Dr. David Fajgenbaum looking through a microscope at his lab.
They're running a publicly accessible database called the CORONA Project (COvid19 Registry of Off-label & New Agents) that to date tracks 400 different COVID-19 treatments that have been tried somewhere in the world, along with the frequency of their use, and the outcomes.
"There's so many drugs being used all over the place, in different ways, with different outcomes," says Fajgenbaum. "We're trying to add some order to the madness."
20,000 people have accessed the registry—other physicians and researchers, those in the pharmaceutical industry, and even curious lay people—and the data are now being shared with the U.S. Food and Drug Administration in the hopes of launching large-scale trials that would lead to approving a constellation of treatment options for COVID-19 faster than any new drugs could come online.
"What David's group has done with the CORONA Project is on a scale that I don't think has ever been seen before," says Heather Stone, a health science policy analyst at the FDA who specializes in drug repurposing. She was not involved in establishing the project, but is now working with its data. "To collect and collate that information and make it openly accessible is a massive feat, and a huge benefit to the medical community," she says.
On a Personal Mission
In the science and medical world, Fajgenbaum lives a dual existence: he is both researcher and subject, physician and patient. In July 2010, when he was a healthy and physically fit 25-year-old finishing medical school, he began living through what would become a recurring, unprovoked, and overzealous immune response that repeatedly almost killed him.
His lymph nodes were inflamed; his liver, kidneys, and bone marrow were faltering; and he was dead tired all the time. At first his doctors mistook his mysterious illness for lymphoma, but his inflamed lymph nodes were merely a red herring. A month after his initial hospitalization, pathologists at Mayo Clinic finally diagnosed him with idiopathic multicentric Castleman disease—a particularly ruthless form of a class of lymph node disorders that doesn't just attack one part of the body, but many, and has no known cause. It's a rare diagnosis within an already rare set of disorders. Only about 1,500 Americans a year receive the same diagnosis.
Without many options for treatment, Fajgenbaum underwent recurring rounds of chemotherapy. Each time, the treatment would offer temporary respite from Castleman symptoms, but bring the full spate of chemotherapy side effects. And it wasn't a sustainable treatment for the long haul. Regularly dousing a person's cells in unmitigated toxicity was about as elegant a solution to Fajgenbaum's disease as bulldozing a house in response to a toaster fire. The fire might go out (though not necessarily), but the house would be destroyed.
A swirl of exasperation and doggedness finally propelled Fajgenbaum to take on a crucial question himself: Among all of the already FDA-approved drugs on the market, was there something out there, labeled for another use, that could beat back Castleman disease and that he could tolerate long-term? After months of research, he discovered the answer: sirolimus, a drug normally prescribed to patients receiving a kidney transplant, could be used to suppress his overactive immune system with few known side effects to boot.
Fajgenbaum became hellbent on devoting his practice and research to making similar breakthroughs for others. He founded the Castleman Disease Collaborative Network, to coordinate the research of others studying this bewildering disease, and directs a laboratory consumed with studying cytokine storms—out-of-control immune responses characterized by the body's release of cytokines, proteins that the immune system secretes and uses to communicate with and direct other cells.
In the spring of 2020, when cytokine storms emerged as a hallmark of the most severe and deadly cases of COVID-19, Fajgenbaum's ears perked up. Although SARS-CoV-2 itself was novel, Fajgenbaum already had almost a decade of experience battling the most severe biological forces it brought. Only this time, he thought, it might actually be easier to pinpoint a treatment—unlike Castleman disease, which has no known cause, at least here a virus was clearly the instigator.
"Because [a drug] looks promising, we need to do a well-designed, large randomized controlled trial to really investigate whether this drug works or not ... We don't use that to say, 'You should take it.'"
Thinking Beyond COVID
The week of March 13, when the World Health Organization declared COVID-19 a pandemic, Fajgenbaum found himself hoping that someone would make the same connection and apply the research to COVID. "Then like a minute later I was like, 'Why am I hoping that someone, somewhere, either follows our footsteps, or has a similar background to us? Maybe we just need to do it," he says. And the CORONA Project was born—first as a 10-day exercise, and later as the robust, interactive tool it now is.
All of the 400 treatments in the CORONA database are examples of repurposed drugs, or off-label uses: physicians are prescribing drugs to treat COVID that have been approved for a different disease. There are no bonafide COVID treatments, only inferences. The goal for people like Fajgenbaum and Stone is to identify potential treatments for further study and eventual official approval, so that physicians can treat the disease with a playbook in hand. When it works, drug repurposing opens up a way to move quickly: A range of treatments could be available to patients within just a few years of a totally new virus entering our reality compared with the 12 - 19 years new drug development takes.
"Companies for many decades have explored the use of their products for not just a single indication but often for many indications," says Stone. "'Supplemental approvals' are all essentially examples of drug repurposing, we just didn't call it that. The challenge, I think, is to explore those opportunities more comprehensively and systematically to really try to understand the full breadth of potential activity of any drug or molecule."
