Angelo DePalma Ph.D. Writer GEN
Streamlining development through data-sharing.
A number of cell and gene therapy researchers are attempting to shape regulatory guidance of the field on two critical issues. The first involves reducing FDA-mandated animal studies of biodistribution without increasing risk to patients through identification of well-characterized vector subtypes and dose ranges. The second issue relates to the effectiveness of mouse gene integration models and the possibility of reducing the number of these studies as well.
Terence R. Flotte, M.D., dean of the University of Massachusetts Medical School, is an expert on vector platforms. Dr. Flotte’s research has focused on gene therapy based on the adeno-associated virus (AAV) platform for pediatric genetic diseases including cystic fibrosis, alpha-1 antitrypsin (AAT) deficiency, type I diabetes, and disorders of fatty acid oxidation.
Dr. Flotte became involved in the “standardized pathways” idea through his role in a group that represented ASGCT on creating platforms to promote AAV vectors in diseases that were not yet being addressed, but which were similar enough to those under investigation. “We were looking for a route through which such trials could move forward quickly,” Dr. Flotte says.
For example, AAV vectors have established clinical efficacy in Leber’s congenital amaurosis (LCA), a rare, recessive, single-gene disorder that leads to retinal dysfunction and visual impairment at an early age. For this disease the gene RPE65, when administered with the AAV vector at a certain dose through sub-retinal intraocular injection, restores vision. At least one company and several research groups are moving forward with this treatment.
“That’s great for LCA, but more than twenty similar, rare genetic diseases affect the retina,” explains Dr. Flotte. “Why not just plug in a different gene and attempt to treat them? [I want to help] facilitate parallel development of products for single-gene disorders.”
This would require a degree of transparency and cooperation among researchers, regulators, and funding agencies. The goal would be to create what Dr. Flotte refers to as a “clearly identified, relatively streamlined packet” of preclinical studies with which funders, regulators, and investigators would be comfortable, and which would grease the skids for parallel clinical studies on single-gene diseases. NIH, says Dr. Flotte, is “very interested in this concept.”
This approach has been somewhat inhibited by FDA’s oversight structure, which focuses more on approving or disapproving clinical projects, rather than providing carte blanche for conducting similar studies. Still, says Dr. Flotte, the agency could provide validity to this basic idea where sponsors can demonstrate similarity to either approved products or well-established clinical-stage treatments.
The comparison might be founded on the right combination of similar serotype, dosage, administration route, and gene delivery platform. “In other words, what would be the minimal package to get a treatment into the clinic, based on duplication of what appears to be a successful use of this platform,” adds Dr. Flotte.
What this represents is a standardized set of preclinical studies that demonstrate gene integration and lack of toxicity. To borrow from the small molecule pharmaceutical lexicon, the discovery and proof-of-concept phases are pared down significantly. Clinical development is expected to be similar in scope to current studies.
Whither the Mouse Model?
Frederic Bushman, Ph.D., professor of microbiology at the University of Pennsylvania, studies DNA integration from vectors into humans. Dr. Bushman’s research focuses on understanding the transfer of genetic information between cells and organisms. “If you take bone marrow cells from humans, transduce them with an integrating vector, and put them back in, the stem cells will output blood cells to the periphery,” he says.
Dr. Bushman samples these peripheral cells for sequence integration sites to determine the number of different stem cells giving rise to circulating cells, and to characterize those cell populations. “We look to see if they have any scary-looking integration sites near genes that have been associated with cancer,” he tells GEN.
Dr. Bushman refers to the first SCID-X1 trial, which treated twenty pediatric subjects at two sites. All subjects benefited to the extent that T cells that were missing due to SCID-X1 deficiency returned. But of those twenty critically ill children, five developed leukemia and one died. Four subjects responded to chemotherapy and continued benefiting from the gene therapy. “Given how few options these children had, and how sick they were, you can call that a big success,” according to Dr. Bushman.
Despite this achievement and subsequent improvements in vector technology, SCID-X1 continues to serve as a warning for gene therapy trials. Early vectors contained strong enhancers of tumor-promoting genes that researchers have removed. Another trial, for ADA-SCID, showed no adverse events despite using closely related vectors, and recent trials with lentiviral vectors have not shown serious adverse events. “However, some other trials did have problems. It’s not a universal phenomenon, but still it’s something everybody thinks about,” continues Dr. Bushman.
One particular area of interest in the gene therapy arena involves preclinical studies, especially murine models. Specifically, what preclinical models are needed, and do mouse studies tell us anything worth knowing about human toxicity?
“There are questions about how valuable these difficult, expensive, and time-consuming models are in understanding human gene therapy,” adds Dr. Bushman. “Because of their difficulty and expense, they represent a significant obstacle to the forward motion of human gene therapy.”
