OIVD’s head of personalized medicine shares the agency’s current thinking on regulation of companion diagnostics.
FDA anticipates more and more companion-diagnostics submissions in the coming months and years. Currently no formal process for these submissions exists, but that may change in the near future.
To learn more about the current and future regulation of companion diagnostics at OIVD, IVD Technology editor Richard Park spoke with Liz Mansfield, OIVD’s director of personalized medicine. In this interview, Mansfield talks about the upcoming draft guidance for companion-diagnostics regulation, as well as the office’s predictions for personalized medicine’s future and advice for those submitting companion diagnostics for agency approval.
IVD Technology: What is FDA and OIVD’s current stance on and policy toward the regulation of companion diagnostics for personalized medicine?
Liz Mansfield: It is both OIVD’s and the therapeutic product centers’ stance that when a test is critical to making a drug safe and effective, that test needs premarket review by FDA. It needs to be cleared or approved and available when the drug or therapeutic product is approved.
Is FDA/OIVD planning to issue a guidance document on companion diagnostics? If so, when will the document be available, and will it be a draft or final guidance?
Yes, the three medical product centers—the Center for Biologics, the Center for Drugs, and the Center for Devices—are actually working together now on a guidance document discussing companion diagnostic policies. I don’t know when it will be available. We never know that for guidances, but it will be published as a draft.
Can you give us some idea of what the current status is for that?
Well, we would like to have it out as a draft at perhaps the beginning of 2011.
I presume, and please confirm this, that once it is published as a draft, there will be a comment period for the industry during which they can review it. And how long is the comment period?
It is usually 60 days.
And then after the agency receives the comments, it will take those into consideration, revise the guidance, and then publish it as a final guidance, correct?
That is our hope.
Do you have any information about when an actual final guidance will come out?
It’s always dependent on many factors that we don’t control. So no, we do not have that information.
What sort of issues are you currently trying to work out for this guidance? What are the real big issues and areas that you’re particularly focusing on and wrangling over?
The biggest issue that we want to get right is to make sure that we have adequately captured the tests that we believe require oversight without getting into tests that we don’t really want to apply additional oversight to. We are trying to make sure that we’ve made the right cut, that we’re looking at the right tests and not looking at ones we don’t need to.
What exactly do you mean by, “We’re looking at the right tests?”
There is a lot of testing used in medical practice. Many patients are on prescriptions now, and they may have testing related to that. Not all testing related to use of a therapy is what we consider a companion diagnostic, so we’re trying to write the guidance so that it adequately captures the ones that we think are companion diagnostics and adequately eliminates the ones that we don’t want to categorize as companion diagnostics. We are trying to develop the criteria now so that it would be clear when you looked at a test whether it was a companion diagnostic or not.
Has there been any sort of differences of opinion or philosophy between the Center for Drugs and the device centers of CDRH and OIVD on this matter?
We have not had any big differences of opinion. We have certainly had discussions where we each hear the other side, but I don’t think there’s any major difference of opinion going on.
Is there currently some sort of general policy or review process for a companion diagnostic at FDA?
There is no formal policy, but this is the process that we’ve been following: When a therapeutic product submission of any kind comes in, whether it is very early or late in the process, if within that package there is some mention of having used an IVD test to select a population or predict a response or something like that, the therapeutic product center will issue a consult to us. We will take a look at it and give our opinion on the submission. We frequently will provide advice directly to the companies involved. We work together closely with CDER and CBER, and we try to make sure that the pharmaceutical company understands what will be required of it regarding the diagnostic.
There are investigational applications; there are preinvestigational applications, so we’re interacting with the Therapeutic center and the sponsors through the whole process, as soon as we recognize that there’s a diagnostic and a therapeutic that are likely to go together. We have not seen many submissions for approval yet, but the way those are currently handled is that the therapeutic approval is going to the therapeutic center, and the diagnostic approval is coming to us.
We are consulting with each other to make sure that the decisions are agreeable to both sides and that the timing is as good as we can make it.
Please describe your partnership with the Center for Drugs during the approval process.
We have ongoing collaborations for particular drug projects that are moving through the drug development cycle. And then once the final approval comes along, it generally is separated into two individual applications, one to Devices and one to Drugs or Biologics—but we keep talking to each other.
And do the submissions for approval usually come from one company only, or are there co-sponsors? For instance, would the co-sponsors be drug company X and IVD company Y?
To date I believe it’s mostly been that there’s a pharma company with the drug and a diagnostic company with the companion diagnostic. It doesn’t have to be that way. The pharma company could submit everything itself.
We also operate in the opposite direction: If we get an IVD that comes in that references a drug, and we’re not aware of the connection between the diagnostic and the drug, we will get the Therapeutic center involved to make sure that we aren’t clearing or approving something that would constitute an off-label use of the drug.
Do you anticipate that the upcoming guidance document will significantly change the current process for evaluating and approving companion diagnostics?
I don’t think this guidance document is going to necessarily change anything about our process. It is just a policy that certain diagnostics will need premarket review and compliance with our regulations. Certain information will go onto the drug label; it doesn’t have anything to do with the process of how we work together.
