Making Use of Precision Medicine for Rheumatic Diseases

The NIH recently commissioned $55 million for studies to investigate precision medicine. While it remains to be seen how many of those dollars will be used in rheumatic diseases, the message is clear: Medicine is moving steadily toward personalized approaches.

Experts who deal with arthritis, lupus and other associated syndromes have taken notice and are embracing precision medicine in the clinic. Researchers across the United States will be collaborating with those from around the world to further understand genetic and environmental factors, the microbiome, novel therapies and novel approaches to optimize currently available therapies in a cross-section of rheumatic conditions.

Despite this mission, there are more answers than questions about precision medicine in rheumatic diseases, according to S. Louis Bridges Jr., MD, PhD, Anna Lois Waters Professor of Medicine, director of the Division of Clinical Immunology and Rheumatology and director of the comprehensive Arthritis, Musculoskeletal, Bone and Autoimmunity Center at the University of Alabama at Birmingham. “The concept of personalized medicine — meaning the presence of assessments can successfully predict which drug will work in an individual patient — in [rheumatoid arthritis] RA has unfortunately not yet come to fruition,” Bridges said. “But, we have made some gains and there is hope for the future.”

Amr H. Sawalha, MD, professor of internal medicine, Marvin and Betty Danto Research Professor of Connective Tissue Research in the Division of Rheumatology and the Department of Internal Medicine & Center for Computational Medicine and Bioinformatics at the University of Michigan, outlined a way forward.

“The single most effective way toward understanding and developing personalized medicine approaches is to conduct research studies in longitudinal cohorts of patients followed individually over time,” he said. “Inception cohorts will be critical. We need to study patients early in the disease process or, even better, before they develop the disease, and then collect follow-up samples. This will be an effective approach to understand these diseases, how they evolve in individual patients, how they can be monitored at the individual patient level and how this information will guide us to the best therapeutic approach based on the biological pathway that is most involved in causing the disease manifestation or disease flare.”

Sawalha added the idea of disease states could come into question. “We have to step away from assuming that, for example, lupus is one disease,” he said. “It is not, and this is why some patients respond and others do not to a particular treatment. In fact, this is perhaps one important reason why most of our clinical trials in lupus have failed.”

With this in mind, Healio Rheumatology explored the way genetic and environmental factors impact one another; the first steps into understanding how the microbiome can impact inflammatory conditions; where drugs are working; the role of government and professional societies in moving the field forward; and how personalized medicine is moving at a faster rate for some diseases than for others.

Genetic Factors

“Our understanding for the genetic basis of rheumatic diseases has exponentially increased in recent years,” Sawalha said. “We know more now about genetic risk loci that predispose to these diseases, but more work is needed to fully explain the heritability of these diseases, and, more importantly, to understand how these genetic risk loci cause or contribute to the disease process.”

Sieberts and colleagues conducted a community-based assessment of the usefulness of single-nucleotide polymorphism (SNP) to predict outcomes for anti-tumor necrosis factor (TNF) therapies. The conducted a model using data from 73 research groups that included 29,880 RA cases and 73,758 controls. Ninety-eight biological candidate genes were found from 1,010 potential risk loci. Investigators found a significant genetic heritability estimate of treatment non-response trait. However, they added there was no observed genetic contribution to prediction accuracy that was deemed to be significant.

“Results formally confirm the expectations of the rheumatology community that SNP information does not significantly improve predictive performance relative to standard clinical traits, thereby justifying a refocusing of future efforts on collection of other data,” they concluded.

“Traditional genetics has not helped personalized approach to anti-TNF treatment of RA, but epigenetics and integrative approaches have not yet been fully explored,” Bridges said.

“Most of the genetic information that we have available to us in rheumatic diseases is still difficult to put into practice,” Timothy B. Niewold, MD, a rheumatologist at the Mayo Clinic in Rochester, Minn., said. “Many associated polymorphisms exert a moderate effect on disease susceptibility by themselves, and interaction models incorporating information from multiple polymorphisms thus far have not been able to provide an improved genetic model of disease.”

