On Thursday, April 10, the FDA held its first public advisory committee on assessing safety risks of cell therapies derived from embryonic stem cells. In a hotel ballroom just outside Washington, more than two dozen committee members and invited specialists weighed in. Separated from the discussion area by a yellow plastic chain, were about 200 prominent academics, consultants and industry representatives, and members of the press.
Prior to the meeting, the FDA had released a 12-page briefing document outlining the safety questions to explore, particularly how cells should be characterized and assessed for safety before transplantation and how patients could be monitored afterwards.
The biggest concern is that once placed in human subjects, cells could proliferate and differentiate in ways that are harmful and uncontrollable, and that studies of human cells in mice and rats won’t reliably predict their safety in human patients. A transcript of the session should be available from the FDA in a few months. Here’s about 2,000 words of what I found salient. (Disclosure: I was fresh off a red-eye from San Francisco, and I’m writing this on the plane back.)
Previously, I ran around asking attendees for their top-level thoughts. You can read that here. The consensus of the advisory committee for cellular, tissue, and gene therapies seemed to me to be both “Onward!” and “Careful!”
The sections that follow are 1) company presentations 2) wanted: fortune-tellers for teratomas 3) why can’t a mouse be more like a patient? 4) cell products are special and 5) once a trial enrollee, always a subject?
Geron, Advanced Cell Technology, and Novocell present
Three companies presented at the meeting, and the strengths each emphasized highlight the general flaws in the field. Melissa Carpenter from Novocell said that cell therapies have already been shown to work in diabetes, so with a convenient, consistent source of cells, successful treatments will be more widespread. Jonathan Dinsmore from Advanced Cell Technology described work to treat visual impairment and emphasized that transplanted cells should have less capacity for dangerous side effects in the eye, the site is relatively isolated from the rest of the body and immune attack is much less likely. Also, the number of cells expected to provide benefit is small. Jane Lebkowski from Geron, which is developing therapies for spinal cord injury, emphasized the sheer amount of time her company had spent working with cell products. Developing a way to transplant cells so that they survived for more than nine months took a year and a half. Geron developed a list of markers they must find in their cells, and a list of markers they must not, and they have assessed many, many antibodies for use in their assays to eliminate cross-reactivity and so false negatives.
Wanted: fortune-tellers for teratomas
Embryonic stem cells are partially defined by their ability to develop into strange lumps called teratomas that contain cells characteristic of all basic tissue types. Obviously, that’s not what clinicians hope the cells will do in patients. Though cells will be differentiated before being transplanted to patients, the advisory committee worried that undifferentiated cells might slip through and start to grow. Teratomas can, rarely, generate cancers growths called teratocarcinomas.
A big question is how many transplanted cells it takes to get a teratoma. Geron’s Lebkowski presented data in which ES cells were spiked into a population of progenitor cells and rodents were monitored for teratomas for twelve months. If ES cells were fewer than 5 percent of the total of two million transplanted, researchers never saw teratomas. There are lots of variables to be considered, though, site of implantation, whether cells are grown as aggregates, the cell line, and the primary cell population.
One committee member suggested that the differentiated cells in the transplant might suppress tumor formation. And so another question was whether the risk of teratomas depended on the absolute number of undifferentiated cells or the percentage; Ken Chien of Harvard Medical School and other committee members suggested this could be addressed by developing teratoma assays in large animals. Some committee members voiced support for developing large-animal assays, either putting human cells into animals or even (much more difficult) creating new species-specific cell lines to test. Results of such studies would be desirable, but requiring companies to develop a cell product other than the one intended for therapy seemed to be considered unreasonable. Some questioned whether the results of either sort of assay could be reasonably applied to humans.
Chien questioned whether studies in large animals need to use cells from the same species. “We could do all this and still leave the safety issues unsolved.” Still, Chien called for the development of more assays that could detect when a teratoma was growing. “We should not settle. We have not been aggressive enough in developing these assays.”
Benign tumors that develop in the brain or spinal cord would be much more dangerous than elsewhere, and the committee wondered if it should be less worried about procedures that would transplant cells to less vulnerable areas. One committee member said that the standard for tumorigenicity should be more rigorous than for toxicity for other toxicities that could result from cell therapy because the tumorigenicity issue was unique to embryonic stem cells and problems there would “be disastrous to the field.”
“We’re all talking about teratomas as if this were the only adverse events we expect to see,” Doris Taylor of the Center for Cardiovascular Repair at the University of Minnesota said. She emphasized that ES-cell-derived products is an even newer field than adult stem cell therapy and so other, unforeseen dangers might be more relevant. In particular, committee members wanted to know how progenitor cells might proliferate dangerously or even induce native tissue to grow aberrantly.
Why can’t a mouse be more like a patient?