We will gain experience over time. Every submission has been unique. And as we gain experience, we may be able to ask better questions and give more targeted advice up front. But unless somebody comes along and tells us they don’t think the process is working, I don’t expect it to change a lot.
And so far you haven’t received any messages from anybody saying that it isn’t working?
I haven’t. I think it is working. I think it is just that we’re inexperienced, and every different IVD-drug combination presents a new situation for us.
Does FDA encourage companies developing a drug and a corresponding companion diagnostic to validate clinically the companion diagnostic in the same clinical trials in which the drug is being developed, or would FDA expect a companion diagnostic to be validated in separate clinical trials?
We want it to be validated ideally in the Phase III drug trial. That gives us direct information about how the diagnostic works with the drug.
It appears a coordination between CDER/CBER and CDRH/OIVD has been lacking when it comes to companion diagnostics. Drug companies may desire to have a validated companion diagnostic in place at the time of their Phase III clinical trials, but they often have to rely on a research test or LDT for this purpose while the provider of the commercial diagnostic test is seeking OIVD approval via the PMA process. To what extent then does CDRH/OIVD provide input into the design of Phase III clinical trials to ensure that test results can be used to support a PMA test?
There are a couple of issues within that question. First of all, the test does not have to be approved prior to use in the Phase III trial. So if people are waiting for approval before they’re using it in the Phase III trial, then they need to talk to us because that’s not necessary.
We actually have been working pretty well with CDER and CBER. Again, this is something fairly new to us, so it’s not completely regimented yet, but I think we’re working pretty well.
And there is no reason to say that an LDT can’t be analytically validated or even clinically validated and put on the market for use with a drug. It just needs to undergo premarket review.
We provide as much input as needed into Phase III trials. You know, generally we can only comment on what we see, so if the company doesn’t tell us something, then we can’t help them.
Typically what we have been seeing is that at phase II a pharma company might figure out which marker it wants to look at as a companion diagnostic. They might do a little bit of work with it then. By the time phase III rolls around, that company should have an analytically validated diagnostic. If they’re going to use it, they’re going to try and clinically validate it in the trial. But there have been all kinds of variations on that.
What particular requirements is FDA considering if companies want the companion diagnostics included in the labeling for a drug, and vice versa? Is there a guidance document coming out on this matter, and, if so, what is the estimated timing for that guidance?
A diagnostic isn’t going to appear in the labeling for the drug unless it’s necessary for the drug to be safe and effective. So if you just come up with a diagnostic and you want it to be in a drug label, I think you’re out of luck unless you can convince FDA that it’s necessary for it to be there. It’s not something that you can just ask for. It’s got to actually have some bearing on the way the drug is used.
Could you talk a bit about the labeling for Warfarin regarding its companion diagnostic?
Yes, there is a recommendation on the label that you do testing. It’s not a requirement. It’s not a black-box label. It’s just a recommendation. There is, at the moment, at least one test that has been cleared for that purpose. There are tests that are cleared for other reasons that can be used. There are laboratory-developed tests that can be used.
The recommendation could either have been initiated by the drug manufacturer because it has new information about safety, or it could have been initiated by FDA because the agency has new information about safety. It could go either way depending on who has the information and who wants it done.
So conceivably, once companion diagnostics become more prevalent, once this guidance document gets finalized, these so-called recommendations to physicians about the availability of companion diagnostics could become more common on drug labels?
There has to be evidence that the test actually provides additional information about the use of the drug. Somebody’s got to generate that evidence. For Warfarin and others, studies have been carried out outside of the pharmaceutical companies, as far as I know.
So as more information becomes available in the research literature and so on, yes, it may become more common. But we’re not going to do it without evidence.
Is FDA/OIVD more concerned about approving companion diagnostics for certain disease areas such as oncology versus, say, arthritis or Parkinson’s disease?
A companion diagnostic is a diagnostic device that’s critical to safe and effective use of the drug. And so whether it’s intended for oncology or Parkinson’s disease or rheumatoid arthritis doesn’t make any difference. It’s how it’s tied to the use of the drug that we care about.
Has there been among the companies submitting companion diagnostics for approval a particular focus on certain disease areas that you have seen?
Oncology tends to be at the forefront of innovation because it’s easier to identify the correct population. Most cancers have specific genetic changes associated with them, and the drug approval process can be more rapid sometimes because of accelerated approval. So we’ve just happened to see more oncology products, but I don’t think it’s because there has been a particular focus on oncology. I just think that’s where people have found biomarkers that are useful.
What about for cardiology? That’s also another big area. Have there been any recent submissions in that area?
I can’t tell you specifically what we’ve had submissions for, but there has been activity in the cardiovascular area as well.
Can RUO IVDs be used as companion diagnostics? If not for reference in the drug labeling, then what about for Phase I or Phase II clinical trials?
RUOs are intended to be used for research and never for clinical diagnostic use, so RUOs shouldn’t be used in clinical trials at all. If you want to use a test in a clinical trial, it should probably be under the investigational label and have some design control and some analytical validation and so on.
But RUO means for research use only, not for use in diagnostic procedures, so we actually frown on people using them for other purposes.