Niewold said the heterogeneity of rheumatic diseases is a major obstacle. “Different patients have different pathways to disease, and large studies that combine all patients together would reduce the effects of all polymorphisms associated with subgroups of patients with the disease,” he said. “Also, it is likely that many polymorphisms exert their influence on the immune system, which is an intensely complex organ, and simple models adding up polymorphisms together are not likely to represent biological reality.”

Studies on biologically similar groups of patients, in which pathogenic pathways undergo investigation, have led to advances in understanding, according to Niewold. There also have been studies to define the molecular impact of disease associated polymorphisms. “While this work will be a large undertaking, it has the potential to reveal human disease pathogenesis in ways that were not previously attainable,” Niewold said.

Assassi and colleagues investigated genetic associations with systemic sclerosis. They suggested IRF5, STAT4, BANK1 and BLK are pathways that underwent investigation due to their involvement in immune regulation, while HLA-DQB1, HLA-DPA1/B1, and NOTCH4 associations with systemic scleroderma are likely confined to disease-specific auto-antibodies. Regarding associations with disease severity and organ involvement, CTGF, HGF, IRAK1, IRF5, MMP-12 and SP-B polymorphisms were implicated, particularly with interstitial lung disease. Pulmonary arterial hypertension was associated with IL23R, KCNA5, TLR2, TNAIP3, and UPAR genes, while HLA-DRB1*04:07 and *13:04 carried associations with scleroderma renal crisis. “However, the above findings need to be replicated in independent studies,” the researchers wrote. While this information can help in researching novel agents and targeted therapies, there have yet to be large-scale studies conducted in drugs such as sifalimumab, rituximab and abatacept. Moreover, data examining which patients may be most likely to respond to such therapies are sparse.

“The authors discuss how the genetic basis of scleroderma is being unfolded,” Sawalha said. “Some of the genes discovered are more important in limited or diffuse scleroderma, or seem to be more relevant to specific disease manifestations. After this review, there have been also some reports of epigenetic differences between patients with limited and diffuse scleroderma. It is likely that the combination of genetic and epigenetic loci will help to better develop a more effective and a personalized approach toward understanding and treating scleroderma in the future.”

Significant work remains in terms of the development and validation of these data before these can be applied in the clinic to predict which patients will respond to which therapies, according to Sawalha.

“There is a lot of important in research happening in genetics right now,” Daniel G. Arkfeld, MD, associate professor of Clinical Medicine and director of Rheumatological Education at the Keck Hospital of the University of Southern California, told Healio Rheumatology. “In RA, we are seeing that if there is the shared epitope on chromosome 6, and they smoke, they are much more likely to have a worse course of disease.”

Harris Perlman, PhD, chief of rheumatology, director of flow cytometry and Mabel Greene Myers Professor of Medicine in the Division of Rheumatology at the Northwestern University Feinberg School of Medicine, echoed this sentiment. “There is the genetic information that you are born with, the DNA that tells you that you are predisposed to X, Y and Z,” he said. “But then there are the environmental factors, like stress, that alter the DNA and allow these things to be expressed, which are multifactorial.”

For Arkfeld, the key is this interplay between genetic and environmental factors. “We are seeing underlying genetic signatures for RA and lupus, but they are obviously complicated and we have not fully solved the equation,” he said. “Because of that, as a clinician, I try to do what I can to control the environmental factors. I tell them not to smoke, to control their gingivitis, to do other things that will trigger the disease.”

Microbiome Information

“‘Microbiome’ is a nice buzzword to use these days,” according to Perlman. “We can certainly use it to show environmental cues going on in the body. But it is difficult to use this microbiome information in a general way because it is so specific to the individual patient — to what they eat, where they live, other environmental factors.”

Sawalha agreed. “Work in the microbiome in rheumatic diseases is in its infancy, though it is gaining momentum and important discoveries are being made,” he said. “It will be of particular interest if we can better define our diseases or predict some of the specific disease manifestations by a microbiome signature. We are not there yet, but I can see how this might become possible in the near future.”

