients with low sICAM1 and high CXCL13.
The researchers said that their data demonstrate that underlying molecular and cellular heterogeneity in RA affects clinical outcome to therapy. Patients with the myeloid phenotype exhibited the most robust response to anti-TNFα therapy in their study, suggesting a future pathway to identify and validate serum biomarkers that could predict response to targeted therapies, they concluded.
Timothy B. Niewold, MD, associate professor of medicine at Mayo Clinic in Rochester, Minn., presented research (Abstract 2927) at the 2014 ACR/ARHP Annual Meeting in Boston that looked at the association of circulating type I interferon levels and response to biologic therapies. Results of that research found that the increased pretreatment serum IFN-β:IFN-α inhibition ratio was strongly associated with nonresponse to TNFα inhibition by EULAR criteria at 12–14 weeks. IFN-β:IFN-α ratio greater than 1.3 was significantly more likely to have a nonresponse by EULAR criteria at 12 weeks, and no patient in the study with that ratio or greater achieved a good response. The conclusion was that the blood test may be useful in making treatment decisions about use of TNFα inhibitors in RA.
DAS scores simply measure current disease activity, & we need biomarkers that can predict response to therapy.
Dr. Ruderman describes that type of research as a fine example of “out-of-the-box” thinking that seems wise in the search for biomarkers. The researchers looked at interferon levels, interferon alpha and beta, “and it turns out the ratio of the two was helpful in predicting response to TNF inhibitor therapy, which you wouldn’t think of because it has nothing to do with TNF inhibitor therapy. Obviously the results have to be vetted, and it has to be replicated prospectively, but it was particularly useful at deciding who was unlikely to have a good response to a TNF inhibitor.”
Other “remarkable strategies” are emerging in the search for biomarkers, according to Dr. Rigby, who pointed to an article in the New England Journal of Medicine for an example.3 In that research, investigators looked at response of patients with melanoma treated with ipilimumab or tremelimumab. The malignant melanoma exomes from 64 patients treated with CTLA-4 blockade were sequenced and characterized for their ability to predict clinical response. Their findings exemplify the potential of these approaches when they are applied to rheumatology, Dr. Rigby says.
Psoriatic arthritis is another type of rheumatologic disease in great need of a predictive biomarker, because skin psoriasis occurs many years before developing into psoriatic arthritis, Dr. Ruderman says. Only about 30% of people with skin psoriasis develop joint disease. “Right now, we can only watch it more carefully for when joint disease develops, but maybe we can then figure out what to do to prevent it from progressing if we could figure out who those people are.”
The search for biomarkers in rheumatic diseases may depend on some kind of breakthrough that is brought on by exome sequencing or by protein analysis or some other strategy. “The power of novel technologies is so striking and advancing so quickly that I think that it will happen. But I can’t decide whether these things are going to happen based on analyzing synovium from affected joints or some completely unexpected tissue in RA,” Dr. Rigby says.
“All of these things are out there. For the moment, the most encouraging thing for me is that the field is becoming increasingly rigorous, that questions are increasingly refined. First reports are always greeted with warmth and a certain amount of enthusiasm, but also with ‘let’s see it confirmed,’” he says.
Dr. Kremer says the Corrona registry, which has data on 40,000 patients with RA and psoriatic arthritis who are on biologic therapy, is looking at genetic or biomarker analyses on 2,800 of those patients. “Do we necessarily expect to find a smoking gun? That would be unrealistic, but if we can advance the ball down the field.”
Predicting Disease Progression
Biomarker research has shown results for a measurement of disease activity that helps predict the risk of future joint damage. Study of biomarkers in blood led to development and marketing of a composite test, Vectra DA (Crescendo Bioscience), which measures 12 serum proteins to provide a score on a scale of 1–100. The proteins, or biomarkers, included in the test are VCAM-1, EGF, VEGF-A, IL-6, TNF-R1, MMP-1, MMP-3, YKL-40, leptin, resistin, serum amyloid A and C-reactive protein (CRP). The Vectra DA score represents the level of RA disease activity and can provide objective information that is independent of CRP or conventional, physical exam-based measures.