The left column shows the path of a repurposed drug, and on the right is the path of a newly discovered and developed drug.
Cures Within Reach
In Fajgenbaum's primary work, promising drugs stand out easily. For a disease like Castleman, where improvement almost never occurs on its own, any improvement that follows a treatment can pretty clearly be attributed to that treatment. But Fajgenbaum says tracking COVID outcomes is less straightforward since "the vast majority of people will get better, whether they take steroids or they take Skittles." That's why the intent of the database is to identify promising treatments only to generate hypotheses and fruitful clinical trials, not to offer full-throated treatment recommendations. Within the registry, Fajgenbaum considers a drug promising if it's being used in humans, not just in lab animals, and a significant proportion of cases report patient improvement.
"It's that sort of combination of rock-solid randomized controlled trial data, plus anecdotal retrospective data, that we combine to say, 'Wow, this drug looks more promising than another,'" says Fajgenbaum. "Because it looks promising, we need to do a well-designed, large randomized controlled trial to really investigate whether this drug works or not ... We don't use that to say, 'You should take it.'"
Experts say that the search for repurposed drugs to treat COVID could have implications for rare diseases in general. Rare diseases, of which Castleman is one, affect 400 million people around the world. 95% of them don't have a tailor-made, FDA-approved drug treatment. Developing one is a lengthy and often prohibitively expensive process. If only a dozen people will benefit from and buy a drug, it's not often worth it to pharmaceutical companies to spend millions of dollars making them. On occasion when they do, however, that overhead shows up in the price tag: the top 10 most expensive drugs in the world are all for rare diseases, often making them inaccessible to patients. Identifying new clinical uses for drugs that already exist is critical for opening a trap door out of a cycle that prioritizes profits over health outcomes.
"COVID is an interesting case where it's demonstrated that when the scientific and medical community really focuses all of its efforts and talents on a single problem, a solution can be identified and in a much faster time period than has ever historically been the case," says Stone. "I certainly wish it hadn't taken a pandemic to do that, but I think it does have lessons for the future in terms of our ability to accomplish things that we might have previously not thought were possible"—for example, mainstreaming the idea of drug repurposing as a treatment tool, even long after the pandemic subsides.
A Confounding Virus
The FDA declined to comment on what drugs it was fast-tracking for trials, but Fajgenbaum says that based on the CORONA Project's data, which includes data from smaller trials that have already taken place, he feels there are three drugs that seem the most clearly and broadly promising for large-scale studies. Among them are dexamethasone, which is a steroid with anti-inflammatory effects, and baricitinib, a rheumatoid arthritis drug, both of which have enabled the sickest COVID-19 patients to bounce back by suppressing their immune systems. The third most clearly promising drug is heparin, a blood thinner, which a recent trial showed to be most helpful when administered at a full dose, more so than at a small, preventative dose. (On the flipside, Fajgenbaum says "it's a little sad" that in the database you can see hydroxychloroquine is still the most-prescribed drug being tried as a COVID treatment around the world, despite over the summer being debunked widely as an effective treatment, and continuously since then.)
One of the confounding attributes of SARS-CoV-2 is its ability to cause such a huge spectrum of outcomes. It's unlikely a silver bullet treatment will emerge under that reality, so the database also helps surface drugs that seem most promising for a specific population. Fluvoxamine, a selective serotonin reuptake inhibitor used to treat obsessive compulsive disorder, showed promise in the recovery of outpatients—those who were sick, but not severely enough to be hospitalized. Tocilizumab, which was actually developed for Castleman disease, the disease Fajgenbaum is managing, was initially written off as a COVID treatment because it failed to benefit large portions of hospitalized patients, but now seems to be effective if used on intensive care unit patients within 24 hours of admission—these are some of the sickest patients with the highest risk of dying.
Other than fluvoxamine, most of the drugs labeled as promising do skew toward targeting hospitalized patients, more than outpatients. One reason, Fajgenbaum says, is that "if you're in a hospital it's very easy to give you a drug and to track you, and there are very objective measurements as to whether you die, you progress to a ventilator, etc." Tracking outpatients is far more difficult, especially when folks have been routinely asked to stay home, quarantine, and free up hospital resources if they're experiencing only mild symptoms.
But the other reason for the skew is because COVID is very unlike most other diseases in terms of the human immune response the virus triggers. For example, if oncology treatments show some benefit to people with the highest risk of dying, then they usually work extremely well if administered in the earlier stages of a cancer diagnosis. Across many diseases, this dialing backward is a standard approach to identifying promising treatments. With COVID, all of that reasoning has proven moot.
As we've seen over the last year, COVID cases often start as asymptomatic, and remain that way for days, indicating the body is mounting an incredibly weak immune response initially. Then, between days five and 14, as if trying to make up for lost time, the immune system overcompensates by launching a major inflammatory response, which in the sickest patient can lead to the type of cytokine storms that helped Fajgenbaum realize his years of Castleman research might be useful during this public health crisis. Because of this phased response, you can't apply the same treatment logic to all cases.
