In the design of any clinical trial, issues of scientific integrity and rigor are balanced against issues of safety, feasibility and logistical applicability. The RAPPORT investigators recognized that some degree of scientific rigor was sacrificed in exchange for enhanced feasibility (improved enrollment and retention of subjects) and safety by allowing variability in corticosteroid dose during the efficacy phase of the study. This approach permitted subjects on higher doses of corticosteroids to be included in the study, in contrast to other JIA trials, as their corticosteroid doses could be decreased by protocol if they responded to study medication, rather than requiring stable higher dose corticosteroids for longer than clinically indicated. This approach also better approximated “real life” challenges of caring for children with SJIA. Additionally, a standardized rate of taper was developed for study drug responders, in order to highlight differences in efficacy between the two treatment arms (those who started active rilonacept initially and those who started active drug 4 weeks later). Because increasing or starting corticosteroids would result in the subject being classified as a non-responder, the panel recognized that the frequency of starting or increasing corticosteroids should be standardized to avoid inadvertently and adversely affecting the retention of subjects in the efficacy phase of the study, while ensuring subject safety.
One key element that improved the ability to reach consensus was the presence of experts present who responded to panel questions and concerns quickly and authoritatively. Specifically, the availability of an expert in the trial design (BF) providing comprehensive information about the impact of starting or increasing corticosteroids on study performance, as well as an expert on the diagnosis of MAS (AG) who shared new data concerning the use of ferritin to distinguish MAS from SJIA flare[17, 18], greatly enhanced the consensus discussions and ensured scientific rigor. These data, which are now published, showed that in 5/16 SJIA patients with elevated MAS biomarkers (sIL2ra and sCD163), the median ferritin level was 2,950 ng/ml the median level in 11 SJIA patients with normal MAS biomarkers was 179 ng/ml. This suggests that ferritin levels can distinguish MAS from SJIA flare to some degree. Experienced group leaders (EG, CW) ensured that the consensus process was robust and not affected by negative behaviors that can detract from group decision making, such as group-think, effects of status or style, obedience, compliance, conformity, group polarization or minority influence[13, 14]. Having site PIs participate in the consensus process as the expert panel enhanced study engagement and “buy-in” regarding the final protocol.
While the modified Delphi NGT process performed well for the majority of the decisions, two major areas required discussion by the entire panel since there was the lack of agreement between the two formal nominal groups. These areas were 1) the determination of criteria for incomplete MAS, as there was little available information on which to base consensus; and 2) determination of criteria for corticosteroid taper. Revisiting the principles of the trial design as a combined group permitted consensus to be achieved. Another unplanned but valuable outcome of this process was development of the first criteria for diagnosis of incomplete MAS, which will require further study and validation.
The expert panel successfully achieved consensus on most items considered despite the highly variable and individualized approach to corticosteroid treatment for SJIA. The consensus resulted in the design of algorithms to standardize changes in corticosteroid dosing that was successfully implemented in RAPPORT and positively accepted by the investigators. The algorithms were designed to be optimal for the RAPPORT clinical trial only and not intended as a guideline for clinical care of children with SJIA outside of the trial setting. In fact, Childhood Arthritis and Rheumatology Research Alliance (CARRA) investigators have recently developed consensus treatment plans for SJIA to be used in clinical practice. While these plans include recommendations on dosing of corticosteroids, they do not suggest specific criteria for corticosteroid dose changes. It may be valuable to combine the CARRA guidelines and the RAPPORT algorithms for future rigorous study.
Although RAPPORT enrollment is complete, the results of the study are not yet available to fully evaluate whether development and implementation of corticosteroid algorithms ultimately had a beneficial impact on enrollment or retention of subjects, or the projected impact on study results. However, near universal positive feedback, as assessed on monthly conference calls with site PIs and research coordinators, and an interim analysis of data from 55 subjects identified only 6 errors in applying the algorithm out of 622 monitored visits. This suggest that the use of consensus methodology to drive research protocol development in controversial areas may be a worthwhile model for other studies, particularly pragmatic trials.