Over the past 20 years, many randomized controlled trials have shown that biologic agents may result in slowed disease progression and improved physical function in patients with rheumatoid arthritis (RA). In clinical rheumatology practice, it’s estimated that 40–70% of patients have been exposed to a biologic.
“However, randomized controlled trials for biologic agents in RA have rigorous and non-standardized entry criteria, which may affect the external validity of generated findings and, thus, limit the generalizability to routine clinical practice.” write Priyanka Vashisht, MD, and colleagues from the VA Nebraska-Western Iowa Health Care System and Nebraska Arthritis Outcomes Research Center, University of Nebraska Medical Center in Omaha. “Therefore, it is critical to evaluate how study populations are assembled for randomized controlled trials in order to determine whether routine clinical patients differ in a meaningful way from those who participate in trials.”
The research, published in the October 2016 issue of Arthritis Care & Research, reviewed the eligibility criteria from 30 regulatory randomized controlled trials of 10 U.S. Food and Drug Administration-approved biologic agents. Additionally, data for each biologic agent were grouped into three categories: tumor necrosis factor (TNF) inhibitors, non-TNF inhibitors and an oral Janus kinase inhibitor.
The inclusion criteria were then applied to patients enrolled in two large, observational clinical cohorts for RA: the Veterans Affairs Rheumatoid Arthritis registry (VARA; n=1,523) and the Rheumatology and Arthritis Investigational Network Database (RAIN-DB; n=1,548). Based on data from their baseline entry into the registries, patients were assessed across three domains of eligibility: RA disease activity, demographics and medication exposure.
The Results
“We demonstrated that the vast majority of RA patients from our clinical cohorts (96.3% in VARA and 92.4% in RAIN-DB) did not satisfy criteria for participation in biologic agent randomized controlled trials,” write the authors in their discussion.
Only 3.7% of patients in VARA and 7.1% in RAIN-DB were eligible for inclusion based on disease activity alone.
For both cohorts, only 28-joint counts (tender and swollen) were available to calculate the Disease Activity Score (DAS28), along with the erythrocyte sedimentation rate and patient global well-being score components. In contrast, a majority of the study’s randomized controlled trials defined minimum thresholds of disease activity with joint counts of 68 possible tender and 66 possible swollen joints. As a result, “the domain most often attributable to patient exclusions in both cohorts was disease activity, specifically the minimum number of swollen and/or tender joints required,” write the authors. “The mean (range) DAS28 score was 6.59 (6.1–7.1) across the 19 randomized controlled trials groups at enrollment, in comparison to 3.8761.56 in VARA and 3.6561.33 in RAIN-DB.”
For demographics, the most common reason for trial exclusion was patients exceeding the disease duration maximum. On average, this aspect affected 10% of VARA patients and 2.5% of RAIN-DB patients across 20 of the trials. Additionally, a mean of 72% of VARA patients and 29% of RAIN-DB patients were excluded due to previous medication use. The authors write, “Prior methotrexate use, which excluded, on average, 16% in VARA and 6.8% in RAIN-DB, and the use of other disease-modifying anti-rheumatic drugs, which excluded, on average, 62% in VARA and 27% in RAIN-DB, were most striking.”
The authors note the importance of the aggressive treatment of RA, and in clinical practice, patients with fewer disease-affected joints are offered biologic agents as first-line therapy and/or when traditional disease-modifying anti-rheumatic drugs are not effective.
“These findings underscore the need for caution in extrapolating trial results to day-to-day management of RA patients,” write the authors. “Patients recruited into these studies commonly have much higher levels of disease activity than seen in clinical practice, and comparator arms may include placebo. … However, when eligibility criteria are overly restrictive, [the criteria] can compromise external validity and sacrifice the generalizability of the results to the patient populations in whom the answer to the question is desired.”
The authors note the value of observational data for putting the results of randomized clinical trials in perspective. “There is a need for more real world/observational studies to guide prescribing protocols and provide comparative effectiveness of the many medications in treating patients with RA in clinical practice,” they conclude.
Vashisht P, Sayles H, Cannella AC, et al. Generalizability of patients with rheumatoid arthritis in biologic agent clinical trials. Arthritis Care Res (Hoboken). 68(10):1478–1488. doi: 10.1002/acr.22860.
