乌鸦传媒

SAN FRANCISCO -- None of the first six drug regimens for COVID-19 tested in the I-SPY adaptive trial platform were successful enough to warrant further evaluation, a researcher said here.

In fact, one was booted from the trial early because hospitalized patients seemed to do worse than on usual care alone, according to Carolyn Calfee, MD, MAS, of the University of California San Francisco.

Patients in needed at least 6 L/min of supplemental oxygen to be eligible for inclusion, but only if the duration was less than 72 hours at enrollment, Calfee explained at the American Thoracic Society annual meeting. Those thought to have less than 50% probability of 6-month survival were also excluded. The trial's primary endpoint was time to recovery, defined as oxygen requirement less than 6 L/min for two consecutive days.

The six therapies were:

• Cenicriviroc, an investigational CCR2/5 receptor antagonist
• Icatibant (Firazyr), a bradykinin receptor antagonist
• Apremilast (Otezla), a phosphodiesterase-4 inhibitor
• , an investigational anti-CD14 monoclonal antibody
• Inhaled dornase (Pulmozyme), a DNA-cleaving enzyme
• Celecoxib (Celebrex), a cyclooxygenase-2 inhibitor, plus famotidine (Pepcid), an H2 blocker

The last regimen was the one terminated early for harm. Interim data showed markedly slower time to recovery and higher mortality compared with usual care, which consisted of dexamethasone and remdesivir (Veklury), along with standard hospital care.

Some of the others showed weak trends toward a positive effect, but none met the study's criteria for "graduation," defined as a strong likelihood that benefit would be confirmed in a conventional placebo-controlled phase III trial. Calfee and colleagues estimated that none of the therapies had more than a 23% chance of meeting that standard; they had set a 97.5% chance as the threshold for likely benefit.

Those same calculations produced probabilities of 99.8%-100% that further testing would prove futile.

While not finding an effective COVID therapy is obviously a disappointment, Calfee portrayed the effort as a win for the adaptive trial concept.

Several besides I-SPY COVID were launched in 2020, amidst a frenzy to find treatments for this lethal new viral disease. The concept is to develop a system for testing new treatments -- networks of trial sites and investigators who agree to follow prespecified protocols. Particular treatments can be added without designing an entirely new trial, and they can be withdrawn once enough data accumulate to judge its efficacy and safety.

Calfee offered an analogy to sports stadiums: it makes more sense to build one and play many games in it, rather than (as is done for most drug studies) build a new stadium for each game.

It's not a new concept. An was developed to test novel breast cancer therapies; organizers have said data from these studies helped underpin accelerated approvals for three such agents. In July 2020, it was expanded, with new sites and investigators, to examine potential COVID-19 therapies.

Others with similar designs include , , and the series. In fact, it was RECOVERY that first demonstrated in June 2020 that dexamethasone improved outcomes in patients with severe COVID-19.

I-SPY COVID was designed to play the role of phase II testing, such that relatively small numbers of patients (40-125 per treatment arm) and relatively short study periods were used to identify therapies with substantial promise. Calfee said the leaders understood that such a strategy holds some risk of Type II error, that is, missing a small signal of benefit with a treatment that does actually work. This, she suggested, was a necessary tradeoff during the burgeoning pandemic when time and resources were limited and everyone wanted a "big win."

In August 2020, icatibant, cenicriviroc, and apremilast became the first three drugs to enter I-SPY COVID. By early in 2021, it was clear that all had failed to meet the admittedly stiff criteria for success. The other three then launched in the trial, also eventually yielding negative results by later in 2021. (A said a seventh treatment, nebulized aviptadil, had also fallen short, but is still being evaluated in one of the ACTIV trials as an intravenous treatment.)

Across all of these trials, participants' baseline characteristics were similar -- mean patient age was about 65, two-thirds were men, about half were white, and roughly 20% were Black. Mean time since symptom onset was 9 days. About 12% were mechanically ventilated, and the same proportion received non-invasive ventilation.

Two additional agents are still being evaluated separately in I-SPY COVID, according to the Clinicaltrials.gov listing: the immunosuppressant cyclosporine, and , a monoclonal antibody targeting part of the complement pathway, under development to treat complications of hematopoietic stem cell transplant.

As the pandemic has evolved, so has I-SPY COVID, said Calfee. Initially, up to four agents could be tested at one time, and in the rush to get the study moving, patients could be randomized before they had given consent. (Those who refused were immediately dropped from the main trial, but became part of an observational cohort.) Time pressures have eased somewhat, so study organizers have cut the number of simultaneous treatments being tested to two, and consent is now obtained before randomization.

Calfee said investigators are also raising the number of patients in each treatment arm, so smaller effects can be detected, and performing more lab studies to enable "real-time biological phenotyping" of participants' conditions.

Comment