Background/Purpose:

Rheumatoid arthritis (RA) is among the most intensively studied chronic inflammatory musculoskeletal diseases. Over the past two decades numerous new compounds have been tested in RA. Although the results from any new clinical trial in RA is usually interpreted in the context of existing data from previous trials, it is not clear whether trial populations are necessarily historically comparable. We investigated secular trends in characteristics of patient populations enrolled in RA clinical trials and inclusion criteria as possible influencing factors.

Methods:

We performed a systematic literature review of randomized, controlled, double-blind trials, investigating biological therapies in RA. Reports were identified using PUBMED, EMBASE and the Cochrane Library. Populations were stratified into conventional disease modifying anti-rheumatic drug (DMARD) naïve, DMARD inadequate responders (IR), and biological DMARD IR. The following variables at baseline were extracted from reports: swollen and tender joint counts (SJC, TJC), pain, patient and evaluator global (PGA, EGA), acute phase measures (erythrocyte sedimentation rate, ESR and C-reactive protein, CRP), as well as the Health Assessment Questionnaire Disability Index (HAQ). In addition, we obtained the year of publication and the inclusion criteria (IC) of each trial. We then performed a mixed model meta-regression of year of publication on each of the mentioned variables and IC.

Results:

Out of 697 abstracts selected for screening, 73 studies were chosen as relevant; 3 studies with mixed populations were excluded, resulting in 70 studies included for analysis. Table 1 shows the medians and quartiles for the baseline characteristics. Meta-regression showed a significant decrease of SJC (β=-0.415;p<.001), TJC (β=-0.378;p<.001), and CRP (β=-0.123;p<.001) over the years (figure 1); for all other core set measures, there was no trend or significance. IC showed similar trends over time for SJC (β=-0.154;p<.001), TJC (β=-0.243;p<.001), and CRP (β=-0.065;p<.001). Furthermore, there were significant linear associations of IC with SJC (β=1.33;p<.001), TJC (β=1.14;p<.001) and CRP (β=1.16;p<.001) at baseline.

Conclusion:

There is a progressive drift towards lower number of swollen and tender joints and lower CRP-levels at trial entry of time, which is at least partly related to a similar trend in inclusion criteria for RA. The constancy of patient-reported outcomes suggests that the baseline activity is still perceived as similarly high. Differences in overall baseline inflammatory activity may pose a challenge for comparing newer with older trial results.