Thus far, research into the microbiome has documented specific microbe species that associate with disease and how changes in the community correlate with disease status, according to Niewold. “Most studies to date have examined bowel flora,” he said. “There is still complexity around what factors contribute to the changes in bowel flora, such as medication, diet and disease state. But, these studies are encouraging and they suggest the possibility of modulating the microbiome as a therapeutic strategy. This kind of effort is farther along in infectious diseases of the bowel, such as Clostridium difficile, but this approach represents an exciting frontier in rheumatic disease.”

For Arkfeld, it comes down to controlling what can be done until more information is available. “I still talk to my patients about how their diet can cause inflammation. I tell them to avoid [genetically modified organisms] GMOs,” he said. “This is my way of manipulating the microbiome. I think more sophisticated ways of manipulating it is where we are headed, but we are not there yet.”

Arkfeld also noted models have shown mice raised in a germ-free environment do not develop psoriatic arthritis. “You have to believe that if certain microbes are a key influence in psoriatic arthritis, they will be important in RA and lupus, as well,” he said. “The problem is that this is still theoretical at the moment. We are still figuring out which probiotics might be helpful. A lot of the data are not clear, so it is not of much help in the clinic.”

The interplay between genetics and the microbiome is also something clinicians should be thinking about, according to Arkfeld. “We cannot influence their genes, but we can influence their environment,” he said. “Every day we are telling people to control their environmental factors.”

Closer Look at RA

Muskardin and colleauges studied sera from a test set of 32 patients with RA from the Auto-immune Biomarkers Collaborative Network Consortium and from 92 patients with RA from the Treatment Efficacy and Toxicity in Rheumatoid Arthritis Database and Repository Registry to predict response to biological agents. Patients in the test set had a good response or no response to TNF inhibitors at 14 weeks, while the validation set included patients with good, moderate or no response at the 12-week mark. Results indicated an increased ratio of interferon [IFN]-beta to IFN-alpha in serum before treatment was associated with a non-response to anti-TNF therapy. The researchers noted neither anti-cyclic citrullinated peptide (anti-CCP) antibody titer nor class of TNF therapy impacted this association. The optimal cutoff for a receiver-operator curve ratio was 1.3, according to the findings. Results from the validation set indicated a ratio of IFN-beta to IFN-alpha greater than 1.3 was associated with a significant risk for a non-response compared to a good response. The test yielded a 77% specificity and a 45% sensitivity for predicting no response vs. moderate or good response, according to the findings. The predictive capacity of the ratio was confirmed in a meta-analysis. “This study supports further investigation of serum type 1 IFN in predicting outcome of TNF inhibition in RA,” the researchers concluded.

“This study addresses a major gap in our understanding of RA treatment,” Niewold said. “Responses to all treatments are variable and there are always non-responders, but we do not know how to identify the non-responders prior to treatment. So we treat, and expect a certain number of non-responders, but for the individual patient we cannot tell until we try.”

The researchers examined blood samples from RA patients taken before treatment with TNF inhibitors. “We found circulating levels of type 1 interferon could predict non-response to anti-TNF treatment,” Niewold said. “This was tested in one group of patients and then validated in a second independent set of patients. We are hopeful this kind of testing may allow for more personalized treatment decisions, which would reduce the burden of unnecessary treatment for those who would not have responded in the first place.”

Nair and colleagues aimed to develop a treatment algorithm for predicting clinical response to RA therapies using measurement of MRP8/14 serum levels. They measured these levels in 170 patients who initiated therapy with infliximab, adalimumab or rituximab and developed a predictive score for 16-week clinical response. Higher baseline MRP8/14 complex levels were associated with the probability of response. In addition, the probability of response with higher baseline MRP8/14 complex levels was different between TNF-blockers and rituximab. Higher DAS28 at baseline also improved the probability of response. Decreases in the probability of response were reported for rheumatoid factor positivity, functional disability and previous use of a TNF inhibitor. Results of the treatment algorithm indicated 80 patients would have been recommended for therapy with a TNF inhibitor. Eight patients would have been recommended for rituximab and 13 patients would have been recommended for another biological agent. There was no treatment recommendation made for 69 patients. The researchers observed a relationship between the predicted and observed responses.

