In a series of rooms in the heart of the Stanford Cardiovascular Institute, incubators the size and shape of a dorm-room refrigerator hum quietly. Inside each, a surprise: hundreds of people.
Well, hundreds of people’s cells, that is.
The incubators house a multitude of palm-sized, clear plastic trays dotted with circular wells, which are filled with liquid ranging from pink to yellowish. Each well contains languidly beating heart cells that genetically match one person who has donated them for research. One person per row; 96 wells per plate; 10 to 20 plates per humid, warm incubator shelf; two to four incubators per room.
For the past decade, researchers in the laboratory of institute director Joseph C. Wu, MD, PhD, have been using these trays of cells to investigate the molecular causes of common heart disorders and test the effects of newly designed drugs on heart and blood vessel cells. They envision a future where the tray-based screening — a technique that Wu, the Simon H. Stertzer, MD, Professor and professor of medicine and of radiology, calls “clinical trials in a dish” — reduces the need for large-scale, expensive and time-consuming experiments on humans and laboratory animals.
Like boxy time machines, the incubators, and their contents, are poised to drastically fast-forward drug development by making it much quicker, less expensive and more precise than current methods.
Clinical trials in a dish may also reduce racial and ethnic disparities that often plague real-world clinical trial enrollment, streamline drug development that’s aided by artificial intelligence and even predict which of several possible drug treatments would work best for an individual — the ultimate in precision health.
It’s a tall order, but the vision is timely. Late last year, Congress passed the Food and Drug Modernization Act 2.0, authorizing drug developers to seek alternatives — among them, cell-based tests like Wu’s — to animal testing, which is now the gold standard when first determining whether potential drugs are effective and safe. The act is a nod to the growing realization that dish-based drug screening of human cells may offer a better way to identify promising new treatments for many illnesses, from cancer to heart disease.
There’s certainly room for improvement. It’s estimated that the cost of bringing a new drug from the earliest laboratory experiments to a final approval by the Food and Drug Administration hovers between $1 billion and $2 billion. The cost is high, in part because failure is baked into the system; fewer than 10% of drugs that succeed in animal trials go on to be approved for use in humans. And none of them are reliably effective for everyone.
“Right now, much of clinical medicine is based on trial and error,” said Wu, who is also the president of the American Heart Association. “No one drug is effective for everyone. People think because a drug is approved by the Food and Drug Administration it will work for them. But this is not always true.”
What exactly is a clinical trial?
Clinical trials come in many flavors, including double-blind, randomized and placebo-controlled. There can be multiple groups, or arms, to test various medical interventions, and outcomes may be assessed at multiple time points or triggered by certain clinical readouts. But at their core they are all basically the same: People with the same disease or condition are separated into two or more groups, some of whom will be given an experimental intervention and others who will receive either the current standard of care or no treatment. Doctors then assess which group fares better.
Such commonsense comparisons have a long history, stretching back thousands of years to biblical times. But the father of modern clinical trials is considered by many to be British naval surgeon James Lind, who in 1747 tested six possible daily interventions on 12 sailors with scurvy: seawater; cider; a fiery concoction of horseradish, mustard and garlic (yum!); an elixir of dilute sulfuric acid in alcohol; vinegar; and oranges and a lemon.
After about a week, the two sailors who had received citrus were in markedly better health than their peers. When the British navy began incorporating the routine distribution of citrus fruits and lemon juice to sailors, they virtually eradicated scurvy in their ranks.
In 1938, Congress passed the Food, Drug and Cosmetic Act, which mandated that drugs be proven safe before they can be approved and marketed for routine clinical use. First, laboratory experiments must show the molecule behaves as expected — that it binds to a protein important in disease, for example, or rectifies a disease-associated problem in cellular signaling or function. Often these first rounds of testing incorporate what are known as cell lines, or common, well-characterized lineages of cells grown for years, sometimes decades, in a lab. These predictable, genetically identical cells are the bread and butter of cell-based laboratory studies, but the bread is white and all the slices are exactly the same.
Next, testing in laboratory animals is required to determine that the molecule is safe and, ideally, that it has the predicted physiological effect, although not every human disease can be accurately mimicked in animals. Like the Wonder Bread-like cell lines, however, many laboratory animals are highly inbred and genetically similar. This similarity ensures that the cells, and animals, behave biologically in predictable ways and that experimental results are reproducible over time and among different laboratory groups.
If things go well with the animal testing, a stepwise series of studies, or clinical trials, in humans are then conducted over months or years to further confirm the drug candidate’s safety and effectiveness. Molecules that pass these hurdles — about 1 out of every 4 — are tested in a phase 3 clinical trial, which is often conducted at multiple institutions and enrolls hundreds or thousands of genetically diverse participants. These trials are meant to ferret out rare or unusual side effects that wouldn’t be obvious in a smaller, more homogeneous group. If the drug remains safe and effective, the manufacturer can apply to the FDA for approval.
Only about 3 in 10 drugs that enter phase 3 trials reach this milestone.
It’s a far cry from handing lemons to a scurvy-afflicted sailor, and the threshold for success can vary.
