Originally published Sept. 20.
By Turna Ray
Arlington, Va. — An official from the US Food and Drug Administration last week demonstrated that the agency is making progress on issuing formal guidelines on the codevelopment of drug and diagnostic personalized medicine products by discussing detailed strategies for navigating the regulatory process for Rx/Dx combination products at an industry conference.
Since issuing a white paper on drug/diagnostic codevelopment five years ago, the FDA on several occasions has failed to meet its own projected timelines for issuing draft guidelines on the topic. The latest word came from FDA Commissioner Margaret Hamburg earlier this year, when she committed the agency to completing its long-awaited drug/diagnostic codevelopment guidance by year end (PGx Reporter 03/03/10).
When it was released in 2005, the white paper on Rx/Dx codevelopment was criticized by both drug and diagnostic industry stakeholders, who felt that the agency's proposed guidelines were too idealistic, and failed to account for the challenges of aligning competing interests, different development timelines, and divergent regulatory pathways for therapeutics and diagnostics.
Last week, at a conference hosted here by the Cambridge Healthtech Institute, Donna Roscoe, a scientific reviewer at the Office of In Vitro Diagnostics Device Evaluation and Safety at FDA's Center for Devices and Radiological Health, discussed how companies can gain the agency's input on an investigational companion diagnostic ahead of a regulatory submission; provided advice on which FDA divisions to submit test data to; and pointed out certain pitfalls to avoid when filing for regulatory approval for a companion test.
Roscoe's comments reflect stakeholder feedback that the agency has been gathering over the past several years in order to gain a more real-world understanding of the barriers to inking successful Rx/Dx collaborations. To aid FDA in this effort, the Personalized Medicine Coalition in December issued a white paper based on input from 40 of its members that recommended the FDA embrace regulatory flexibility and propose alternative clinical trial models to accommodate a variety of co-development scenarios (PGx Reporter 12/16/09).
Also, in recent months, the agency has formed a new personalized medicine division and has said that it is trying to align the activities of its drug and diagnostics divisions. These internal changes may help the agency solidify a regulatory framework for combination products.
Still, FDA has challenges when it comes to reviewing Rx/Dx products. For example, at the end of a Phase I or Phase II trial, while FDA's Center for Drug Evaluation and Research or the Center for Biologics Evaluation and Research may be ready to give sponsors the go-ahead for a drug, CDRH may not believe that the test is ready for use.
"That's a complex issue that has to be dealt with," Roscoe acknowledged. "We're saying that we really are committed to helping the pharmaceutical companies. We don't want to hold anyone up. We want to work together."
During her talk, Roscoe said she did not know when the Rx/Dx codevelopment guidance is slated for release. However, the detailed advice she provided at the meeting offered the strongest sign in recent months that the agency is moving steadily toward making good on Hamburg's promise of issuing guidance by year end.
Pathways to Market
The FDA has informed sponsors that all diagnostics that pick out the intent-to-treat population for a drug need to be cleared or approved by the agency.
[ pagebreak ]
"The drug and test depend on each other to work. If the test doesn't work, then the drug cannot be administered to the appropriate population," Roscoe said. "The FDA believes that if you're going to use a test to identify a specific population [for a drug] then the test has to be approved." Even if a drug developer is using a test already cleared or approved by the agency, according to Roscoe, the sponsor might still need to file for regulatory approval of the diagnostic if it is being used off label.
Based on FDA's three-tiered risk classification system, companion diagnostics are considered by the agency to be in the highest-risk category, Class III, "because the companion device carries the risk associated with the drug." The FDA has various regulatory requirements for diagnostics based on the intended use and risk level of the test.
"We have some flexibility with the pre-market approval for Class III devices, which are particularly useful for companion diagnostics," Roscoe said. "First you can obviously get a PMA approval for when you're looking to manufacture and sell a device IVD. But if you are designing and implementing a test at a single site, also known as a laboratory-developed test, we can expect to approve those tests as well."
The FDA has said it intends to regulate LDTs, but it hasn’t outlined any specific regulatory requirements. The public comment period on FDA's expressed intent to regulate LDTs ended on Sept. 15.
Additionally, the modular PMA review — the pathway Qiagen is taking to garner approval of its TheraScreen KRAS test for detecting non-responders to certain colorectal cancer drugs — is another option for companion diagnostics developers to consider (PGx Reporter 03/04/09).
"This is very useful for companion diagnostics, because the timing requirements are so critical for a [biologics license application] or a [new drug application]," Roscoe said. "It allows the manufacturer to submit the analytical and clinical performance manufacturing modules independently of each other over time."
Generally, CDER or CBER will lead the review of combination products, and CDRH will take a consulting role when it comes time for conducting a regulatory review of a combination product. "The approval of the device has to be concurrent with the drug. And that's because the drug and the device need to be co-labeled, to reflect each other," Roscoe said.
However, when a drug developer or the agency first identifies the need for a companion diagnostic in early-phase studies, sponsors can submit information on the test either as part of the IND to CBER/CDER, or directly to CDRH as an investigational device exemption form.
"I can tell you from my personal experience … that CDER and CBER like the use of IDEs because then they can just call us up and ask, 'What's the status of this device?'" Roscoe said. "And we can say, 'Oh, we've already approved that IDE,' and [CDER] is happy, because they don't like having things tied up or worrying about the device aspect."
CDRH must look at an IDE and respond to applicants within 30 days. "If we don't get back to you in 30 days about the deficiencies or additional information we want, then [the IDE] is automatically approved," Roscoe noted. If sponsors decide to submit test data as part of the IND application for the drug, CDRH will still ask for a copy of the information submitted on the device.
