NEW YORK (GenomeWeb) – The US Food and Drug Administration last year issued a long-awaited draft guidance outlining the principles for simultaneously developing a drug and diagnostic — the paradigm that the agency has maintained is the ideal way to get a personalized medicine product to market.
The draft document, entitled "Principles for Codevelopment of an In Vitro Companion Diagnostic Device with a Therapeutic Product," is intended to be a practical guide for companies developing a treatment that relies on a test for its use and to help FDA staff reviewing such products. Although the FDA has recognized that simultaneous development and launch of a drug and test may not always be possible, the agency notes in this guidance that this will be the path for sponsors in most circumstances and, with this in mind, outlines the regulatory requirements for clinical trial design and statistical analysis that drug and diagnostic firms should consider.
It is unclear when the FDA would finalize this particular draft document, since it isn't on the agency's priority list for finalization in 2017, and the Trump Administration has issued a number of executive orders to lessen regulation at government agencies. Still, codevelopment is a complex process in which industry players and FDA have limited experience and all parties would all benefit from more discussion.
Since the draft guidance was released, 15 stakeholders, including labs, diagnostic companies, and drug and biotech industry groups have provided the agency feedback on how the document might be improved. Stakeholder comments suggest a significant gap between how precision medicine is developed and delivered on the ground, and how the FDA would like to see it done.
Diaceutics, a firm that consults drug companies advancing personalized medicine products, has been tracking industry views on this draft guidance. Jordan Clark, managing director at Labceutics — a Diaceutics' division focused on creating a lab infrastructure that supports personalized medicine — discussed industry's sticking points in the hopes of advancing public dialogue on some long-standing scientific and operational challenges in the field.
Below is an edited transcript of the interview.
The FDA issued the codevelopment guidance last July, but it was a long time coming. Do the general principles outlined in the guidance reflect how drug and test developers are codeveloping products or do they reflect how the agency wants the companies to do it?
I think the draft guidance reflects how the agency wants the industry to be able to conduct these studies. I think that's evident in the sections addressing bridging studies and clinical cut offs for IVDs. These are ideas that have been heavily pushed by the agency, whereas in real life practice, clinical cutoffs, the contemporaneous development [of a drug and test], and bridging studies are very hot topics of debate, and we see many drug and diagnostic companies, and laboratories taking different views on them.
I don't think the FDA has taken into consideration what is going on in real-life practice, but is more about trying to push out their message, which I guess is what a guidance document should be doing. They want people to conform to their way of thinking to speed up regulation and are trying to drive that. But perhaps there needs to be a little bit more support for the diagnostic and pharma companies in terms of bridging from what's going on now to how the FDA wants the testing to be done, so the process can be streamlined and produce faster results.
Complementary diagnostics is a relatively new category advanced by the FDA that has received a mixed reaction from stakeholders. The codevelopment guidance doesn't mention complementary diagnostics, but by the time it was issued last year the FDA had already approved a few tests (see here and here) in this category. How important are complementary tests to pharma's drug development efforts right now and should FDA address this in the final guidance?
The FDA should definitely address this in the final guidance because this has been a really confusing issue for the prescribers of these drugs, knowing whether to test or not. We have done market research showing that 34 percent of oncologists didn’t understand the difference in labeling [with regard to companion and complementary testing] and 15 percent understood it incorrectly.
That causes big issues about whether or not they should be testing patients. We may find that patients are unduly being delayed treatment due to having to wait for test results. When it's a complementary diagnostic, patients can go on to therapy without a test, and its speeds up that process. It's a really big issue and it was a big shock to Diaceutics that [a discussion of complementary diagnostics] wasn't in this draft guidance.
This document is here to help codevelopment and help sponsors understand the data that needs to be submitted to the FDA. It would have really helped to understand where the agency draws the line in terms of a what is a companion and complementary test, and how industry can develop them in a robust statistical manner.
The FDA has used the term complementary diagnostic, but hasn't officially defined it. What should the agency focus on when defining this new category?