Research (Abstract 2973) presented at the 2014 ACR/ARHP Annual Meeting included 143 patients in the BRASS registry at Harvard who had received a stable treatment over two years. They were evaluated at a single visit in the BRASS registry for the Vectra DA score and the conventional measures of CRP, DAS28-CRP, CDAI, SDAI and RAPID3. X-rays of their hands and wrists were obtained at the visit and two years later. Vectra DA was a better predictor of radiographic progression over two years when compared with the standard measures based on the exam or patient-reported outcomes. The odds ratio for predicting progression was highest for Vectra DA and lowest for RAPID3.
Eric Sasso, MD, vice president of medical and scientific affairs at Crescendo Bioscience, says the research indicates that a high Vectra DA score can occur even when CRP for conventional clinical measures show low disease activity or even remission. “When those types of discordance occur (i.e., high Vectra score indicating high disease activity in a patient with low CRP or low clinical measure of disease activity), those patients have an increased risk of ongoing joint damage.” These findings, he says, suggest that the Vectra DA test is detecting disease in some patients who appear to be in remission.
Vectra can provide additional information not readily available from conventional assessments, he says, such as when a patient appears to be in remission but actually has a risk of further joint damage. It can also be useful in the opposite kind of situation where the patient has a great deal of pain and appears to be doing poorly. “In some cases, the pain may be noninflammatory and the patient may not have much or any activity of their RA,” Dr. Sasso says. “Vectra DA can provide a completely objective way of assessing the contribution of rheumatoid inflammation.
“The role of Vectra DA is to measure the amount of disease activity. That information can be particularly important when deciding whether to change treatment, and the result can provide assurance as to whether the patient is doing well on current therapy,” Dr. Sasso says. “Vectra DA does not help with the decision about which therapy to select. That remains an area of great interest for ongoing research.”
Dr. Ruderman noted that rheumatologists who use Vectra do so to help them rule or rule out active disease and to quantify active disease. Those who don’t use it, he says, think it is an interesting approach but may not be much better than what they are currently using, such as CRP and DAS28 or a swollen and painful joint exam, which can be less expensive.
According to Dr. Rigby, the greatest application of Vectra is for predicting people with poor outcomes or for predicting that a patient is in low disease activity even when they are still reporting lots of symptoms. The big question in a patient with painful hands but no swelling is whether the patient’s “amplification of pain pathways is talking or whether it is the synovial inflammation talking. Some say Vectra is very good at that. Others would say why not do musculoskeletal ultrasound of that particular joint and confirm the inflammation.”
But of course the problem remains that, even if it is the inflammation talking, there is still no biomarker that can predict the best treatment for the patient whose disease does not respond adequately to methotrexate or even triple therapy.
The search for the holy grail in RA is a complex one, but one worth doing, Dr. Kremer says. “The stakes are so high. People are suffering with a chronic lifelong disease, and these interventions are very expensive. In fact, the Corrona data show that 55–60% of all RA patients in the U.S. are on some type of biologic drug. In an era of shrinking resources for these expensive drugs, it would save society a lot of money if individuals can be identified up front as better or worse candidates for a particular intervention.”
Kathy Holliman is a medical journalist based in Beverly, Mass.
- Robinson WH, Lindstrom TM, Cheung RK, Sokolove J. Mechanistic biomarkers for clinical decision making in rheumatic diseases. Nat Rev Rheumatol. 2013;9:267–276.
- Dennis G Jr., Holweg CTJ, Kummerfield SK, et al. Arthritis Res Ther. 2014;16:R90;1–18.
- Snyder A, Makarov V, Merghoub T, et al. Genetic basis for clinical response to CTLA-4 blockade in Melanoma. N Engl J Med. 2014. Doi:10.1056/NEJMoa1406498.