Patients live longer, are bigger, and will likely be on immunosuppressive and other drugs. Human cells like to grow into bigger structures than mouse cells; introduced human cells will interact with native human tissue differently than with animal models of disease. It quickly became apparent that the necessary animal experiments would depend on the disease, cell product, and potential risks and benefits (an experimental therapy for patients on the brink of death may not require 5-year monitoring studies in pigs). Committee members sparred here, arguing for either the necessity or irrelevance of various animal models in various situations, and whether it made sense to monitor animals for a set period of time or for their entire lives.
Many factors need optimization to make cell therapies safe and effective, emphasized In his introductory talk before the committee, invited speaker Ole Isaacson who studies neurodegeneration at Harvard Medical School. “Cell concentration, dose, delivery, and implantation don’t sound very exciting to an academicians,” he said, “but are very important.”
All the more reason that experiments must be rigorous. For example, control studies for these experiments should not just mean injecting animals with saline solution but injecting them with other populations of cells, said one committee member. Ideally, researchers should be blinded so that they don’t know which animals receive the potential cell therapy.
“Preclinical studies have to be clinically relevant,” said Taylor. No one argued with that principle, but no one seemed to agree on how to apply it either. Depending on who was speaking about what, additional animal studies were either unnecessary barriers or the most responsible way to move to humans.
“It’s easy to say we want animal models with integrity to human diseases, but the truth is they are just crude approximations,” said Salomen. “What a sponsor should do with an animal model is answer specific questions that approach what they will do with a patient.”
“We need to be careful not to add an extra burden,” Taylor said, adding that the committee also had to recognize that no one know how long it will take for cell transplants to display adverse events. And it’s not just the cells that need to be considered; the age, sex of co-morbidities of patients will matter too.
Cell products are special
Small molecules interact chemically with cells; their presence in the body fades over time. Cells interact chemically and physically with cells; they travel through the body; they may persist for years and even expand in numbers. They behave differently at different densities and in different tissues.
While fully differentiated cells that are incapable of further divisions might behave most predictably, progenitors might be the only ones capable of integrating into degenerating tissue and surviving, particularly in the brain.
Committee members suggested repeatedly that a mixture of cell types is not only unavoidable but desirable for cell therapies. ( This is an issue that the leaders of CIRM raised in a recent commentary. ) Everyone accepted that the final cell products would be mixtures of cells. While heterogeneity can be tolerated, they said, it must be consistent from batch to batch.
Committee members stressed the need to use precedents from other cell therapies whenever possible. Matthew Allen of the Department of Veterinary Clinical Sciences at the Ohio State University urged members to remember which issues had already been hashed out and “only make alterations if we think the risk-benefit ratios merit it.” After all, other living cells are already being moved into patients without the extensive characterization being considered for ES-derived products. “This can’t be dissociated from the cell product and initial composition. Maybe we set more rigorous guidelines for understanding the composition of the product and then come up with a few assays that have to be employed.”
Both the starting material and the final product must be characterized for cell therapies derived from embryonic stem cells. For embryonic stem cells, that means assessing contaminants, genetic identity, and genetic stability. Committee member Meri Firpo of the Stem Cell Institute at the University of Minnesota suggested that cell banks were making useful progress with doing this. (Incidentally, the April issue of Cell Stem Cell has two reviews on stem cell banks.)
Genetic stability is important because genetic changes could affect tumorigenecity, but the size of genetic changes that should be monitored was unclear. Gross chromosomal abnormalities can be detected readily; smaller insertions and deletions may need to make up a large population of a culture before they can be detected. Ultimately, companies will be responsible for making sure their lines are characterized. One suggestion was that these products should be limited to early passage numbers.
An invited speaker, Jeff Bulte of Johns Hopkins University suggested several techniques to monitor cells that might be applicable for monitoring stem cells in vivo, mainly in animals, but also in patients. The committee members were highly interested, but frustrated by the number of potential artifacts and other limitations. For example, the half-life for the only FDA-approved cell tracker is only three days.
Once a trial enrollee, always a subject?
Committee members worried about their inability to remove cells from patients if they started awry, but a proposal for inserting a “suicide gene” into cells intended for therapy won little enthusiasm; one committee member suggested it could introduce as many risks as it eliminated.
“We should not create too many standards,” urged Jeffrey Chamberlain of the University of Washington School of Medicine. “A quarter of people who get the cells will die of cancer anyway.” (He also suggested that all cells come with a DNA profile.)
“We’re putting in those cells, and we don’t know what they’re going to cause in terms of trouble, so we need to be broad in terms of what they’re going to cover,” said one committee member. A patient registry would potentially allow clinical trial participants to be tracked long past trials’ end.
Kurt Gunter of Hospira suggested that following patients for a lifetime is “a lot to ask for.” He noted that when considering trials with retroviruses, patient follow-up was put at 15 years.
Richard Chappell, a biostatistician at the University of Wisconsin Madison Medical School tried to impress upon the committee how little information clinical trials can provide; no finite number of patients can demonstrate zero toxicity, he reminded them.