"In COVID, drugs that work late tend to not work if given early, and drugs that work early tend to not work if given late," says Fajgenbaum. "Generally this … is not a commonplace thing for a virus."
"There are drugs that are literally sitting in every single hospital pharmacy in the country that, if a study shows it's effective, can be deployed that evening to patients on a massive scale."
This see-sawing necessitates tracking a constellation of drugs that might work for different stages of the disease as a patient moves from the weak immune response stage into the overzealous immune response.
"COVID is difficult, compared to other diseases, because there are so many different levels of disease severity, and recovery at different rates," says Stone, the FDA researcher. "That makes it hard to see the patterns or signals and it makes it very important to collect very, very large numbers of cases in order to really reliably identify signals."
This particular moment in the pandemic feels like a massive tipping point, or the instant a tiny pinprick of light finally appeared at the end of the tunnel: several vaccines are already here, with more on the way imminently. In the U.S., more than 65 million doses of the vaccine have been administered, and positive COVID cases are finally falling back to levels not seen since October. On the hopeful surface, it might seem a strange moment to be preparing to launch trials that will validate treatments for a virus it seems the U.S. may finally be beating back. But at best, Americans are still months away from reaching herd immunity through vaccination, and new circulating variants may threaten to upend our fragile progress.
"In the meantime, there are drugs that are literally sitting in every single hospital pharmacy in the country that, if a study shows it's effective, can be deployed that evening to patients on a massive scale. It wouldn't have to be newly produced, it wouldn't have to be shipped, it's literally there already," says Fajgenbaum. "The idea that you can save a lot of lives by finding things that are just already there I think is really compelling, given how many people are going to die over these next few months."
Even after that, not everyone can or will be vaccinated, and, as the Wall Street Journal recently reported, "The pathogen will circulate for years, or even decades, leaving society to coexist with Covid-19 much as it does with other endemic diseases like flu, measles, and HIV." Neither vaccines, personal behavior, or treatments alone is a panacea against the virus, but together they might be.
"It's important to explore all avenues in this public health emergency, and drug repurposing can continue to play a role as the pandemic continues and evolves," says Stone. "I think COVID variants in particular are a big concern at the moment, and therefore continuing to investigate new therapeutics, even as the vaccines roll out, will continue to be a priority."
A recent study in The Lancet Oncology showed that AI found 20 percent more cancers on mammogram screens than radiologists alone.
Published in The Lancet Oncology, the study analyzed the scans of 80,000 Swedish women with a moderate hereditary risk of breast cancer who had undergone a mammogram between April 2021 and July 2022. Half of these scans were read by AI and then a radiologist to double-check the findings. The second group of scans was read by two researchers without the help of AI. (Currently, the standard of care across Europe is to have two radiologists analyze a scan before diagnosing a patient with breast cancer.)
The study showed that the AI group detected cancer in 6 out of every 1,000 scans, while the radiologists detected cancer in 5 per 1,000 scans. In other words, AI found 20 percent more cancers than the highly-trained radiologists.
Scientists have been using MRI images (like the ones pictured here) to train artificial intelligence to detect cancers earlier and with more accuracy. Here, MIT's AI system, MIRAI, looks for patterns in a patient's mammograms to detect breast cancer earlier than ever before. news.mit.edu
But even though the AI was better able to pinpoint cancer on an image, it doesn’t mean radiologists will soon be out of a job. Dr. Laura Heacock, a breast radiologist at NYU, said in an interview with CNN that radiologists do much more than simply screening mammograms, and that even well-trained technology can make errors. “These tools work best when paired with highly-trained radiologists who make the final call on your mammogram. Think of it as a tool like a stethoscope for a cardiologist.”
AI is still an emerging technology, but more and more doctors are using them to detect different cancers. For example, researchers at MIT have developed a program called MIRAI, which looks at patterns in patient mammograms across a series of scans and uses an algorithm to model a patient's risk of developing breast cancer over time. The program was "trained" with more than 200,000 breast imaging scans from Massachusetts General Hospital and has been tested on over 100,000 women in different hospitals across the world. According to MIT, MIRAI "has been shown to be more accurate in predicting the risk for developing breast cancer in the short term (over a 3-year period) compared to traditional tools." It has also been able to detect breast cancer up to five years before a patient receives a diagnosis.
The challenges for cancer-detecting AI tools now is not just accuracy. AI tools are also being challenged to perform consistently well across different ages, races, and breast density profiles, particularly given the increased risks that different women face. For example, Black women are 42 percent more likely than white women to die from breast cancer, despite having nearly the same rates of breast cancer as white women. Recently, an FDA-approved AI device for screening breast cancer has come under fire for wrongly detecting cancer in Black patients significantly more often than white patients.
As AI technology improves, radiologists will be able to accurately scan a more diverse set of patients at a larger volume than ever before, potentially saving more lives than ever.