sign in sign up
<<
Home , RECOVERY Trial

NEWS & ANALYSIS an audience With… RECOVERY 1 year on: a rare success in the COVID-19 landscape Peter Horby, co-​lead on the RECOVERY platform trial, discusses the origins, lessons learned and future plans for one of the most informative studies of Credit: David Fisher, COVID-19 therapeutics with Asher Mullard. Fisher Studios, Oxford

On 23 March, the UK National Health colleagues in Saudi Arabia, and adapted In the UK, we have the advantage of Service’s RECOVERY trial marked its those quickly. Within 20 days of the first data linkage. We knew that we could keep first anniversary. Initially set up to test four announcement of the outbreak, we managed the follow-up​ very simple, and ascertain possible treatment options in a few thousand to enrol the first patient into a trial of outcomes like death, particularly, but also hospitalized COVID-19 patients, the platform [the HIV drugs] and . intensive care admission and ventilation trial is still going strong. It has now enrolled Not long afterwards, we started a second trial from existing data streams. That gave us the nearly 40,000 patients — just under 10% of with [Gilead’s antiviral] remdesivir in Wuhan. ability to simplify things. the UK’s hospitalized COVID-19 patient But epidemics are a bit unpredictable, and The context was very clear: we were population — into 13 treatment arms. very aggressive public health control measures going to have a really big outbreak. There In a landscape marked by poorly designed in China actually meant that case numbers was a very real chance that the NHS would and underpowered trials, RECOVERY plummeted in Wuhan. We didn’t reach the be completely overwhelmed. Only a simple stands out as a rare bright spot. Its results target sample size for either of those studies. trial would work in those very stressed have demonstrated the benefits of the steroid I had applied for additional funding circumstances. If people don’t have time, and the IL-6-targeting​ to continue those trials, to turn these into they are not going to enrol patients into very , while showing the a platform trial in China with Chinese complicated trials. As we didn’t have much lack of efficacy for , , colleagues. The funders rang me up and said time, we also needed to get this thing started convalescent plasma, “Yes, we’ll fund you. But the cases are now in very quickly. And we realized that we were and lopinavir–ritonavir. Ongoing arms are Europe. You can do the trial in the UK”. going to need a large number of patients, testing other agents, including an antibody We had a sort of design and a pot because there was not going to be a miracle cocktail and the JAK inhibitor . of money. And that was the point that drug. Severe viral respiratory infections Peter Horby, co-lead​ of the trial and [University of Oxford’s] Martin Landray, are generally pretty difficult to treat. So we Professor of Emerging Infectious Diseases the other co-chief​ investigator of RECOVERY, needed to have a trial that would detect a and Global Health at the University of and I got in contact. He’d had a discussion modest benefit. All of that pushed us to Oxford, discussed the origins of the trial, with The Wellcome Trust’s Director Jeremy do a very simple trial. the lessons learned and the future prospects Farrar [on a bus], and Jeremy knew I’d been for large, simple platform trials. funded to do this research. This brought Did you expect this trial to get so big? together two traditions within Oxford — No. We estimated that we should look for How did RECOVERY come together? emerging infections, which I’ve been treatment effects of about a 20% reduction I’ve been working on epidemic infections working on for a long time, and large-scale​ in mortality, and to see that on a background for quite some time, and we’ve been trying pragmatic cardiovascular trials. It has been a of around 20% mortality we would need to improve the speed with which we can wonderful marriage. about 2,000 patients versus 2,000 patients. start clinical studies for some years. If you So we always anticipated we were going to go back to the 2009 influenza pandemic, How did the protocol change with this need thousands of patients. If I remember it was clear that there was a big failure to do union? correctly, I think the original approval was any good, meaningful RCTs. The next big We had been in discussion with the World for 12,000 patients. And we’ve had to keep challenge was the Ebola outbreak, where Health Organization about working with pushing that up, and up, and up. we managed to get trials started within a them on their Solidarity platform trial, When we started we had four drugs and a few months. But, still, these were started but Martin and I looked at the protocol and control group, and we weren’t sure we would towards the end of the epidemic, and in that decided it was too complicated. Martin picked finish all those. West Africa outbreak we really didn’t get any up the protocols for the ISIS (International We hadn’t expected how things would good answers. studies of Infarct Survival) trials, mega-trials​ turn out. When this outbreak started, I was working from the 1980s looking at the effect of various with a Chinese colleague and his team to set drugs on survival following heart attacks. How do you decide which drugs to test? up trials in Wuhan itself right away. We used Martin got those protocols out of a drawer, With the first four drugs that were chosen, some pre-prepared​ protocols that we had and we decided that we should do something we wanted drugs that were available in the developed for the MERS coronavirus with very simple like those. cupboard. There was a lot of discussion

