Shared Control: Novel Trial Design Works In COVID, But Will Sponsors Buy In?
Executive Summary
COVID-19 has helped demonstrate the viability of platform trials with a shared control arm. That model is easier to extol than it is for sponsors to embrace, especially outside of crisis mode.
The lessons of the COVID-19 pandemic will be examined for years to come, but one message is already clear: the value of large platform trials involving multiple potential treatments in a single protocol.
Specifically, the United Kingdom’s RECOVERY (Randomised Evaluation of COVID-19 Therapy) trial, which has now delivered three definitive outcomes just six months into the outbreak. Specifically, dexamethasone has significant benefits in critically ill patients, while lopinavir/ritonavir in combination and hydroxychloroquine as a single agent do not.
The latter finding had the salutary effect of ending a highly-charged and ultimately distracting debate over the role of the antimalarial agent that would likely still be raging without the study – a clear demonstration of the true value of “negative” results. (Also see "Hydroxychloroquine EUA Is First Rx Victim Of Hahn’s ‘Move Fast, Course Correct’ Philosophy" - Pink Sheet, 15 Jun, 2020.)
Duke Margolis Center for Health Policy Director (and former US Food & Drug Administration Commissioner) Mark McClellan highlighted that track record during the recent Drug Information Association annual meeting to underscore the value of clinical trial networks to deliver rapid results. (Also see "COVID-19 Puts Spotlight On Clinical Trial Networks: 'We Really Need Them Right Now'" - Pink Sheet, 2 Jul, 2020.)
Learning A Lesson Many Already Knew
Of course, like many other lessons learned from COVID, the need for enhanced clinical trial infrastructure in the US was already evident to many long before the pandemic. The longtime head of FDA’s Center for Drug Evaluation & Research, Janet Woodcock, has emphasized that subject for at least a decade – helping to drive efforts to create master protocols in oncology and other disease areas to help increase trial participation and speed development.
And, indeed, she was urging the rapid adoption of COVID-19 master protocols as the outbreak ramped up in the US this spring. (Also see "Woodcock Warns Research Apparatus Could Be Overwhelmed If Coronavirus Trials Don’t Use Master Protocols" - Pink Sheet, 27 Apr, 2020.)
Woodcock has now transitioned to a more hands-on role as part of the Operation Warp Speed project – though, for the time being at least, there is precious little information about exactly what that federal effort is doing to coordinate therapeutic development amid an alphabet soup of public-private consortia. (Also see "For COVID, Consensus On Value Of Platform Trials, But A Confusion Of Consortia" - Pink Sheet, 1 Jun, 2020.)
It may be that, in the US at least, the moment has already passed to replicate the remarkable success of RECOVERY. Just like it is too late to start stockpiling necessary medicines and supplies once the panic buying has begun, it may be too late to try to organize a dedicated clinical trial network when the hospitals are already filling up – at least, in the context of the US health care system.
Sharing Control – Of More Than Just The Placebo Arm
In the UK, on the other hand, having the National Health Service architecture in place made it much simpler to roll out a truly national study protocol. The trial enrolled its first patient in March; it now has enrolled nearly 12,000 subjects at more than 170 hospitals across the UK.
In the US, it is inherently more difficult to organize networks of hospital into a single overarching protocol. Consider how little data collection actually occurred with the HCQ distribution program in the US – a tiny percentage of treatments under the Emergency Use Authorization were tracked and reported in compliance with the directions for use. (Also see "Missing Data: One Long-Term Lesson From US FDA Action On Hydroxychloroquine" - Pink Sheet, 16 Jun, 2020.)
Still, it is easy to imagine a system that could work with a little advance planning. Perhaps one takeaway from COVID so far is that the US needs a “pandemic trial emergency network” in place and ready to activate if and when the next major infectious disease outbreak occurs.
But that approach would not by itself address the underlying issues about the US clinical trial infrastructure, just as the Strategic National Stockpile doesn’t address concerns about “routine” drug shortages in the US.
And that gets to the more truly revolutionary component of the RECOVERY trial: not the infrastructure so much as the underlying design. And the key feature can be captured in a single phrase: shared control.
In a technical sense, the trial uses a shared control arm, which greatly increases the power of the study – many fewer patients are needed than would be the case if each of the agents was studied separately against usual care.
That is an idea that has been championed in the US in different settings, with little success. FDA Oncology Center of Excellence Director Richard Pazdur in particular has urged sponsors of cancer therapies targeting rare tumor subtypes to collaborate on trials using a shared placebo arm to maximize the ability to complete studies – and even proposed the potential of a class-indication for products like PD1 inhibitors that participated. (Also see "PD-1 Fatigue: Cross-Labeling May Help Pare Checkpoint Inhibitor Pipeline" - Pink Sheet, 15 Aug, 2018.)
The Risks Of Head-to-Head Trials
Sponsors, however, have proven reluctant to embrace the approach – knowing that head-to-head comparisons across treatment arms will be inevitable, even if the statistical approach does not allow for that.
It is noteworthy that the one definitive therapeutic result delivered outside of RECOVERY, the National Institutes of Health study that showed efficacy of remdesivir, was a straightforward, head-to-head design. It is worth asking how the data from that study would be viewed if it had come from RECOVERY instead. The remdesivir trial showed a significant impact on hospitalization time, but did not reach statistical significance on mortality. How would that result be viewed if it came out of RECOVERY, with same control arm as was used for dexamethasone (where a survival benefit was proven)?
And that points to a far broader notion of “shared control” that underpins the RECOVERY design: commercial sponsors have very little role in the conduct of the study, beyond contributing the potential treatment if they choose. The investigator team selects the potential treatment arms and controls the release of information on success or failure.
They also have no recourse but to accept the implied head-to-head comparisons that will inevitably be made. Should another treatment show efficacy in the weeks and months ahead, that result will immediately be contrasted with the approximately 30% mortality benefit shown for dexamethasone. The comparison may not be statistically valid given the nuance of the RECOVERY design – but it will at least have some measure of plausibility that is lacking when comparisons are drawn across two completely independent studies based on the assumption that the control arms are identical.
Amid the COVID pandemic, the urgency for answers is paramount for everyone. But sharing control is likely to prove much more difficult for sponsors when the pandemic passes and they are considering how best to develop their most important pipeline projects.