The statistical analysis and what the [drug and test] labeling looks like are key things the FDA does already regulate. When a pharmaceutical company is submitting the draft label … a lot of input is given by the FDA about what's allowed and not allowed. What's needed is perhaps more consistent terminology between companion and complementary categories.
Ultimately this comes down to the clinical trial design and the data submitted. What is the clinical trial design and the data that the agency is poring over to see if they're going to approve something with a companion or complementary test? What different statistics are used? Are there any cutoffs on the overall survival or progression-free survival endpoint that pushes them toward a complementary rather than a companion test? So far, these things are unclear for many people.
In your survey of stakeholder comments to the draft guidance, it seems that they are also questioning FDA's basic premise that a drug and test should be codeveloped and approved at the same time. One criticism in this vein came from the Association for Molecular Pathology (see pdf below), which asserted that the agency's requirement for contemporaneous approval of a drug and CDx doesn't allow companies to keep pace with innovation and would hinder patient access to appropriate care. This is also AMP's rebuttal to FDA's bid to oversee all lab-developed tests. Do you think that because the agency has backed off of regulating LDTs, groups historically against the agency's oversight are now challenging its authority to regulate companion diagnostics, even though the agency has said companion diagnostics are likely to be high-risk tests that need premarket review and has issued final guidance on the topic?
There were a number of comments from AMP that were rebuttals to FDA's intentions to regulate LDTs … In this case, the think the association made a good argument. For example, in the case of AML [acute myeloid leukemia], the latest phase 3 clinical trials have been 11 years in the making. If you used a technology that was available 10 years ago for your study, by the time the drug comes to the market, the test is not going to be fit for practice for the laboratories.
Technology moves very quickly. And as we are moving to an era where we are molecularly subtyping all of our [cancer] patients, we are segmenting them into smaller and smaller cohorts. That means our clinical trials have to recruit far more patients to get statistically significant numbers in these rarer mutations. That means our clinical trials are taking a lot longer when we look, for example, at AML and lung cancer and all the different molecular subtypes we have now.
There is a big risk that the technology used in the trial will be outdated by the time it's [ready for] market. This is something we'll see pharmaceutical clients go against and do more bridging studies, which again the FDA doesn't want. But it ensures that the life cycle management of a drug is less risky and pharma won't introduce outdated technologies that will never be used when the drug comes to market.
One of the trickiest parts of the codevelopment process is the early work when drug and diagnostic companies are trying to align their development timelines, making decisions about whether to submit an investigational device exemption, and whether to have pre-submission meetings with FDA's drug and device divisions. Industry players have had some experience now going through this process for various Rx/Dx products. Do they have any new ideas for making this process more efficient than what the FDA is suggesting in the guidance?
I think while some of the companies do have a lot of experience going through this process … we're still seeing that for many this is their first time. Even some big pharmas are only just launching their first companion diagnostics. That's something these companies are all trying to learn how to do.
From Diaceutics' perspective, the biggest recommendation we can give is to have early communications between the pharma and a diagnostic partner. Identify the correct diagnostic partner early on, engage with them as a partner and have very close collaboration with them, so you go to the FDA as a unified voice.
It's still a huge learning curve. And there are few in the industry with a lot of experience.
Now, getting into some of the nitty gritty aspects of the guidance: The FDA has been concerned with prescreening for some time. The draft guidance strongly discourages enrolling subjects to a trial based on a local test due to the problem of selection bias. But prescreening is happening and a reality of community-based oncology practices in the era of precision medicine. What do drug and test makers feel about this issue and what does the FDA says about this in the guidance?
The discussion about prescreening was one of the highlights of the draft guidance for Diaceutics. We know the FDA is against it, and they want the IVD to be used in the clinical trial … But when you have someone who is acutely unwell, these patients need to be treated within days. Acute myeloid leukemia patients, for example, are usually treated within a 48 hour period after diagnosis. It's very hard for a centralized lab to return a result that can allow these patients to be enrolled in a clinical trial in that situation.
