NEW YORK (GenomeWeb) – Based on the experiences of drug and diagnostic firms developing personalized medicines, the US Food and Drug Administration is refining its thinking on codevelopment of treatments and companion tests, follow-on companion tests, and complementary diagnostics.
In 2014, 45 percent of the drugs FDA approved were targeted, a dramatic increase from the early 1990s when 5 percent of approvals were for targeted agents. But with all these targeted drugs, there is also a need to identify the "intent-to-treat population" using diagnostics, Aaron Schetter, a scientific reviewer in the Division of Molecular Genetics and Pathology at the FDA's device center, said at a recent conference on companion diagnostics hosted by the Drug Information Association.
The FDA defines a companion diagnostic as a test that is essential for the safe and effective use of a drug. So far, the FDA has approved 23 drug/diagnostic combinations, according to the agency's own count.
In the final CDx guidance the agency issued last year, the FDA indicated that the ideal situation is when the drug and the test are approved at the same time. But the agency also recognized that this is not always possible.
"That guidance described the expectation for contemporaneous approval, but we didn't go into any details on how you might do that," Pamela Bradley, staff fellow at the Office of In Vitro Diagnostics and Radiological Health within FDA's device division, said at the meeting. "We realize that this can happen in any number of ways, at any point in the drug development process. It can happen early … and it can happen late."
A 'how to' on codevelopment
Ultimately the goal from a regulatory perspective is to align the development of the drug and test so that the safety and effectiveness of both are proven in the same study and the two products reach the market simultaneously, Bradley said. But industry experiences have shown that this isn't easy, and so the agency is in the final stages of writing a draft guidance that details the "how to" of developing a drug and test together. For years the agency has talked about releasing this guidance, and although FDA representatives at the meeting said it would be released soon, they couldn't provide a more precise timeframe.
The draft guidance will discuss how to plan ahead for test validation and try to clear up confusion around investigational use of IVDs in drug studies. "Most companion diagnostics used in [drug] trials will not have already been approved for [how] they are used in the trial, and therefore, they are investigational," Bradley said.
In such a scenario, the test would need an investigational device exemption (IDE), which test developers are increasingly applying for because IVDs are more readily used to stratify patients in drug trials. "The purpose of that regulation is to protect public health," Bradley explained. "We want some information that this test is doing what it says it's doing before it's put to use in the trial."
If the investigational test poses "significant risk" — for example it impacts trial accrual or a treatment decision — then the sponsor must file an IDE application that includes information on the test's analytical validity. The analytical validity and test cutoffs have to be locked down before a test is used in a drug study, Bradley noted. The FDA is planning to elaborate on these aspects in a draft guidance specifically on the use of investigational IVDs in drug studies.
The codevelopment guidance will also discuss the importance of banking and consenting samples — both marker-negative and -positive samples — along the course of a drug trial, so they may be used to address any diagnostic issues that arise later on. Over the years, the agency has repeatedly warned drug developers to plan ahead and collect samples, so if necessary in late-stage studies, they can use them to conduct the necessary bridging analysis showing concordance between a research assay and the CDx intended for commercialization.
Jolette Franco, regulatory affairs manager at Myriad Genetic Laboratories, said at the meeting that in her experience drug developers don't think ahead about banking samples when they're using an assay in Phase II studies. "Then they get done with the Phase II, they realize they have great data and they want to go to the FDA, … [but] now we have no samples for a bridging study" for a CDx, she said. "You should always plan for a bridging study from Phase II on and you've got to have extra samples for that, stored appropriately and ready to be tested on."
The codevelopment guidance will also emphasize the need to collect sufficient marker-negative samples and be vigilant for prescreening. At the meeting, several presenters discussed prescreening bias as a particularly difficult challenge for sponsors conducting personalized medicine studies.
Prescreening bias — a topic FDA has been cautioning industry about for several years — occurs when enrollment sites send patients to partake in a precision medicine trial knowing ahead of time that they have the marker of interest. This skews the patient population for the drug trial, hinders evaluation of patients who don't have the favorable marker, and it can make the companion test look like it's performing much better than it actually is.
"When we're running an assay, we want to get both the positive and the negative results for our testing," Douglas Robinson, global head of biomarkers and diagnostics biometrics at the Novartis Institute for Biomedical Research, said at the meeting. However, in cases where Novartis researchers have asked doctors if they can run eligible patients on its assays for clinical trial enrollment, the physicians balked, saying that that they didn't want to wait for the results from Novartis, that their local assay "is as good or better than the enrollment assay," and they have demanded that known biomarker-positive patients be enrolled in the study.
Robinson explained that this often results in insufficient accrual of marker-negative patients, whose samples are needed to accurately prove the ability of the CDx to distinguish between marker positive and negative groups, and show that patients tested in the trial represent the intent-to-treat population. Patients who are marker negative by local prescreening are often not recommended for enrollment and simply become "ghosts," he said. "We can kind of sense they're there because of the skewed distribution, but we know nothing about these patients."
Although this is a challenge for drug and test developers and the FDA in terms preserving the integrity of the data supporting medical product approvals, there don't appear to be any easy solutions. In the era of precision treatments, more and more patients are having their tumors analyzed at local treatment centers. If test results don't point to treatments that match the molecular features of patients' tumors, their next option is to find a clinical trial. Several sponsors noted during the meeting that despite efforts to discourage the practice, prescreening continues to occur at enrollment sites.
Complementary and follow-on tests
The agency's three most recent CDx approvals include: Ventana's ALK IHC assay to determine eligibility for Xalkori (crizotinib); Roche's Cobas KRAS test for Erbitux (cetuximab) and Vectibix (panitumumab); and Myriad's BRACAnalysis CDx for Lynparza (olaparib).
Roche's KRAS test "was an example of a follow-on diagnostic," Schetter said. The Roche test came on the market following FDA's approval of Qiagen's TheraScreen KRAS Mutation test in 2012 and 2014, for the same intent-to-treat population.
He defined a "follow-on" test as a diagnostic "intended to identify the same therapeutic indication as an originally approved companion diagnostic on the market," and noted that the field will see more tests of this kind. In reviewing such a companion diagnostic, the agency will assess its performance against the original CDx or a reference standard.
Given the pace of technological advancements in the diagnostics space, a clear path for follow-on companion diagnostics that offer alternative options and advantages over the original CDx can help drugmakers improve access to therapies. One barrier to showing that a follow-on CDx is comparable to the original CDx has been access to patient samples from the drug trial. Moreover, industry competitors don't easily agree to head-to-head comparisons of their technologies.
FDA is willing to be flexible on this front. "Not all follow-on diagnostics are going to have access to samples from the original clinical drug trial, which would be the ideal design," Schetter said. "So, there will be different ways of handling clinical performance depending on the samples that are available." For example, according to the label of Roche's KRAS test, the agency accepted comparison of the follow-on test against Sanger sequencing using 94 procured colorectal cancer FFPE tissue specimens.
Another category of diagnostics that the agency is thinking about is complementary diagnostics. While companion diagnostics are required for the safe and effective use of the drug, complementary diagnostics are not but can be used in patient management. The FDA is having internal discussions right now about drafting a formal definition for the "complementary" category of diagnostics.
At the meeting, Schetter said that a preliminary definition for a complementary diagnostic is a biomarker test that defines a subset of patients that respond particularly well to a drug and that aid in risk/benefit assessments but isn't a prerequisite for receiving the drug. For now, complementary diagnostics would still require premarket approval, the same as companion diagnostics.
"Historically, it's been a one-test, one-indication paradigm," Schetter said. But as more personalized medicine are developed, the field is "getting far more complicated,” he recognized.