Regulatory Editor

Francis Collins, director of the U.S. NIH, said in a public forum that his agency is "really bullish" about precision medicine. Precision medicine requires mounds of data to be viable; however, the necessary data may soon be available. Collins said the NIH's All of Us research program has drawn the interest of more than 300,000 willing participants to date, adding that the target enrollment of 1 million should be accomplished before the end of 2022.

Collins gave the keynote speech at the 2019 FDA Science Forum, noting that he has been witness to "an extraordinary partnership between NIH and FDA." While the NIH is enthusiastic about precision medicine, researchers "need a lot of data about a lot of people" to make precision medicine a clinical reality, Collins said. In terms of enrollment, the NIH initiative is aiming for rural communities and the economically disadvantaged, a focus Collins said will help ease disparities.

There is a need to guard enrollee data and privacy, but unlike the participants in most clinical trials, "they will be getting their data back" from All of Us, Collins said.

Collaboration a focus

Much of Collins' speech highlighted the collaboration between the NIH and FDA. He noted that he has been part of the FDA-NIH joint leadership council since being appointed in 2009. In a reference to the churn at the FDA commissioner's post, Collins quipped, "somehow, FDA keeps changing, and I'm still here." The remark was especially poignant given the rumors that the Trump administration is considering at least one candidate to take over as the permanent FDA commissioner. Acting Commissioner Ned Sharpless, who could not make the meeting, took the job despite his work as the director of the National Cancer Institute, which at present is being overseen by Acting Director Douglas Lowy.

Collins reviewed some of the challenges facing the two agencies, such as gene editing and CRISPR, as well as neurotechnology and the BRAIN Initiative. Of the latter, Collins said, "this is going to have some very significant impact in the field of diagnostics." Also among the programs Collins reviewed was the Partnership for Accelerating Cancer Therapies, a five-year, $220 million pre-competitive effort that enjoys the support of the NIH, FDA, Foundation for the National Institutes of Health and 12 companies. Collins said this program has accelerated immunotherapies for a number of cancer types and patients. He did note, however, that solid epithelial tumors have proven resistant to immunotherapies. He described immunotherapy as an attempt to "take the immune system to graduate school and teach it what to look for."

Also of interest for Collins was gene therapy, about which he said, "the exciting thing is, it's really starting to work, and some dramatic examples of that" are in clinical use. He said the NIH's focus is on resolving the barriers to basic research, particularly for ultra-rare diseases, adding that he and Peter Marks, director of the FDA's Center for Biologics Evaluation and Research, are of the view that manufacturing science must be improved as well, given the high cost of producing some therapies. While the ex vivo approach to gene therapy largely has taken center stage, Collins said the NIH also is focused on developing the science behind in vivo approaches, noting that this is more challenging in some respects.

The NIH Heal (Helping to End Addiction Long-term) Initiative, Collins said, is looking for scientific solutions to both opioid addiction and the development of alternative analgesics. He said the agency had expected it would be able to access $500 million in fiscal 2018, but that the budget dilemma in Washington impeded the agency's ability to review proposals before the end of that fiscal year (FY). Consequently, the NIH will spend $994 million in FY 2019, some of which is carryover from the FY 2018.

In silico study more economical

Among those who presented at the forum was Aldo Badano, laboratory leader of the Office of Science and Engineering Laboratories within the FDA's device center. Badano described the Virtual Imaging Clinical Trials for regulatory Evaluation, or VICTRE, Project, the first all-silico imaging clinical trial. "We've reached a point where some of these computational modeling tools are very, very advanced," Badano said. He advised that while this approach to clinical trials is less costly in terms of time and money, the regulatory decision made with the help of in-silico tools "has to be at least as good as the ones we're making today." He said there is an interest in regulatory efficiency as well, adding that simulation eventually will become an essential component in regulatory review.

VICTRE was intended to demonstrate that computational modeling can play an increasingly predominant role in regulatory review of imaging systems. The agency examined a number of completed clinical studies and settled on the pivotal trial for the PMA application for the Mammomat digital breast tomosynthesis system from Siemens Medical Solutions Inc., of Malvern, Pa. Badano said the VICTRE effort required the use of in silico phantoms, a simulated imaging system and a machine reader. The FDA approved the Mammomat in 2015 – about the same time it began internal discussions of how it would approach this project.

The Siemens study enrolled more than 400 subjects and relied on more than 30 radiologists at six sites who read the images. The trial took six or seven years, Badano said, noting that the VICTRE study made use of a model that produces random images of the dummy breast as well as random lesions. The algorithm employed finite element analysis to produce a virtual breast as compressed by a conventional mammography machine. The FDA staff designed the trial for about 3,000 virtual subjects, excluding 14. Half of these dropouts were due to failed insertions of lesions into the dummy breasts, while the another seven were taken out for other reasons.

"Overall, the results are very consistent" between the two studies, Badano said. He added that the FDA study took only two years vs. the nearly seven needed by Siemens.

Badano said the VICTRE results suffer from "much less uncertainty" than the Siemens study due to the ability to repeatedly run the algorithm against the virtual images, although depiction of microclusters favored digital breast tomosynthesis. One obvious source of bias was that the FDA researchers knew in advance where the "lesions" were in the study articles in VICTRE. However, the economics of the virtual study were quickly obvious.

Badano also noted that Siemens had disclosed the amount of money it had expended toward the pivotal study for the Mammomat, adding, "we think at most we were able to do this with one-third of their expenditures." The cost of this kind of effort should fall over time, he asserted, noting that all the source code and datasets the FDA used in this trial are available. The actual trial duration was three weeks, Badano noted, adding that this kind of virtual trial "has great potential to reduce, refine and possibly replace clinical trials for regulatory evaluation."

No Comments