We have to have pre-screening where local labs can meet those shorter turnaround times because they are close to the patient. This is a big issue if we want to drive recruitment in clinical trials, segment our patients by molecular lesions, and progress personalized medicine. There is going to be a need to do this prescreening so that it's actually useful for the physician. Otherwise we could find the uptake of clinical trials in these molecularly targeted therapies being a problem since the physician won't wait for the clinical trial because he has a very sick patient in front of him that he needs to get on treatment.
This is another point that AMP addressed in comments to the draft guidance. Coming from a laboratory background myself, I agree with them that laboratories are well regulated in the US … and the quality of results from laboratories, when they're properly validated and monitored using proficiency testing, is high.
If prescreening is happening in local care, then it behooves drug firms to do studies comparing the IVD and the local test. So, what is the debate when it comes to bridging studies?
The FDA allows bridging studies, but they would want the chosen diagnostic partner's IVD to be used to enroll all patients in the trial. But they understand there is a need for bridging studies. What this draft guidance doesn't do is give enough detail of the statistical analysis and the situations where bridging studies are acceptable and where they aren't, to allow the design of those studies.
Right now there is not enough guidance on how the diagnostic and pharmaceutical companies should conduct those studies if they are being used. One of the reasons for this is the FDA is trying to move away from bridging studies, so by giving little guidance on them, it makes it more difficult to do them. They're trying to drive people to simultaneous codevelopment. But in reality, we know bridging studies are being used and particularly in Phase III studies that take a long time and were set up five or six years ago, it's important to bridge.
Have stakeholders tried to suggest solutions, maybe ways in which local tests can be validated and used as companion tools? Has anyone tried to solve for this dilemma, where companies are investing money and time to develop an IVD that the companies will pretend will be used but won't really be used alongside the drug on the ground?
I don't think anyone went as far as to comment on that … I think we still firmly believe that the agency will continue to [reference] FDA-approved companion and complementary tests within the drug labels. In the real world, as you say, we see labs not using those companion or complementary diagnostics to test the patients. We have the FDA world and the real world and sometimes they don't marry up. I can't see them coming close together and being aligned in the near future.
The FDA also would like companies to pre-specify cut offs for tests. What does industry have to say about this?
One of the issues with having a clinical cutoff is that it lends itself very well to molecular lesions in EGFR or BRAF where you have a binary result — you're either mutated or wild-type, positive or negative. The situation starts looking a lot grayer when you start looking at immunohistochemistry and fluorescent in situ hybridization. If we take HER2 as an example, the guidelines from ASCO and AMP use a scale for IHC measurements … where 2+ patients are equivocal. That's not a positive or a negative, it's an equivocal result, and then you have to do a FISH test to get a confirmation based on the copy number. This is what the labs are dealing with every day. So, to put a clinical cut off on there, we're finding that very arbitrary and difficult from the labs' perspective because that's not how biological disease works.
This is no more evident than in the PD-L1 space. If we consider a cut off of 50 percent PD-L1 expression [as a way to determine who gets immunotherapy,] a patient that has 49 percent PD-L1 expression wouldn't get the therapy. At the moment, I just don't think we have the data showing there is that hard and fast line. This is where the FDA has tried to use the complementary diagnostic to get over this gray area, but instead of making it clearer and making it more black and white, I think they've made it even more complex. We have to think of these as biological diseases and there is a continuum. There is not a binary yes or no in some of this biomarker testing like the FDA wants everything to be.
Is there anything else that you wanted to highlight about the draft guidance that we didn't discuss?
I think the prescreening and clinical cutoff problems are important scientific discussions that need to be had. And there are lots of questions around timing and bridging studies, and the definitions of companion and complementary testing. Hopefully, this guidance will help us have those conversations … but I think we are still on that journey to better understanding. There will need to be more guidance and reviews, so we can get to a place that works for everybody — diagnostic companies, pharmaceutical companies, the FDA, and the general public.