On March 16, the US Food and Drug Administration rolled out its Critical Path Opportunities List 76 projects that it expects to help set priorities for the Critical Path Initiative, which it kicked off in 2004 to help speed the drug-development process [BioInform 03-24-06].
The list includes eight projects under the topic of "Harnessing Bioinformatics," which could be a key driver for bioinformatics efforts that are aligned with those recommended projects.
This week, BioInform spoke to Felix Frueh, associate director for genomics in the Office of Clinical Pharmacology and Biopharmaceutics at the FDA's Center for Drug Evaluation and Research, to get a better idea of what the list means for the field of bioinformatics.
Can you discuss the significance of the Critical Path Opportunities List? Bioinformatics is one of the topics singled out in this document, so what kind of impact might this have in terms of guiding research priorities?
The Opportunities List starts out with the bigger headings, one of which is, "Harnessing Bioinformatics Data Pooling and Simulation Models." In particular, under this topic, there are eight opportunities listed that pertain to the broader effort in bioinformatics. Now, those are fairly specific examples for how we envision bioinformatics to be used, such as modeling and clinical trial simulation, and so forth.
So in the broader context, you would almost see this as [falling under] the use of biomarkers in clinical trials and what informatics tools can help us to better integrate the use of those biomarkers into the trials. We're talking about adaptive trial designs and this type of thing on the one hand, but then also on creating better drug and disease models for predicting the benefit of using those biomarkers.
In other words, you have bioinformatics playing a critical role at various stages. First, you need to have and develop a disease model. Secondly, you need to have a drug and you need to be able to model the effect of the drug on the model that you create. And thirdly, you might want to bring in biomarkers of various different types and see how it affects the use of the drug in the model.
So you can envision, and I think it's in the opportunity list, a couple different ways how bioinformatics can be used in situations like that.
In addition to that, and I'm not sure if it's listed in the Critical Path Opportunities list, but there are a whole number of bioinformatics activities at the FDA it's more informatics and less bio to streamline the submission processes with electronic submissions and so forth, but also in the area that I'm most familiar with, which is pharmacogenomics, we have several initiatives to standardize the submissions, to come up with common formats and a common language that we can use in such a way that the bioinformatics piece that follows those submissions becomes easier.
For example, we're talking about streamlining the process of submitting DNA microarray data, where the bioinformatics piece essentially is downstream, but heavily dependent on the format and standards that are being used in the first place to submit data to us.
Does the FDA have any internal bioinformatics projects underway that might align with or overlap with any of the specific bioinformatics projects in the Opportunities List?
Sure. From the top of my head, looking at the list Modeling Device Performance, Identification and Qualification of Safety Biomarkers, Clinical Trial Simulation, Adverse Event Data Mining, and so forth in all of these areas, we have activities that are going on.
The Opportunity List I would describe as a call to arms. In other words, it's a priority list that FDA sees for pushing the field forward in these specific areas because we feel that they are most critical to improve the drug-development process.
So some opportunities you find are ongoing activities at the FDA, but others may not be. But definitely in the field of bioinformatics, I would say that we are fairly eager to make sure that we have certain aspects already covered, so we're not starting from ground zero.
For example, in the area of disease models, we've built a couple of models already and want to follow up with several more, and include additional information with the existing ones to improve those.
Some of the bioinformatics projects involve developing databases for sharing data. Is there a mechanism in place for developing and maintaining those databases? Is that something that the FDA would do, or is it more likely to be an outside party like the C-Path Institute?
I think it depends on the database, quite frankly. I can think of examples for each of the scenarios you mentioned. It depends on if the database is being built on, say, proprietary data that we receive in IND and NDA type of submissions. That would remain FDA internal. However, you could envision quite easily databases being developed with more or less public, let's call it pre-competitive, information that would benefit everybody and would be accessible to everybody. Building the infrastructure for those I would expect that perhaps not necessarily the FDA is going to do that, but a third party such as C-Path.
Some of the recommended databases would require pharmaceutical companies to share internal data, which is not something that the pharmaceutical industry has historically been willing to do. Do you have a sense that these companies will be willing to place this information in the public domain to support the Critical Path Initiative?
