This method of matching drugs to patient starts with RA but may eventually be used for many other conditions
In 1800, Augustin Jacob Landre Beauvails, a doctor at the famed Salpetriere Hospital in Paris, described what is now known as rheumatoid arthritis (RA). But he did not name it. That was left to Alfred Bring Garrod, a British rheumatologist who called it RA because it has some of the characteristics of rheumatic fever, a totally different disease.
Virtually everyone reading this article knows someone with this dreaded disease, which affects more than 1.3 million people in the United States. Mediated by an overreaction of the immune system, RA can affect virtually every organ system and shorten one’s lifespan. RA targets the synovium, causing creation of a toxic joint fluid that leads to degradation of the articular cartilage and bone and eventually to crippling joint damage.
Scoring systems to assess RA consist of measures of disease activity involving the joints as well as patient-reported activity and laboratory findings. Among the instruments used to measure RA disease activity are the Patient Activity Scale (PAS); the Routine Assessment of Patient Index Data (RAPID); the Clinical Disease Activity Index (CDAI); the Simplified Disease Activity Index (SDAI); the ACR20, ACR50, and ACR70; and the Disease Activity Score in 28 Joints (DAS 28).
These instruments are used in varying degrees as primary and secondary endpoints in clinical trials and to determine remission and treatment modifications. In addition to these physical exam metrics and patient-reported data, physicians have relied on a number of individual blood tests, including ESR, rheumatoid factor, CRP, and others. Research has demonstrated that tight control of RA leads to better long-term outcomes. Determining when tight control has been achieved when many of the measures are subjective has led to a search for a true biomarker similar to a HbA1c for diabetes. To date, there has been no gold standard.
But a biomarker test — actually a multibiomarker disease activity (MBDA) test for RA — might finally have become a reality through the work of a small South San Francisco company, Crescendo Bioscience. The search for this MBDA started with a hypothesis: Measurements of multiple serum protein biomarkers combined via a mathematical algorithm into a single score could “quantitatively and objectively characterize RA disease activity and enhance current disease activity assessment.” The developers further hypothesized that “periodic monitoring of this score could complement existing approaches to patient care, facilitating quantitative tracking of patient status and treatment.”
A multistage approach was used to develop the MBDA. The first stage consisted of a literature-based search for potential candidate biomarkers and the optimization of corresponding assays. This manual review was based on literature found using PubMed and in bioinformatics databases.
Stage 2 included two steps: clinical studies to “assess and prioritize the biomarkers based on their relationship to clinical disease activity” followed by a study to verify that the MBDA would reflect joint findings. Stage 3 further optimized the assays and biomarker selection and final algorithm.
To develop the MBDA, stored serum samples were screened. They had been obtained from patients in numerous clinical trials and cohorts representing a diverse set of patients from across the United States and Western Europe who were being treated according to the current standard of care.
Patients had to have at least 4 of the 7 criteria for RA in the 1987 revised American College of Rheumatology criteria for the diagnosis of RA. Exclusions included using high-dose oral corticosteroids, pregnancy, nursing, organ failure of numerous types, history of demyelinating disease, HIV, malignancy or active infection, or substance abuse. All have the potential to influence the inflammatory markers.
One hundred thirty candidate biomarkers were identified. Each stage sequentially eliminated candidates until 25 were selected for algorithm training and development. The final 12 biomarkers were selected from development and then validated in independent patient cohorts. No MBDA would be of use if it did not correlate with disease activity.
Hence, the development involved stratifying patients to low, moderate, and high disease activity levels based on the routine scoring for RA, which considers: DAS28-CRP, DAS 28-ESR, CDAI, swollen joint count, tender joint count, and patient global assessment (PGA). Statistical testing was a foundation for the development, but the testing is beyond the scope of this article.
While the biomarkers were chosen based on a statistics-based process, they also reflect key biological pathways activated in RA.
In addition, the development ensured that there was “no direct interference due to interaction from RA drugs, heterophilic antibodies such as RF, or other substances.”
Various methods of statistical analysis resulted in a final algorithm using different combinations and weights of the 12 biomarkers in separate linear models for each of the clinical components of the DAS28-CRP (TJC28, SJC, Patient Global Assessment).
Several publications have demonstrated that the MBDA is, in fact, significantly associated with the disease activity in RA (both seropositive and seronegative patients). The simple scoring system offers a unique benefit not only to patients and their physicians but also to managed care organizations. It has been shown to quickly distinguish clinical responders from non-responders. How might this test be used?
The drugs used for treating RA can be divided into small (and relatively inexpensive) DMARD molecules and biologic (and considerably more expensive) DMARD molecules. The use of biologic agents has skyrocketed since the first anti-TNF agent was approved by the FDA in 1998. Biologic agents focus on several distinct pathophysiologic processes and each has efficacy in a portion of patients treated, but it is hard to know objectively which people respond and how well they are doing.
The clinical response of individuals with RA is not uniformly measured in a comparative way across populations. If we use diabetes as an example, the HEDIS measures have significantly improved the way we can comparatively measure control of this disease among different health plans over time. HbA1c, lipid levels, and use of preventive screening are now part of the fabric of diabetes treatment.
Can an RA MBDA provide the same level of comparability for measuring RA? Will it lead to more rapid termination of “good enough” treatment in search of an optimized outcome? Will the widespread use of this MBDA prevent or delay the development of joint destruction? Will it allow clinicians to slowly withdraw expensive medication once a “remission” is achieved?
Perhaps some of these questions prompted Medicare to approve payment for this novel approach to the monitoring of this crippling disease a few weeks ago. But this approach is not going to be limited to RA.
We will soon be in a world where many chronic diseases are measured not by one biomarker but by a validated, reproducible array of biomarkers. Such is the world of Tomorrow’s Medicine.
The author is a director in the value-based health department at Genentech. He has had no other industry affiliations in the past three years. The views expressed in Tomorrow’s Medicine are the author’s alone.