“There’s a phrase in the clinical trial world: number needed to treat, or NNT,” Wu said. “This refers to the number of people that need to be treated with a particular intervention or treatment to prevent one bad outcome — death, or stroke or whatever is being measured by the trial. People mistakenly think because a drug is approved by the FDA, it will always work for them. But for many drugs, the NNT may be 20 or 30 or even higher. So if the NNT is 30, that means 30 people need to take the drug for one person to benefit from it. And right now we simply don’t know which drug will work best for which individual.”
Clinical trials in a dish stand to upend this system. The first rounds of testing a drug candidate for safety and efficacy can be as simple as applying it to each well of a tissue culture plate populated with cells from many people and watching how the cells respond. If they die or stop functioning, the candidate probably doesn’t warrant further investigation; if diseased cells begin to function better, researchers can explore how or why — all without involving laboratory animals or asking human participants to roll up their sleeves or gulp a pill.
Or flip that previous scenario on its head. Test three, or five, or 10 different available drugs on a tray of cells all from one individual and see which is most effective. Now you have an idea as to which treatment might work best for that individual.
The concept can even be extended beyond cells in a dish to include organoids — micro versions of brains or intestines or pancreas made up of several cell types in a tissue.
“We are not going to eliminate the need for safety testing in animals or for clinical trials in people,” Wu said. “But if we first test a potential treatment on trays of cells or organoids from hundreds of people in a laboratory, the screening process can be much simpler and more efficient. We can identify subgroups of people who might benefit more than others and quickly weed out drugs with unacceptable safety risks on human cells.”
Heart cells from skin cells
Clinical trials in a dish became feasible only when stem cell pioneer Shinya Yamanaka, MD, PhD, showed in 2007 that it’s possible to take a specialized cell like a skin cell and wind it backward to its infancy. Unlike mature skin cells, locked into their dermatological destiny, the newly infant cells have potential to become nearly any cell in the body. Scientists deem these cells pluripotent, from the Latin “pluri,” meaning many, and “potent,” meaning power.
A decade ago, Wu and his colleagues showed that lab-generated cardiomyocytes mirror the genetic profile of their donors’ heart muscle. This indicated the cells could be a good proxy on which to test medications for genetic disorders.
It also meant these cells could be used to screen drugs of all sorts for likely impact on the heart. This is important because drugs that fail clinical trials or are pulled from the market after approval are often removed due to cardiac complications — regardless of what condition the drug was intended to treat.
“Now, instead of testing drugs on me as the guinea pig, we can test these drugs on my surrogates — the beating heart cells that are genetically identical to me,” Wu said. A tray in which every well contains Wu’s iPS-derived heart cells can be easily treated with therapeutic doses of several candidate drugs and assessed for a favorable response.
It’s more than a hypothetical scenario. Since 2017, Wu has gotten up close and personal with his own iPS cells — characterizing their responses to various medications. The cellular introspection is driven in part by curiosity, but it’s also a testament to what Wu sees as the power of cell-based testing. “Our postdocs have taken my iPS cells and differentiated them into my brain cells, heart cells, endothelial cells and liver cells,” Wu said. “I’m asking them to test some of the medications that I might need to take in the future.”
Since Yamanaka’s 2007 discovery, scientists have shown that iPS cells can also be induced to become many types of cells — meaning that the clinical trial in the dish can be used for many conditions. After you make a genetically identical copy of a person’s brain, heart, liver or other type of cell, you only need a way to assess a drug’s effect on that cell type.
Clinical trials in a dish can also be used to screen and validate the increasing numbers of drug candidates designed with the help of artificial intelligence. Some of these computational approaches generate dozens or even hundreds of molecules
“We can now essentially tell a computer program, ‘This is the structure of the protein we’re targeting; design some molecules that can bind to it,’” Wu said. “And then we can use these clinical trials in a dish to screen these molecules on cells from hundreds of genetically diverse people at one time.”
Artificial intelligence-enabled strategies like those Angela Zhang, an MD-PhD student in Wu’s lab, is using could facilitate and standardize the use of iPS-derived cells in small labs as well as large pharmaceutical companies. Zhang, who began working in Wu’s laboratory as an undergraduate, is using machine- and deep-learning techniques to predict which iPS cell colonies are likely to successfully become functional cardiomyocytes that can be used for screening.
“There’s no better place than Stanford to apply machine learning to standardize and scale up these clinical trials in a dish,” Zhang said. “And we can do it with cells from diverse backgrounds.”
Diversity is a priority for Wu and his colleagues.
“Most of the drugs approved by the FDA in the past were developed and tested primarily on white people, especially men,” Wu said. “But we have now the world’s largest iPS cell biobank — with cells from more than 2,000 people from a variety of ethnic backgrounds. Some are healthy, some have heart diseases, and others have rare or orphan diseases that are difficult to study. I don’t think any other university or single lab has access to a resource of this magnitude.”
That biobank is a big part of the reason Stanford Medicine leads the effort to explore the value of clinical trials in a dish. It exists in part due to funding from the National Institutes of Health and the California Institute of Regenerative Medicine and because of people like Wu, who straddle the medical world and the research lab — merging access to patient samples with the lab space and expertise needed to generate such a resource.
“I am a physician scientist,” Wu said. “If I only studied mice, my MD might as well stand for mouse doctor. My goal before I retire is to figure out ways to improve our current system of drug development. Right now, time and money are the two biggest obstacles. If I can help decrease the time it takes and the amount of money it costs to get successful treatments to patients, then I will feel that I have made some type of meaningful contribution to biomedical science during my career.”