“On group level the predicted response based on the algorithm resulted in a modest 10% higher response rate in our cohort with much higher differences in response probability in individual patients treated contrary to treatment recommendation,” the researchers wrote. “Prediction of response using MRP8/14 levels along with clinical predictors has potential in personalizing treatment for RA patients starting biological anti-rheumatic treatment, and might increase cost-effectiveness.”

“Although we do not have a commercially available test to tell which RA patients might have a good response, or a poor response, we are making some progress,” Bridges said.

Strides in SLE

Sanchez and colleagues examined 16 potential genetic loci in a cohort of 4,001 individuals of European heritage, 1,547 of Hispanic heritage, 1,590 African Americans and 1,191 Asian Americans with systemic lupus erythematosus (SLE). Researchers observed significant associations with clinical manifestations of disease in the FCGR2A, ITGAM, STAT4, TNSF4 and IL21 genes. “The findings suggest genetic profiling might be a useful tool to predict disease manifestations in lupus patients in the future,” they concluded.

“We have initial data for genetic risk loci that might predispose to a particular disease manifestation in some of our complex diseases, some data to suggest specific genetic variants associated with response to particular therapy and epigenetic marks that might distinguish between specific manifestations, for example the presence of renal involvement in lupus,” Sawalha said.

“This is still in early stages, but there has been some recent progress in personalizing approaches to SLE,” Niewold added. “Antibodies targeting the type 1 interferon pathway have been used in phase 2 trials, and the results so far support the idea that baseline pre-treatment interferon levels predict response to agents that block this pathway. In the case of rontalizumab (Genentech), the low interferon group showed a stronger response to treatment as compared to the high interferon group, which seemed somewhat paradoxical.”

Niewold said sifalimumab (MedImmune) and anifrolumab (AstraZeneca) had a better response in SLE patients with higher pre-treatment interferon levels compared to SLE patients with lower pre-treatment type 1 interferon levels.

“Overall, while the results differ in direction between the agents to some degree, pre-treatment interferon levels were important to response with each of these agents,” Niewold. “These data taken together suggest assessment of pre-treatment interferon levels may be the future in SLE, at least when considering blocking that pathway for therapeutic effect.”

Current Therapies

In 2009, Burgos and colleagues conducted a comprehensive review of clinical and environmental factors, along with laboratory parameters and genetic markers in rheumatic disease. Studies investigating the Health Assessment Questionnaire and non-response to disease-modified antirheumatic drugs (DMARDs) underwent analysis, as did those looking at smoking, lower educational level, low socioeconomic status, rheumatoid factor, positivity for anti-CCP antibodies, erythrocyte sedimentation rate or C-reactive protein levels, and the presence of erosions in baseline assessment.

“The knowledge of the severity profile in RA patients, based on specific predictors, is the first step to make a therapeutic decision; in addition, it is necessary to determine the predictors of efficacy and toxicity to both traditional DMARDs and biologic agents,” they wrote. “All medications have the potential to produce side effects; the ability to identify those patients who will benefit the most with biologics is imperative, given the high cost of these medications, their increased use and the development of new compounds aimed at specific target molecules in the pathway of RA.”

“This is a nice review summarizing some of the most important developments that will help utilize a personalized medicine approach in patients with rheumatoid arthritis,” Sawalha said. “The authors also discuss some of the limitations of these studies, and suggest larger studies take into account confounding factors, such as baseline therapy and race/ethnicity, will be needed before conclusions can be useful in the clinic are possible. As genotyping will be critical to provide information regarding personalized medicine approaches, ethical issues surrounding genetic data access and storage, and financial aspects will be also important to consider.”