How will FDA/OIVD handle the regulation of companion diagnostic or personalized-medicine LDTs that are assembled, tested, and used at each site?
Well, once a test is approved, as long as you show reproducibility you can send it or sell it wherever you like. So I don’t foresee a problem there unless somebody comes in with an LDT and says it’s only going to be used at a single site and we say, “Okay, you don’t need to do reproducibility because it’s never going to be run anywhere else,” and then they take it somewhere else.
I think the premarket review process is going to solve that problem of moving it between labs. But I have to tell you that we don’t, in fact, consider an LDT that’s developed by one lab and transferred to another to still be an LDT. So that whole premise is a little bit weak.
How will the expansion of the regulation of personalized-medicine LDTs at FDA/OIVD be funded, and is FDA/OIVD expecting to add personnel? If not, how will this affect FDA/OIVD’s timeline for the review process?
Our timeline is statutory and agreed to up front. It is what it is, and we’ve got to stay within our timeline. We intend to undertake this process by steps. We know what our resources are. We know what we can do and so we don’t intend to overburden ourselves, and whatever we need to do to make sure that doesn’t happen is what we’ll do.
The more prevalent, well-known companion diagnostics are for drugs such as Herceptin, Gleevec, and Warfarin. What sort of lessons is OIVD taking from these companion diagnostics in terms of moving forward with the regulation of companion diagnostics for personalized medicine, as well as for the process of developing the guidance document?
Our regulation is pretty much the same regardless of whether it’s a companion diagnostic or not. We look for the same performance characteristics and so on. I think the biggest problem we’ve had with companion diagnostics that have come along, perhaps after clinical trials are done, is that there aren’t enough clinical-trial samples to validate the test as well as we would like. Sometimes the samples won’t be available if they need to bridge to a new diagnostic or something like that. But in general our regulations are the same whether it’s a companion diagnostic or not.
Is any effort being made to address this issue concerning this lack of samples for the clinical trials?
Well, we’re telling people, “Try to get samples from everybody and save them.” But I think that’s generally good advice that’s been out there for a couple of years now.
Please explain how it is that evaluating an IVD device is generally the same as evaluating a companion diagnostic, from FDA’s perspective.
What we look for in IVDs is analytical validity and clinical validity. Analytical validity is, How well does this test measure what it’s supposed to measure? That could be a mutation. That could be an expression pattern. That could be a protein. It doesn’t matter. A companion diagnostic is measuring a biomarker, and a regular IVD is measuring a biomarker, so that’s all the same.
The clinical validity is a little bit different. Clinical validity for a companion diagnostic will generally come out of the drug trial, where they’ve used this diagnostic to, let’s say, select patients.
And they say that the biomarker they’re measuring can distinguish between people who will respond to the drug and people who won’t respond. Well, if they run the test on the patients and then they see who will and won’t respond, then we see how well the diagnostic works for picking people who will and won’t respond.
For other kind of diagnostics, instead of looking at who responds we look at who actually has cancer or who actually has Parkinson’s disease or whatever the illness is. So it’s not really very different.
To this point, in the various submissions that you’ve been involved in and that you have seen, what do you think are some of the biggest challenges that companies are encountering in putting together the submissions for companion diagnostics for personalized medicine? And what are the biggest challenges that you’ve encountered in working with these companies in developing their submissions for these companion diagnostics?
Let’s see, the biggest challenge I suppose so far is that the pharma company may not engage diagnostic regulatory advice early enough. And so they don’t set up their trials to be informative, or they don’t have the right test in place at the right time. So most of the trouble is in the planning stage.
Is that issue something that you will address in the guidance document?
No, the guidance document is just policy that does not say anything about planning.
But if companies come to you for consultation and suggestions, I presume that is something that you would tell them—to start as early as possible?
As personalized medicine becomes more and more prevalent, which is what many people are anticipating, do you expect that FDA/OIVD will get more and more submissions of this nature?
That is our expectation. We’ve been seeing the number of submissions and consults go up every year, and it doesn’t seem like there’s any idea that we should abandon this. So I expect we will see more. We are in the fourth year of our tracking now, I believe, and every year the number of submissions has gone up.
On the user side, what sort of response or questions from users such as physicians, hospitals, and clinics has FDA received?
There are some people who think it’s great, and there are some people who aren’t sure what to do with the results. And there are some people who think that they do pretty well without a test. It’s varied. There is a huge variation in opinion out there.
Does any lack of understanding of companion diagnostics still exist among healthcare providers?
I don’t think there’s a lack of understanding. I think there’s sometimes a lack of understanding of how to use the test result, but you don’t need to know what a diagnostic test is in order to know that the label says you should get this test before you give this drug to the patient.
What suggestions does FDA/OIVD have for IVD companies that are currently in the process of developing companion diagnostics for personalized medicine?
Talk to us early. Talk to us before you start spending money on studies, because once you start spending money you don’t want to be doing the wrong study.
Elizabeth Mansfield is director of the personalized medicine staff in the Office of In Vitro Diagnostic Devices in the Center for Devices, FDA (Rockville, MD). She can be reached via Erica Jefferson at firstname.lastname@example.org.
to post comments