It remains unclear, notes Dr. Bushman, that mouse studies provide enough useful information to justify their time and cost. He suggests that the most effective way to fast-track the proof-of-concept stage for gene therapies is to apply clinical experience retroactively to validate mouse models.
“The question is what preclinical data do you need, particularly as we continue to accumulate more clinical data that’s directly relevant,” he says. “Can we circle back and ask which preclinical tests were most useful, and which didn’t really help?”
Unintended Consequences?
David A. Williams, M.D., chief, division of hematology/oncology at Children’s Hospital Boston and director of the gene therapy program at Dana-Farber/Boston Children’s Cancer and Blood Disorders Center, was part of an international team to develop a safer vector for SCID-X1 treatment. Dr. Williams presented an interim analysis of a clinical study using the improved gamma-retrovirus vector at the ASH conference in New Orleans last fall.
One of Dr. Williams’ research interests involves the predictive ability of mouse models for human gene therapy clinical trials. “FDA requires many of these expensive, time-consuming studies,” he says. “The question is whether they are reliably predictive. One could question whether it makes sense require those studies at all. FDA is itself questioning this.”
Dr. Williams is “as sure as one can be” that the leukemia cases in SCID-X1 arose from the viral vectors, in which the replacement gene was expressed in a region known as the long terminal repeat, which contained a potent enhancer as well as the promoter. “Because of this very strong enhancer, genes close to the insertion site were expressed even when they should not have been. The enhancer overrode the endogenous regulatory sequences of the human gene,” explains Dr. Williams.
Moreover, the time it took for leukemia to develop, around three years, pointed to an initial event which, while not by itself responsible for leukemia, induced additional genetic abnormalities that led to the illness.
Dr. Williams’ team redesigned the vector to eliminate the implicated sequence. “It sounds simple, but we had to redesign several characteristics of the vector, including how the correcting gene was expressed,” he says. Since the sequence was relevant to how the virus was manufactured, the production method had to change as well.
Will future trials in gene therapy always be haunted by the specter of unintended consequences? “Yes and no,” Dr. Williams says.
In the initial SCID-X1 trial the mouse virus used to make the vector causes leukemia in mice, but since the implicated genes were removed—and the insertion rate not very high— researchers did not expect the rate of leukemia that they observed. “In retrospect, it’s easy to say everyone should have expected this, but at the time it was believed unlikely that we would observe leukemia-transforming events,” continues Dr. Williams.
Still, he notes, mouse studies may not predict side effects in humans. “The only way to know is through human testing.” Ethical considerations are complex for such studies, but in the case of many pediatric genetic diseases the alternative to treatment is often the death of the patient. “The standard for Phase I trials moving forward is at least some possibility of improvement,” he points out.
Scientist at work at the Core Manipulation Cell Facility at Dana-Farber Cancer Institute. [Photo by Sam Ogden, Dana-Farber Cancer Institute]
Data Leveraging
Joy Cavagnaro, Ph.D., president of Access BIO, which specializes in preclinical pharmacology and toxicology studies, agrees that further development of gene therapy can gain much through data sharing. However, she prefers to apply “targeted” rather than “abbreviated” to studies enabled by data-based collaborations. “It’s more about leveraging data than having a standardized protocol,” she says.
When Dr. Cavagnaro worked at FDA she was part of a team that helped formulate ICH guidelines on preclinical safety assessment for biotech products. That experience drove home the need to examine complex therapies case-by-case, rather than automatically adopting platform technologies.
“There are many unknowns in gene therapy, and lots of questions to answer. You can’t have a standard study design,” she explains, regardless of how close protocols appear to be. She believes firmly in learning from previous studies, necessarily duplicating them exactly.
The rub, given that FDA does not provide “competitor” data, is creating and maintaining a robust database for sharing clinical data and research experience. Several databases have cropped up, such as National Gene Vector Laboratory (NGVL), which was replaced by the National Gene Vector Biorepository and various offshoots. Then there is the Genetic Modification Clinical Research Information System (GeMCRIS), which provides access to information on gene transfer trials.
The databases generally suffer from a lack of active participation, which is not surprising given the competitiveness of both academic and industrial developers of new therapies. Another issue is maintaining these resources, and making their data readily available.
Dr. Cavagnaro notes another issue, related to the academic nature of gene therapy. “With small molecule drugs you pick a candidate and advance it into development. Maybe you have several backups as well, but you stick with the lead. In gene therapy the ‘product’ is a moving target. It’s constantly being tweaked.” That’s not necessarily bad, adds Dr. Cavagnaro. “Deciding to end research and begin product development is difficult, especially for academics.”