According to Roscoe, the IDE pathway may also be a "useful tool" if drug and diagnostics partners need to keep some professional distance. For example, if the device developer has proprietary data that it wishes to not turn over to its pharma partner, or if the drug company doesn't want to work closely with the device manufacturer, the IDE is a good way to limit such interactions, she noted.
Gaining FDA's Ear
As pharma and diagnostics firms gain more experience developing Rx/Dx products together, the general consensus is that the earlier the collaboration starts, the greater the chance for a successful combination product launch. Most Rx/Dx partnerships start during Phase II trials for the drug, but finding ways to ink collaborations earlier is likely to minimize the risk of late-stage drug failures.
[ pagebreak ]
Recognizing this, FDA is encouraging drug and diagnostics companies to link up earlier by offering companies a way to talk to the agency about their companion diagnostic strategy without regulatory repercussions. "Ultimately, the recommendation is to consult CDRH as early as possible, even in your Phase I when those tests may not be considered a significant risk. You can make all sorts of changes, and it's good because we'll alert you to the sorts of expectations we will have," Roscoe said.
She recommended that sponsors submit pre-IDE data to gain FDA's ear on a companion device.
"Pre-IDEs are wonderful tools. They are free. They are a mechanism for getting feedback from CDRH," Roscoe said. Although CDRH is supposed to provide feedback to sponsors' pre-IDE requests within two months, she acknowledged that the timeline might be longer since the agency is "inundated" with pre-IDEs and is "a little behind," because "companion diagnostics has just exploded."
The agency will want to see preliminary analytical and clinical validity data for the device being used in the clinical trial; a detailed description of the intended use of the test; as well as a description of the biomarkers, platform, drug, and the intent-to-treat population.
"For companion diagnostics, the clinical validity of the test is the success criteria that you've laid out for [the drug] to CDER or CBER," Roscoe said. "We don't have independent study requirements … Whatever has been used to define the efficacy of the drug is what we use for the clinical validity of the test."
The information that sponsors provide in pre-IDEs is non-binding, confidential, and will not be used in regulatory review of the combination product. Furthermore, the reviewer assigned to discuss the pre-IDE with the sponsor will be the same reviewer that handles the regulatory submission for the test.
"Hopefully this will prevent waste of time and resources," Roscoe said. "We've found a lot of situations where if the sponsor had just known" certain things to avoid while conducting clinical trials, "they would have saved an enormous amount of expense."
The FDA has identified a number of red flags that would cause a sponsor's companion diagnostic to receive a negative review. The biggest pitfalls sponsors should avoid include using different samples in the study than the intent-to-treat population; changing preclinical and extraction steps throughout the development process; failing to show concordance between the test being used in the study and the marketed test if the two are different; and using research-use-only instrumentation in clinical trials.
"We like to see the data with real clinical samples. You can purchase the samples but they should be the same population that is in the [drug] trial," Roscoe said, adding that sometimes sponsors may supplement data from cell lines if not all alleles are represented in the patient population being studied.
Still, particularly for genotyping assays, FDA would like sponsors to show that all the alleles they are claiming in the test indication are represented in the study sample set. "If your clinical trial doesn't include patients that include representations of those claimed alleles then you won't be getting that claim," Roscoe informed. Also, analyte instability may also be a problem if RNA is being used, "because we're going to ask you to prove that the samples are stable," she added.
The FDA will also make sure that changes to the pre-analytical reagents and extraction steps from early to late studies are not impacting test results. "Frequently, your companion diagnostic will evolve between the time that you've started the trial and when you bring it to market. If you can stick with the [same] pre-analytical steps and extraction steps, you will save yourself a lot of trouble," Roscoe noted. "When some of the performance steps start with archived DNA, you won't be concerned that variability exists in the pre-analytical steps because you haven't really changed them."
Labs involved in performing a test for a trial should also follow the protocol laid out in the sponsor's intended final label, "otherwise changes will have to be accounted for," she warned. Additionally, if a sponsor's companion device is operating on a platform not cleared or approved by the FDA, then that, too, will have to be submitted for approval.
[ pagebreak ]
Additionally, Roscoe emphasized that device manufacturers "avoid research-use-only components," as all systems used as part of a clinical trial involving human subjects need to be developed under good manufacturing practices and quality systems regulations. If RUO products are used, "then you'll have to work with the manufacturer of that particular instrument or make arrangements yourself to bring it in to your [quality systems regulations], which is not trivial," she noted.
When designing a trial, sponsors should keep in mind that CDRH prefers a study design that tests both marker-positive and marker-negative patients, since this confirms the predictive claim of the test, according to Roscoe. Under certain circumstances, however, where the biology of the biomarkers and the disease are well understood, FDA will accept a targeted study design that looks at half the patient population. Still, the agency believes it's important to study both marker-positive and -negative patients and will want to see this type of analysis, as long as there is no serious safety issue that makes it unethical to look at both subsets of patients, Roscoe said.
While FDA prefers that companies conduct prospective trials for companion diagnostic validation, there have been instances when the agency has accepted retrospective study data. In the case of companion KRAS testing for Amgen's Vectibix and Merck's Erbitux, FDA accepted retrospective analyses for clinical validation but delineated certain criteria for submitting such data (PGx Reporter 07/22/09).
"In general, for predictive diagnostics, the first one of its kind seeking approval has to be associated in some way with a [drug] trial," Roscoe noted. "After that, manufacturers may be able to procure samples … and demonstrate concordance or bridge to the approved test."
Finally, in an effort to avoid bias in studies, FDA recommends that pharmaceutical and device companies develop risk-mitigation strategies to educate physicians about not pre-screening patients for testing before enrolling them in trials.
"Understandably, in the practice of medicine, physicians test patients as soon as they come to their office," Roscoe said. "CDRH will look at the prevalence of the [marker being tested] in the overall population to see that it matches up with the literature to ensure you didn't have a biased set of people."