Kiely and colleagues built on these findings in a review of the factors that influence outcomes in patients with RA treated with biologic therapies. They suggested patients with higher BMI, including those treated with TNF inhibitors, may have poorer outcomes in RA. However, adiposity has less impact on outcomes with abatacept and tocilizumab. Although smoking impacts outcomes with TNF inhibition, there is less of an impact on rituximab and tocilizumab (Actemra, Genentech), according to the researchers. They added it is unknown how smoking effects abatacept (Orencia, Bristol-Myers Squibb). Anti-citrullinated protein antibodies (ACPA) and RF positivity yields improved responses with rituximab (Rituxan, Genentech) and abatacept, while serotype shows less of an impact on tocilizumab, according to the findings. It is also suggested the association between ACPA and RF is “more complex,” they wrote. Methotrexate appears to improve the performance of all biologics, according to the researchers.

The researchers wrote that monitoring TNF inhibitors is “an exciting new field,” particularly in RA and psoriatic arthritis, and BMI, adherence to therapy, use with DMARDs and anti-drug antibodies are factors that are part of this field. Despite the advances in understanding, Perlman believes there is still work to be done.

“We still do not fully understand why we give which drugs [and] which patients are more susceptible to which therapies,” he said. “Right now, it is still much a guessing game. There is a lot of trial and error. Patients have to go through 12 weeks to a year or more to find out which drugs work and which do not. We need to do better than that.”

Synovial Biopsy

For Perlman, ultrasound-guided synovial biopsies may be a key to the future of personalized medicine in rheumatic disease. “You can get good data with few side effects, it requires only a small tissue sample and a small gauge needle,” he said.

It is critical to get the word out on this therapy, Perlman said. “When people hear the word ‘biopsy,’ they get worried because of the association with cancer,” he said. “But we tell them it is a 30-minute procedure that requires just a Band-Aid when it is finished. We tell them we can take a sample from a knuckle, a knee, wherever, with no pain. We put lidocaine in the joint. It is simple and quick. We need to get patients on board. We have literature for them. We are trying to spread information every day.”

The advantage of synovial biopsy is it allows clinicians to look at the joint, and not just a protein, according to Perlman. “We are trying to figure out a core signature of multiple genes and pathways to determine who will respond to therapy and who will not,” he said. “We are building on methods that we have seen in oncology.”

He noted clinicians in Europe have moved ahead of those in the United States with this technology. “They are doing it in RA patients before and after therapy,” he said. “But it is becoming more common here. Many groups now have someone doing these injections. In our group, we are seeing 15 patients a day.”

Perlman said this may be the route to the next wave of truly personalized medicine, but it is important to keep larger goals in mind, as well. “The idea of precision medicine, among other things, is to understand why only 50% of patients respond to therapy,” he said. “What this translates into is $5 billion a year wasted for ineffective therapy. Of course, we want to improve therapy for patients, but we also want to reduce the burden on the system. This approach will allow us to target therapies more effectively.”

Role of Professional Societies

The NIH is “on board” with synovial biopsy, according to Perlman. “They are leading a training center in this approach,” he said.

This, then, raises the question of the role of professional societies in advancing precision medicine in rheumatic disease. The NIH study will include genomic and biological data, health history and status information, a range of environmental and lifestyle factors, and blood and urine tests to monitor ongoing health outcomes. The study will involve a host of strategic and operational partnerships and include funds devoted to expanding health care infrastructure, according to a NIH press release.

“This range of information at the scale of 1 million people from all walks of life will be an unprecedented resource for researchers working to understand all of the factors that influence health and disease,” NIH Director Francis S. Collins, MD, PhD, said in a press release.

“From the website, I do not see that arthritis or rheumatic diseases will specifically be a big focus of this,” Bridges said. “However, because there are a lot of people with arthritis, including osteoarthritis, RA, etc., it might have an impact eventually.”

The FDA has issued two draft guidances pertaining to the use of next generation genomic sequencing. The first offers recommendations for the development and validation of next generation tests for hereditary diseases, while the second pertains to approach allowing developers of such tests to use clinical evidence from FDA-recognized public genome databases to support clinical claims for their tests, according to an FDA press release.

