Imagine a New FDA
Say a drug, a device, or some other health product has come out of a lab with its efficacy and cytotoxicity demonstrated in vitro and perhaps in vivo in animals or in silico; the mechanism of action has been elucidated and a patent was filed. At this point we would be looking at IND approval for the drug and then seven plus years until approval, during which millions would be spent on clinical trials.
Is the Phase I/II/III/IV model really the best for a wide variety of potential phenotypes? How many departments do we need?
I envision an FDA with a unified, gradual pathway to market (removing confusion about which procedure to use for unusual submissions) and an extensive, linked knowledge base. Starting at the IND approval point (or its equivalent), a new product would be tested in its smallest target patient population—a few patients, extensively characterized with genotyping and other biomarkers—that most closely matches the nonpatient test conditions.
If the drug was tested on cells with mutations x, y, and z, the patients would ideally have similar genotypes, at least in relevant loci. “Relevant” for drugs might mean cytochrome P450 genes, which play a prime role in drug metabolism and biological half-life. A mutation in one of these could spell the difference between life and death for a given dosage. Careful measurement of biological indicators can give early warning of adverse effects, and, as dosage is ramped up from nontherapeutic “safe” levels to therapeutic levels, efficacy can be measured using these indicators.
More patients can be added as safety and efficacy accumulate more evidence in support. And at some point, when the patient population is no longer expanding exponentially, the product will switch over from experimental to marketable. The patients get their drugs sooner, and Phase I/II/III data is wrapped into one and returns sooner, allowing faster approval.
If adverse reactions are detected in the post-approval monitoring, these reactions can be analyzed to identify risk factors, the mechanism examined, and measures taken to prevent at-risk cases from receiving the treatment, all without interrupting the treatment of those not at risk. If further investigation is needed, researchers would be able to request additional information about the participants through questionnaires. The collected data can then be applied to similar compounds, allowing improved formulations to hit the market sooner.
This would truly be personalized drug (and device) development, and it relies on several technologies: secure sharing of personal health records, cost-effective DNA sequencing, better instrumentation and biological indicators, and connected research and clinical trial databases along with software to analyze the whole mess of data. I would argue that these technical developments will not be the limiting factor.
The challenge lies in how organizational frameworks will have to be laid out. Patient rights in this new context will also have to be evaluated. Additionally, exclusivity will have to be adjusted to ensure incentives are preserved.
Now is the perfect time to start figuring this out, with a pilot case of flexible trials (I-SPY 2) and several technological improvements in progress. Will the FDA fix the critical path in time for the next race?