Models for Evidence Generation During the COVID-19 Pandemic: New Opportunities for Clinical Trials in Cardiovascular Medicine

Abstract

Much has been made of the “unprecedented” challenges that the coronavirus disease 2019 (COVID-19) pandemic presented to established models for clinical evidence generation. The pandemic necessitated rapid evidence evaluation and portrayed a disease manifesting a wide spectrum of illness severity and subtypes, as well as one for which its natural history and care evolved rapidly over time, and for which treatments appeared to differ in efficacy on the basis of both illness severity as well as concurrent treatments. However, a countervailing viewpoint may be that such challenges were more “unprecedented” in their time course than in their inherent nature. The pandemic pushed the clinical research enterprise into full throttle. As it did, the clinical community sped through an accelerated, but familiar, sequence of evidence dearth, equipoise debates, expert recommendation–dominated practice guidance, conflicting observational studies, and a substantial number of underpowered and overlapping randomized clinical trials (RCT) alongside a much more modest number of larger ones. Notwithstanding substantial effort, 2 years later, the disease had changed considerably, leaving uncertainty as to how to even apply relatively more convincing evidence generated earlier in the pandemic. Inefficiency in the evidence generation process predates the pandemic. Similarly, common cardiovascular conditions such as atherosclerotic cardiovascular disease are also characterized by considerable change through time—in pathobiology, epidemiology, treatment, and outcomes 1 —challenging the evidence evaluation cycle to keep pace with ever-changing diseases. Recognizing that many of these “unprecedented” challenges were, in fact, already endemic in cardiovascular research, the pandemic may present opportunities to consider new approaches to clinical evidence evaluation. Several pandemic efforts did lead to practice-changing clinical evidence; notably among these were adaptive platform RCTs. In a previous Viewpoint, 2 we presented a primer on adaptive platform RCTs, including uses and methodologic considerations. The current article discusses operational considerations related to future application of these designs in cardiovascular medicine, and what an architecture for a globally integrated cardiovascular platform RCT ecosystem may look like one day. While the Viewpoint is forward-looking, signs of movement toward such an ecosystem are already evident in critical care, oncology, and elsewhere. With increasing interest in learning health systems, pragmatic RCTs, and enhanced global integration and harmonization in clinical research, platforms RCTs may come to represent important pistons in the engine driving cardiovascular evidence generation. Platform RCTs employ master protocols that facilitate recruitment of patients with a single disease or syndrome. 3 Participants are screened for eligibility for at least one of often multiple investigational treatments available on the platform, randomized to one or more treatments, and followed for common outcome measures. Beyond statistic efficiencies, 2 platform RCTs also afford important operational efficiencies. Platform trials allow participants to be concurrently randomized to multiple treatments, creating efficiencies via common clinical data collection and harmonized follow-up, and allowing for treatment interactions to be evaluated. Concurrent randomization may offer advantages both to trial participants, who may receive more than one potentially effective treatment, and to patients outside of the trial, as evidence is generated more quickly. Platform trials run longitudinally. This permits the potential use of nonconcurrent controls to increase efficiency. Trials must account for potential secular changes in patient and disease characteristics over time, which could otherwise bias treatment estimation. Statistic adjustment for time may address this. However, some have questioned whether bias can be completely removed by adjustment, 4 and therefore, some platform trials may choose to employ only concurrent control comparisons. Common data collection and harmonized follow-up reduces trial costs. After the platform infrastructure is built, it may host the evaluation of multiple interventions. The current approach of building the required infrastructure for RCTs one at a time has been likened to building a new sports stadium each time a new match is played; by contrast, platform trials build a stadium that hosts multiple matches. Platform trials may host both phase 2 and 3 comparisons, and may be particularly well-suited for comparative effectiveness research. The RECOVERY trial (Randomised Evaluation of COVID-19 Therapy) exemplified the latter, employing simple inclusion criteria, repurposed investigational treatments, and a primary end point (28-day mortality) that was