Perlman said professional societies have offered more targeted help in terms of precision medicine. “The ACR Research Foundation has been amazing. [It is] the backbone for research in rheumatic disease,” he said. “They have put funds toward a number of projects, including those in precision medicine. They are willing to offer grants to people who have great preliminary data and want to try to explore further. They are re-entering this game.”

As more data comes out, there will be more opportunities to reach out to patients about precision medicine programs, such as those using synovial biopsy, according to Perlman. Management of these patients will come down to the individual doctor-patient relationship. “We still need to determine what components are feasible,” he said.

Looking Forward

Perlman suggested the way clinical trials are conducted may demand fresh thought. “The issue we have now is a plethora of drugs, but if you look at the clinical trials, they all have the same endpoints: ACR20, 50, 70,” he said. “But these drugs are all different. Anti-TNF inhibitors, Janus kinase inhibitors — these drugs have different mechanisms, and maybe we need to think about different outcome measures.”

Another issue is many experts are seeking a simple blood test to predict response. “It is provincial to think that this is going to predict who will respond or not,” he said. “What is happening in the blood does not tell you what is happening in the patient. What is happening in the joint will tell us a lot more. Go to the joint, that will predict who responds to therapy.”

For Bridges, incorporating all data is the key. “Focusing on gene expression in the form of transcriptomics or possibly microbiome, or using multiple sets of diverse data may be helpful,” he said.

Sawalha agreed. “Understanding environmental triggers and how they interact with genetic and epigenetic disease risk loci will be of particular interest to understand disease mechanisms,” he said.

Arkfeld said there is still ground to be covered in this regard, particularly with genetics. “We are gaining information that helps us conceptualize how to treat patients, but we do not have a formula yet where if you have a certain marker, you get this medication,” he said. “It is not yet an evidence based approach.”

He offered some practical advice. “It may be easier to improve our use of biomarkers not in terms of which initial therapy to choose, but in terms of changing therapies,” he said. “They can help decide whether to switch from anti-TNF to T-cell therapy or when to try triple therapy.”

There will be challenges, but Arkfeld remains optimistic. “Past research has shown a lot of confusing data,” he said. “But if we remember that not all patients are the same, I think we can move forward. I am always hopeful that the next discovery is just around the corner.” – by Rob Volansky

• References:
• Assassi S, et al. BMC Medicine. 2013;doi:10.1186/1741-7015-11-9.
• Burgos PI, et al. Ther Adv Musculoskelet Dis. 2009;doi:10.1177/1759720X09351778.
• Kiely PD. Rheumatology (Oxford). 2016;doi:10.1093/rheumatology/kev356.
• Muskardin TW, et al. BMJ. 2016;doi:10.1136/annrheumdis-2015-208001.
• Nair SC, et al. Plos. 2016;doi:10.1371/journal.pone.0152362.
• Sanchez E, et al. Ann Rheum Dis. 2011;doi:10.1136/ard.2011.154104.
• Sieberts SK, et al. Nature Comm. 2016;doi:10.1038/ncomms12460.
• www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm509814.htm?source=govdelivery&utm_medium=email&utm_source=govdelivery

• For more information:
• Daniel G. Arkfeld, MD, can be reached at 1500 San Pablo St., Los Angeles, CA 90033, email: meg.aldrich@med.usc.edu.
• S. Louis Bridges Jr., MD, PhD, can be reached at 1720 Second Ave. South, SHEL 178, Birmingham, AL 35294; email: lbridges@uab.edu.
• Timothy B. Niewold, MD, can be reached at 200 First St. SW, Rochester, MN 55905; email: theimer.sharon@mayo.edu.
• Harris Perlman, PhD, can be reached at 240 E. Huron St., McGaw M314, Chicago, IL 60611; email: h-perlman@northwestern.edu.
• Amr H. Sawalha, MD, can be reached at 5520 MSRB-1, SPC 5680, 1150 W. Medical Center Dr. Ann Arbor, MI 48109; email: asawalha@med.umich.edu.

Disclosures: Arkfeld, Bridges, Perlman and Sawalha report no relevant financial disclosures. Niewold reports he receives research grants from EMD Serono and Janssen.