Patients with rheumatoid arthritis who would have been excluded from phase 3 randomized controlled trials still achieved comparable adjusted outcomes with biologic or targeted synthetic disease-modifying antirheumatic drugs in routine care, although between-drug comparisons were more sensitive to patient composition.
Researchers analyzed 5,462 patients in Japan’s FIRST Registry who initiated biologic or targeted synthetic disease-modifying antirheumatic drugs between 2003 and 2023. Trial ineligibility was defined using exclusion criteria that were consistently measurable in the registry, including prednisolone use of 10 mg/day or more, renal impairment, hepatic dysfunction, anemia, lymphopenia, thrombocytopenia, and low white blood cell count.
Overall, 40% of patients met at least 1 exclusion criterion. The most common were anemia, reported in 16% of patients, and lymphopenia, reported in 15%. Patients meeting exclusion criteria had higher baseline disease activity, greater functional disability, higher inflammatory marker levels, lower methotrexate use, and greater glucocorticoid use compared with patients who met no exclusion criteria. Glucocorticoids were used by 39% of patients meeting exclusion criteria vs 17% of those meeting no criteria.
At 26 weeks, unadjusted Clinical Disease Activity Index remission, defined as a score of 2.8 or lower, occurred in 25% of patients who met no exclusion criteria and 21% of those who met at least 1 criterion. Low disease activity, defined as a score of 10 or lower, occurred in 56% and 48%, respectively.
The researchers used nonresponder imputation for missing 6-month Clinical Disease Activity Index data, including discontinuations related to insufficient efficacy or adverse events, meaning those patients were counted as not achieving remission or low disease activity.
Following propensity score adjustment, however, remission rates were similar between groups: 24% among patients who met no exclusion criteria and 23% among those who met at least 1 criterion. Adjusted low disease activity rates were 54% and 51%, respectively, and treatment persistence did not differ statistically.
The researchers also found that trial-ineligible patients were unevenly distributed across treatment classes. They accounted for 47% of patients receiving interleukin-6 inhibitors, 45% of those receiving CTLA4-Ig, 36% of those receiving tumor necrosis factor inhibitors, and 31% of those receiving Janus kinase inhibitors.
The researchers suggested this pattern may reflect safety-driven prescribing in routine care, with interleukin-6 inhibitors and CTLA4-Ig more often selected among patients with anemia or organ dysfunction, and Janus kinase inhibitors more often prescribed to patients with fewer trial-exclusion features.
To reduce confounding related to calendar time, between-drug comparisons were limited to patients who initiated therapy in 2013 or later, when all evaluated drug classes were available in Japan.
Among patients who met no exclusion criteria, Janus kinase inhibitors were associated with higher remission rates than tumor necrosis factor inhibitors following adjustment. Remission occurred in 39% of patients receiving Janus kinase inhibitors vs 32% of those receiving tumor necrosis factor inhibitors. Low disease activity occurred in 68% vs 51%, and treatment persistence was higher with Janus kinase inhibitors.
Among patients who met exclusion criteria, remission still numerically favored Janus kinase inhibitors vs tumor necrosis factor inhibitors, at 34% vs 29%, but the difference was not statistically significant. Low disease activity, however, remained statistically significantly higher with Janus kinase inhibitors, at 63% vs 51%.
A similar pattern was observed in comparisons of Janus kinase inhibitors vs interleukin-6 inhibitors. Among patients who met no exclusion criteria, remission occurred in 32% of those receiving Janus kinase inhibitors vs 23% of those receiving interleukin-6 inhibitors. Among patients who met exclusion criteria, remission occurred in 26% vs 22%, a difference that was not statistically significant; however, low disease activity remained numerically higher with Janus kinase inhibitors.
The researchers noted that the attenuated remission differences in trial-ineligible patients should be interpreted cautiously because subgroup precision was limited, particularly among patients receiving Janus kinase inhibitors. The point estimates continued to favor Janus kinase inhibitors, and the findings should not be read as evidence that comparative effectiveness disappeared in more clinically complex patients.
“Between-drug comparative effectiveness estimates were attenuated in the RCT-ineligible cohort relative to the eligible cohort,” wrote lead study author Satoshi Kubo, of the University of Occupational and Environmental Health in Kitakyushu, Japan, and colleagues.
The study was limited by its observational design, and residual confounding could not be excluded despite propensity score adjustment. The operational definition of trial ineligibility was also limited to variables consistently captured in the registry and did not include some trial-related exclusions, including psychiatric disorders, pregnancy intention, language-related barriers, or some comorbidities.
The findings may also have limited generalizability outside Japan, where methotrexate dosing practices and patient characteristics may differ.
“In observational comparisons of molecular targeted therapies, the differential distribution of trial exclusion criteria across drug classes should be recognised as a potential source of channelling bias,” the researchers concluded.
Disclosures: The study was supported by JSPS KAKENHI, the Japan College of Rheumatology, and the Mochida Memorial Foundation for Medical and Pharmaceutical Research. Several researchers reported speaking fees, consulting fees, or research funding from pharmaceutical companies.
Source: RMD Open