ascertained through administrative healthcare linkage. The RECOVERY trial was able to enroll approximately 10% of the entire population of the United Kingdom hospitalized with noncritically ill COVID-19. By contrast, an enhanced level of data collection, site monitoring, and outcome collection may be required for investigational new drugs seeking registration, such as would be pursued by pharmaceutic companies performing phase 2 and 3 registration pathway trials. Platform trials could also host such registration-pathway evaluations by providing an integrated site network and operational infrastructure to enable efficient patient recruitment, with tailored extension of data capture, monitoring, and outcomes ascertainment as required. This latter infrastructure may be conceptualized as being a “trial platform,” in addition to a being “platform trial” (Figure). This hybridized approach may offer significant cost savings for both public and industry funders. Figure. Schematic of 2 platform randomized controlled trials (RCTs). The RCTs have collaborated to implement 1 common protocol studying Intervention a compared with control in commonly-eligible subjects (a multiplatform RCT). The multiplatform trial approach involves prospectively harmonizing trial inclusion criteria, interventions, data collection, and outcome measures, and performing a single joint analysis. Also demonstrated here is the possibility that one of the platform trials provides infrastructure to evaluate an investigational new drug for registration purposes, including the use of expanded data collection, monitoring, outcome collection, and a potential distinct statistical model for a separate outcome. Further detail reflecting the iterative and perpetual nature of platform trials more generally is depicted elsewhere. 2 Multiple, overlapping platform RCTs may be developed. Distinct platforms may nevertheless collaborate to undertake a single RCT protocol (Figure). This concept of a multiplatform RCT was demonstrated by the ATTACC/ACTIV-4a/REMAP-CAP trial of therapeutic-dose heparin for COVID-19, whereby 3 operationally distinct platforms prospectively adopted a common RCT protocol that was independently implemented by the 3 participating platforms, harmonizing eligibility, data collection, treatments, and outcomes measures with a single, common statistical model pooling all data. 5 The protocol was developed collaboratively with representation from the platforms and their collaborating networks, and overseen by data and safety monitoring boards working in coordination. This distributed model increased operational efficiency and extended generalizability. The multiplatform RCT approach may offer a strategy for collaboration amidst the anticipated emergence of multiple overlapping platform trials globally. In conclusion, challenges posed by the COVID-19 pandemic highlighted opportunities for novel approaches to evidence generation in cardiovascular medicine. Platform trials demonstrated numerous operational advantages—advantages that may be afforded in nonpandemic settings as well. Platform trials may host a range of comparisons and collaborate to effect more efficient evidence generation. Already, a rapidly increasing number of platform trials are emerging globally across diverse fields of medicine. In the years to come, they may be seen as a pandemic legacy, positively impacting the way that clinical evidence is generated in cardiovascular medicine. Article Information Acknowledgments Dr Lawler thanks Gail Rudakevich, freelance medical illustrator, for generating the figure. Sources of Funding None. Disclosures Dr Lawler was a lead investigator in the adaptive multiplatform trial of therapeutic-dose anticoagulation with heparin in patients with coronavirus disease 2019 (COVID-19), including as coprincipal investigator for the participating ATTACC (Anti-Thrombotic Therapy to Ameliorate Complications of COVID-19 ) trial (https://www.clinicaltrials.gov; Unique identifier: NCT04372589), which received related funding from the Canadian Institutes for Health Research, National Institutes of Health, Peter Munk Cardiac Centre, LifeArc Foundation, Thistledown Foundation, and Province of Ontario. Dr Lawler also served on the Protocol Development Committee for the ACTIV-4a (Accelerating COVID-19 Therapeutic Interventions and Vaccines 4 ACUTE) trial (https://www.clinicaltrials.gov; Unique identifier: NCT04505774) and on the International Trial Steering Committee and as a domain chair/cochair for the REMAP-CAP (Randomized, Embedded, Multifactorial Adaptive Platform Trial for Community-Acquired Pneumonia) trial (https://www.clinicaltrials.gov; Unique identifier: NCT02735707). Dr Lawler is supported by a Heart and Stroke Foundation of Canada national new investigator career award, and reports unrelated consulting fees from Novartis, CorEvitas, the American College of Cardiology Foundation, and Brigham and Women’s Hospital, and unrelated royalties from McGraw-Hill Publishing.