336 | May 2021 | volume 20 www.nature.com/nrd

0123456789();: NEWS & ANALYSIS

The failure that I found so far most I have also been a bit surprised that some disappointing is convalescent plasma. There of the trials sponsored by pharmaceutical There’s been a huge amount were observational data that looked as if it companies appear to be quite underpowered of wasted resource, on might be beneficial. When we didn’t see an and small, given that there’s no shortage overall benefit, that was pretty disappointing. of patients. It’s hard to see the rationale for multiple underpowered or doing trials in fewer than 1,000 patients for a poorly designed trials If you could start the trial again, would pharmaceutical company that’s trying to get you make any design changes? regulatory approval for a drug. I’m not sure There are some routine data that we what the barriers are to bigger trials, and about repurposed drugs, because we could started to collect later — such as C-reactive​ I think that’s an important point that needs use them very quickly and they would protein levels and oxygen saturation some reflection. have a well-understood​ safety profile. measurements — that we could have These were [the HIV drugs] lopinavir and collected from the start. The other change What’s next for RECOVERY? ritonavir; azithromycin, a safe but is with regard to factorial design [in which There have been a lot of learnings from also an immunomodulator that is used in treatment arms are tested both alone and in RECOVERY. Rather than run this trial in inflammatory lung disease; dexamethasone, combination with one another]. We did this an academic research center, we took this because there was controversy about whether later, but we perhaps could have done this trial to the frontline. It was done across over steroids should be used in lung infections; from the outset. 175 hospitals in the UK, by frontline nurses. and hydroxychloroquine, because we thought We sort of democratized the research process. everyone’s starting to use it and we ought to Are there signals RECOVERY can’t assess, We’ve had very positive feedback from that. find out if it really works or not. because of its relatively simple setup? I think that’s a model for the future. We also After that, we were just inundated with I think we took the right approach, which leveraged data linkage. emails of things to test, from the quite was to be all inclusive and big enough so We have linkage to the hospitalization sensible to the, frankly, absurd. It quickly that we could do subgroup analyses. But we data, ICU admissions data, prescribing data became clear that we couldn’t manage that get asked a lot about activity in subgroups of and national fatality statistics. Almost 100% triage process. The UK government set up an subgroups, and things like that. You can take of the primary and secondary outcomes data independent committee called the COVID-19 two views on this. One view is that if a drug is have come through national data streams. Therapeutics Advisory Panel to look at all saving lives, does it matter? The other view is This has been enormously powerful, and of these things. Anyone can submit a drug, that we’ve only very crudely stratified patients shows that clinical trials don’t necessarily and the committee screens these and does by disease severity, but what about if a patient need to be intensive form-filling​ exercises. very good due diligence. Then they make does or doesn’t have a certain biomarker? The next thing for us is to export the recommendations, which we are free to I think we made the right trade off. model to other contexts. This is one of accept or reject. So far, we’ve accepted pretty What you don’t want — which is what you the reasons that we recently expanded much all of their recommendations. have got in some trials — is an uncertain internationally [into hospitals in result with a great deal of granularity. What and Nepal]. We want to see how we can Why didn’t you test remdesivir? you want is a certain result, and you can try export the model to places that haven’t got We tried to get it into the trial, but we could and resolve the granularity later if the result a , national research not get access to it. is positive. For example, dexamethasone infrastructure or electronic data linkage. Another drug that caused a bit of offers a clear benefit. Now, there are ongoing They could still benefit enormously from controversy is [Synairgen’s] inhaled questions about the dose, the patient simplified trials that focus on what really interferon. This was in the original protocol, subgroups that benefit most and the duration matters. but was dropped because the company of treatment. But these can be ironed out in But other contexts also means not just and the investigators who were developing subsequent trials. in emergency and in infectious disease it decided that RECOVERY wasn’t the settings. appropriate vehicle. They started their own What do you make of the rest of COVID-19 I also hope that RECOVERY — as well as phase II trial, which subsequently showed clinical trial landscape? the REMAP-​CAP trial — have made some a potential benefit. But it was a small trial. It’s very clear that there’s been a huge pharmaceutical companies rethink clinical They’ve moved on to a larger trial. amount of wasted resource, on multiple trials. Initially, companies would say to us underpowered or poorly designed trials. “we’re not very interested in your trial. It’s too To date, two out of seven completed That’s really disappointing, but I suppose not simple, too big, and it won’t meet regulatory treatment arms have shown efficacy. surprising because we’re in a pandemic and requirements”. But they come back to us once What’s your take on this hit rate? everybody wants to try and do something. their very complicated, underpowered trials I guess that’s not bad, based on normal drug There’s also been a huge amount of noise have given them an equivocal result. I think discovery pipelines. Obviously, we want generated from those underpowered trials, this will make drug companies think again every one of them to work, and it’s always as well as from unreliable observational about whether they want to spend tens of disappointing when they don’t. But taking a data. The noise around hydroxychloroquine, millions on a trial in 1,000 patients, or work step back, it’s probably not surprising given vitamin D and [the antiparasitic drug] with a national platform trial that will cost that they’re mostly repurposed drugs to date, ivermectin distracts from just doing big trials them a lot less and will give them a more with probably a low probability of efficacy. and getting convincing answers. definitive answer.

nature RevieWs | Drug Discovery volume 20 | May 2021 | 337

0123456789();: