Pandemics aren’t usually Martin Landray’s job. A physician and researcher in the University of Oxford’s Nuffield Department of Population Health, Landray designs clinical trials—cardiology, mostly, the kind of industry-funded studies that suck in tens of thousands of people and extrude new drugs or procedures. But in early March, Landray and his colleagues could see what was coming. People were dying in Wuhan, China; the reports trickling out of intensive care wards in Italy were horrifying. In about two weeks, fighting a new coronavirus was going to be everyone’s job, including Martin Landray’s.
So what would that job actually entail? “We had to make some fairly fundamental choices. We couldn’t see any one treatment that was going to be a cure,” he says. “We knew there were a number of treatments that had some evidence of benefit. … We knew that none of these were proven—a lot of drugs that could work, but none that we knew did work.”
So Landray and his colleagues set about creating a new kind of a drug trial. The gold standard for testing medical therapies today is the double-blinded, randomized controlled trial, which pits a treatment against a placebo given to a control group. But other options exist; Landray recalled that in the 1980s, people tested a bunch of different options for treating heart attacks against each other in a sort of randomized multiplayer death match. Landray’s team realized they could try the same thing here, testing a half-dozen contenders to treat Covid-19.
Plus: What it means to “flatten the curve,” and everything else you need to know about the coronavirus.
In just nine days, the team put together the Randomised Evaluation of Covid-19 Therapy Trial, or “Recovery” for short. In 160 hospitals across the UK, they started recruiting patients presenting with Covid-19, who consented to be randomly assigned to get one of several drugs: the HIV antivirals lopinavir and ritonavir, the anti-inflammatory steroid dexamethasone, the antimalarial immune suppressant hydroxychloroquine, or the antibiotic and antiinflammatory azithromycin. They’d later add another anti-inflammatory called tocilizumab. Patients might also be randomized to get the standard of care—none of those drugs. An independent data monitoring committee would keep track of who would get better, who ended up on a ventilator, and who died. Unless one of the options looked wildly good or terribly bad, even Landray wouldn’t see data until it showed useful outcomes.
There is, however, a catch: Marginal, hard-to-perceive effects only show up in a really big study population. “One does need large numbers, because the numbers you need are driven by how effective you think the drugs are going to be,” Landray says. Today, five weeks after the first patient got randomized, his team has nearly 7,000 people across all the arms of the study, with about 2,000 more signing up every week. “I have no idea whether any of these treatments are effective, no idea at all,” he adds. “They all have a reasonable chance. None is likely to be stunningly effective.”
Broadly, this kind of trial design is called “adaptive,” an idea that many researchers hope will be a hyperdrive for the search for drugs and vaccines against Covid-19. The World Health Organization has launched a multidrug trial similar to Recovery, as has continental Europe. The US National Institute of Allergy and Infectious Diseases is spinning one up too, starting with the drug remdesivir and a placebo, with plans to add new drugs as they become available. They’re happening for vaccines too.
The problem is too many things to test, and a too-slow method of testing them. At least 180 potential drugs are in some stage of trials, and 78 vaccine candidates are in exploratory or preclinical tests. (Six are in early-stage human safety trials). But a spreading pandemic means a need to find and deploy the things that work, like, now. That’s where trial design comes in. Clever statistical and methodological approaches could produce a winner in months instead of years.
A New Kind of Clinical Trial
A couple of decades ago, the agencies and regulators responsible for getting new drugs made and sold noticed that they weren’t getting much return for the amount of money and time researchers and pharmaceutical firms poured onto the problem. It took too long to develop the drugs and recruit enough people to study them properly, and even if you spent the years necessary to do all that, 90 percent of all vaccine candidates failed. One recent estimate put the cost of getting a successful vaccine against an epidemic-causing germ at upward of $1 billion. So planners started coming up with approaches to the development process that might speed things up, or at least remove some friction from the pipeline. Adaptive or flexible trials were one of the ideas—studies that would still tease apart safety and efficacy, but with twists allowing for greater speed. They have lots of subtypes. Researchers can add in or delete drugs on the fly, as Recovery is planning. A “platform trial” tests lots of candidates against a shared control group. A “core protocol” can stop and start, adding new people or whole groups. The list goes on.
In 2015, a biostatistician at the University of Florida named Natalie Dean was working on trials for a vaccine against Ebola, then burning its way through West Africa. The design of the study was unusual. “Instead of gathering people from the general public to participate, it targeted people who were in close contact with confirmed cases,” Dean says, at the University of Florida. Rather than start with young, healthy subjects as is typical, that team went right to the folks who’d most need the vaccine.
The vaccine proved successful, and Dean went on to work with the WHO on an effort to figure out how to improve drug, diagnostic, and vaccine research and development more broadly—“with a focus on emerging pathogens that could cause public health emergencies that we don’t have anything for,” she notes. The Coalition for Epidemic Preparedness Innovations had a list of those possible pandemics, and the 21st century’s previous coronaviruses were on it: SARS and MERS. But so was a placeholder for the devil they didn’t know. They called it “Disease X,” something no one had ever seen before, but that humanity would still need to prepare for. It turned out to be Covid-19.
A big part of the plan for Disease X was the repurposing of existing drugs and vaccine technologies, and that’s what’s happening in the first tranche of Covid-19 treatments. Among them are remdesivir—a drug in multiple trials around the world, which was originally created to fight Ebola—antiretrovirals used against HIV, and the controversial antimalarial chloroquine. Separately, the Pentagon’s mad-science division Darpa, through its Pandemic Preparedness Program, helped the drug company Moderna spin up a vaccine based on messenger RNA, the genetic material that life on Earth uses to translate the code of DNA into proteins.
Moderna was working on vaccines against other pathogens on the WHO’s list, but in theory, its tech can just as easily use mRNA that codes for part of SARS-CoV-2, the virus that causes Covid-19, as it can for the parts of the zika or chikungunya viruses. Moderna was first out of the gate with a vaccine against Covid-19 ready to be tested in humans (it’s in phase I, the safety-check stage), and no vaccine based on messenger RNA has ever been licensed. Another vaccine candidate, from Inovio, is based on DNA and started out aimed at MERS. The other early entrants also had previous lives fighting other diseases.
The problem is capturing the cream from that bumper crop. “You can’t take 200 vaccine candidates to later-stage clinical development,” says Seth Berkley, president and CEO of GAVI, the Vaccine Alliance, an international NGO. “We’re concerned about scaling up manufacture, so the down regulation is going to be critical. You don’t want necessarily to have one candidate, but you also don’t want to have 50.”
Here again is where trial design might help. In one version, for example, multiple vaccines can enter the virtual thunderdome, and only a winner leaves. “You prespecify a milestone or criteria that you’re looking to achieve, to advance the candidate,” says Wayne Koff, president and CEO of the Human Vaccines Project. “If you get a candidate that achieves that, you can advance it to a larger trial—roll that right over from a small phase I to a larger phase II or potential efficacy trial, whereas the other candidates that have not achieved that can be dropped.”
The WHO is spinning up a whole new protocol that does just that: the Solidarity Vaccine Trial. It’s set to have a half-dozen vaccine candidates all working head-to-head, with a shared, unvaccinated control group to make the statistics work. (That’s called a “platform trial.”) So far no pharmaceutical companies have signed on, but it looks like it’ll be one of the higher-profile global efforts.
Trial design can also help you finesse the recruitment of enough people for statistical power. “Normally the way you do a trial is a double-blind, placebo control, and you look at the trial for safety results at multiple times. But you don’t do multiple analyses of the data, because you have to take a power cut if you’re doing that,” GAVI’s Berkley says. In other words, after you grow from a relatively small safety study into a test for efficacy—which usually entails 10 times as many people, because the effect might be subtle—every time you stop and reanalyze the data, you rule participants out of the study, which costs you statistical significance. But in an adaptive design, you can plan for that, build it into the study, and keep the wheel turning with no statistical loss. “You can build in those types of cuts to understand what level you need to get to the next group, and have statistical significance in the subgroups,” Berkley adds.
Read all of our coronavirus coverage here.
In Covid-19 drugs, those subgroups are going to be important. Older people tend not to respond as well to vaccines, but they’re one of the subpopulations most vulnerable to Covid-19. A researcher might want to carve that group out and look at them differently, in case they need multiple doses of a vaccine, even though a young, healthy person—typically, the first group that gets studied—would only need one. Or they might need a vaccine with an adjuvant, an extra ingredient that gooses the immune system and enhances a vaccine. Usually children are major vectors for a disease and need to be immunized early. Nobody knows what role, epidemiologically, they play with Covid-19. Gotta figure that out too.
That’s all demographics; drugs and vaccines also have to cope with geography. “When we’re talking about ‘adaptive,’ we’re talking about bigger-picture flexibility, like adding totally new areas of the world to the trial partway through,” the University of Florida’s Dean says. “The biggest limiting factor with these types of trials is usually that there aren’t enough cases, because outbreaks are very unpredictable. They start and stop. Different areas are hot spots.” That’s why a bunch of Covid-19 drug trials in China have stopped before getting any results; they ran out of patients. The same thing happened with those trials of an Ebola vaccine in the 2014 West Africa outbreak, and with trials of the triple monoclonal antibody cocktail ZMapp—which meant that in the Democratic Republic of Congo in 2018, nobody had good enough data to compare new therapies to ZMapp.
Using what Dean and her biostatistician colleagues call a “core protocol,” though, trials can start, stop, and then start again when new hot spots crop up. An independent group collects the data and accounts for differences in the various arms of the study. “The bigger the trial, the faster you can accrue people and get answers. But the bigger value and innovation in my mind is that this makes it more robust when outbreaks end unexpectedly,” says Dean. “If China had been willing to team up with a larger group or larger effort, then they could have finished their trials, because someone else could have picked up where they left off.”
The Need for Cooperation
These ideas don’t work unless groups that, under most circumstances, might not want to help each other seek out common ground. That could be a problem. “In theory, you can line up a number of the candidates—if they’re ready to go into trials at the same time, which is a big if. And if the companies are willing to run their vaccines head-to-head against somebody else, which is a bigger if,” the Human Vaccines Project’s Koff says. “We tried to do this with HIV way back when, and it was really difficult, because one company really doesn’t like to be in a bake-off against another company—even though it’s a reality that whether you run them in the same clinical trial or sequential runs, it’s a bake-off.”
Vaccinologists and immunologists around the world already have been burning up their Zoom accounts to share information, but Koff says he’s going to try to initiate more formal global information-sharing meetings. “I think this is in a league of its own in terms of the unprecedented efforts and speed and capacity and infrastructure that is involved in advancing the vaccine effort against this disease,” he adds. “But I don’t think it’s as coordinated as it could be.”
That’s particularly true when basic knowledge about the disease is still lacking. No one really understands the role of abnormal blood coagulation, which seems to cause dangerous clots and even strokes in some Covid-19 patients. Nobody really understands what causes the “cytokine storm” in the end stages of fatal Covid-19, in which a person’s own immune system goes haywire. No one knows if a vaccine that merely stops the infection from turning into disease, as most vaccines do, will be good enough—or if one will have to actually halt infection itself. Every disease has “correlates of immunity,” the immunological indicators that a person is disease-proof; without widespread blood testing, nobody even really knows what Covid-19’s correlates of immunity are, or even if that kind of immunity is even possible. There isn’t even a good animal model to test these things in before they ever get near a human, though ferrets are apparently looking promising.
Trial design might not help any of that specifically. But even with all those answers in hand, a drug or vaccine can’t make its way from the lab bench to safety tests to trials for efficacy without cooperation and creativity. “Every country, every scientist, every company will love their own product, for obvious reasons. You have to. So you need to have a set of screens set up,” Berkley says. “What you need is a place where all of this information can be shared in a trusted and neutral environment.”
As that gets assembled by the world’s scientists, the basic epidemiology of Covid-19 itself might actually help, for once. As the disease spreads from hot spot to hot spot, it provides more subjects to build bigger, more powerful tests of things that might defeat it. Landray’s Recover trial has a bit of the core-protocol concept built into it too, it turns out. “Even if there were a lull in the number of cases in a month or two’s time, the platform is still there. As new cases arise in a second phase, whenever that might be, we would be able to rapidly activate again and enroll those patients,” he says. “Everybody is mindful there will be some second phase, some second wave. They don’t know when it will be, how high it will peak, how long it will go on. But we have to go into that knowing more about which treatments work than we did this time around.”
None of the drugs in the Recovery trial is likely to be a magic cure. But like the cardiac treatments Landray used to study, they could be used in concert, and might give a few people a slight edge. The point of “flattening the curve” was to keep people from getting sick and prevent hospitals from becoming overwhelmed, sure, but it was also a delaying tactic. Spreading out cases over a longer duration gives scientists time to try new stuff, pushing new infections forward in time until they pass some therapeutic frontier, where scientists have tools doctors and nurses can actually use.
4/24/20 12:45 PM This story was updated to clarify the dates of Natalie Dean’s work at the University of Florida, the success of the Ebola vaccine, and the methodology of the Solidarity Vaccine Trial.
WIRED is providing free access to stories about public health and how to protect yourself during the coronavirus pandemic. Sign up for our Coronavirus Update newsletter for the latest updates, and subscribe to support our journalism.
More From WIRED on Covid-19