What do we do with our under-enrolled single-center COVID-19 clinical trials?

Given the large number of potential interventions to be evaluated for the treatment of COVID-19, multi-arm platform trials were advocated early in the pandemic.¹ This design framework provides advantages, including shared key design and infrastructure components by creating a master protocol that allows the systematic and simultaneous evaluation of COVID-19 therapeutics.² While ideal for the COVID-19 pandemic, platform trials require advanced and extensive planning, centralized shared governance, and close cooperation among researchers.² In the US, early in the pandemic with the rising number of cases and deaths, and without an organized, collaborative, and multi-site effort, many research institutions and investigators started their own investigator-initiated trials in isolation to evaluate new and repurposed treatments and strategies. At our institutions Columbia University Irving Medical Center (CUIMC) and Yale Health System, over 15 investigator-initiated trials were designed in early Spring of 2020. At both institutions, as cases declined following the initial surge, enrollment decreased or stopped. The recent resurgence of cases coincides with coordinated national implementation of master-protocols and adaptive platform clinical trials under the umbrella of the Accelerating COVID-19 Therapeutic Interventions and Vaccines initiative³; this presents an opportunity to re-think the landscape and hopefully “get it right”.

The questions many institutions like ours have been struggling with are what to do with our existing single-center trials. Should we join these newly established national efforts, should we try to expand to additional sites, and how should we prioritize patient enrollment into trials? The answer to these questions hinges on what can be learned from the single-center investigator-initiated trials that were started and have struggled with under-enrollment and the ethical implications of continuing these studies. The majority of these trials at our institutions are smaller randomized phase 2 treatment trials with sample sizes in the range from 100 to 200, and the control arm being the standard of care with or without a placebo. When we designed these trials, the literature and understanding on COVID-19 was limited. Few trials on COVID-19 had been published, and the literature was changing rapidly making it difficult to specify the event rate in the control arm and define a clinically relevant effect size.⁴ In many instances, instead of calculating sample sizes based on a clinically relevant effect size, the limitations on sample size led to the detectable effect size. Therefore, many of the single center trials may be underpowered to detect clinically relevant differences. Moreover, as new data emerge, the standard of care for treating COVID-19 patients has evolved substantially. In this rapidly evolving landscape, the assumptions made during the design of the trial might no longer be valid and the interpretation of clinical trials randomizing to intervention of interest versus a constantly changing standard of care will be difficult. To clearly understand the effects of the intervention of interest in these trials will require more sophisticated analyses that potentially need larger sample size to account for a host of variables.

To streamline the decision for COVID-19 trials, institutions and investigators should consider the ethical and logistical implications of terminating versus continuing enrollment, and how these single-center trials can advance our knowledge for the treatment of COVID-19. Investigators should consider terminating trials that are no longer relevant based on emerging evidence. For example, with the results from the RECOVERY and Solidarity trials regarding the lack of efficacy of hydroxychloroquine,^5-6 we withdrew our hydroxychloroquine trials considering it unethical to further expose participants to an ineffective treatment. To the contrary, if new evidence suggests the efficacy of treatments under experimentation, trials should be stopped if there is lack of equipoise.

For trials that continue to be scientifically relevant and are well-designed, but underenrolled, investigators should consider adding sites or collaborating with other institutions to complete trials in a timely fashion. For an adequately powered single-center study, the required sample size will likely pose enrollment challenges partly due to the geographic shifts seen during the pandemic. For example, with the decrease in cases in New York last summer, the CUIMC convalescent plasma study added three sites in Brazil to reach its target enrollment (NCT04359810).⁷ The addition of sites should be considered particularly when trials continue to be relevant and do not overlap with other trials. Stopping these trials may hamper potential new directions of research, but without new sites they risk not meeting their enrollment goals and introducing the ethical concerns of the exposure of participants to risks of an experimental treatment and research for which study aims cannot be achieved.⁸ If there are other trials studying the same treatments, merging effort is the best way to accomplish the necessary number of participants to answer the scientific questions posed. With the recent decline in cases across the US and the design limitations mentioned, not merging with other institutions may result in trials not meeting their target accrual and being underpowered, which raises the ethical concern of the exposure of participants to risks from research for which study aims cannot be achieved.⁸ Efforts are needed to identify ways to pool the data and combine the evidence from these single center studies to draw meaningful conclusions. The advantage of starting the collaboration as early as possible is to ensure the collection of common data elements under similar protocols. For example, we encouraged our investigators to share their protocols through the COVID-19 Collaboration Platform (https://covidcp.org) and check the outcomes utilized by other trials in Clinicaltrials.gov to urge the collection of common outcomes. These efforts can facilitate future data-sharing and meta-analysis. Approaches have been proposed to pool data from multiple clinical trials with the same treatment by creating a minimal data set and a central data repository for analysis.⁹ We should promote collaboration among smaller, investigator-initiated studies, and enrollment into centralized trials that have the potential of advancing our scientific knowledge. The aim is not to dampen innovation and investigator-initiated or exploratory studies, but rather to lay the foundation for a harmonized and coordinated approach that will not only help us find promising therapeutics for COVID-19, but also prepare us for future pandemics.