Sure. For example, what was announced at the same time that the Critical Path Opportunities List came out was the Predictive Safety Testing Consortium. … We started working on that probably a year ago, based on our very limited but nevertheless emerging experience with voluntary genomic data submissions. The notion from industry that many of these biomarkers, particularly in the preclinical toxicogenomics area, that we talk about and use are really not very useful if they're only developed and validated at one single place. So certainly from a regulatory perspective, we have no idea how good that information is.
So we felt that there is an opportunity for creating a pre-competitive environment for sharing this type of information, and while doing so, not just taking the information and putting it into a database, but working with it in such a way that members of this consortium actually take that information and validate it with their own technologies and methodologies. So in other words, it's a true cross-validation in the sense of somebody putting something in the pot, and others taking it out and trying to reproduce what the first member found. So there you have a database being developed for, for example, nephrotoxicity biomarkers, or liver toxicity biomarkers, or genotoxicity biomarkers and so forth, that hopefully at some point becomes a repository for information that becomes publicly available and can be used by many others in the drug development process or in basic research.
So you would envision a similar model for other databases, such as the ones in the list for adverse events or failed drugs?
Absolutely, and in fact we're talking about this already to use the very same model to develop an adverse events research consortium, because quite obviously, adverse events are of concern to everybody, and the same types of adverse events, or toxic endpoints if you want, are found across different drug classes and certainly across different sponsors, even if it's in the same drug class. So one would envision that companies would see this as information benefiting everybody if we have better means to address adverse events.
Regarding the simulation and modeling projects that are recommended, what kind of criteria will the FDA be using to validate the effectiveness of these approaches? Are there any examples where you're already seeing a payoff in using these simulation approaches?
We are working with a couple of companies under a CRADA, and I would say that the sophistication for those models is evolving. I don't think we have the one solution at this point that we would feel comfortable relying on for making regulatory decisions that have far-reaching consequences. Because we have large internal datasets from clinical trials from successful trials, but to some extent also from failed trials I would think the FDA is in a unique position to develop such models, and simulate certain events QT prolongation or whatever it may be. I might not be the right person to specifically answer that question with very concrete examples, but I'm not aware of any model that currently is being used or simulation that is currently being conducted that replaces any of the current conduct of, for example, clinical trials.
What it does, and it does that very effectively, is provide information that can be used for designing better clinical trials, for the evaluation of, for example, whether a drug should be given once a day or twice a day based on the pharmacokinetic and pharmacodynamic profiles, and so forth. So these models, to that extent, I think, have been already very successfully used here at the agency.
Now that this list has been available for a little while, what kind of feedback are you getting from the community?
First of all, everybody is saying, 'Oh finally,' once the shock has been absorbed that it's actually out. I would say the response is overall positive. I think many of the opportunities that are listed were sort of obvious. It starts with biomarkers and it continues with biomarkers, and then there is a little bit more about biomarkers I don't think that's a surprise to anyone pushing towards personalized medicine. This certainly is at the core of what we believe is important.
I think we've heard some surprise about the breadth of what's been presented. Also about the level of detail. I think the list is interesting from the perspective of being very general in certain cases, such as we need to find a validation pathway for biomarkers, and then very specific in other cases, such as mentioning individual biomarkers that need to be validated. For example, if you look at Topic 1, Opportunity 1, it says, 'Biomarker Qualification,' and then [Opportunity] 3 says, 'Role of Beta Adrenergic Receptor Polymorphisms in Asthma Treatments.' These are two completely different dimensions one is a very general aspect and the other is extremely specific.
So one is an opportunity that leads to immediate translation into something that is clinically useful. The other one might show a path that we can go down that is generally applicable and will be useful in many situations. I think both of them are justified to be there from the perspective of being an opportunity along the Critical Path.
So it's spelling out short-term and long term opportunities.
Exactly, and again, the list is intended as a call to arms: 'Here it is. This is what we believe is important to address along the critical path in order to develop better medicines faster.' So we certainly have gotten feedback, and I think we've gotten mostly positive feedback.