FDA draft guidance on the Principles for Codevelopment of an In Vitro Companion Diagnostic Device with a Therapeutic Product, published on July 14, 2016, can be viewed as a testament to the Agency’s commitment to incorporate scientific advances into its regulatory framework. In the 90 days this “how-to guide” was open for comment some 15 companies and associations responded, including Diaceutics. Their varied and considered recommendations provide important insights into the challenging, and often complex, regulatory requirements for developing these two products simultaneously.
Personalized, or precision, medicine has the capacity to detect the onset of disease by predicting individual susceptibility, allow greater use of more targeted treatments, and increase efficiencies of healthcare systems by improving quality, accessibility and affordability. For the patient this translates into greater access to safe, effective and novel life-saving therapies. Precision medicines now account for almost a third of FDA’s new drug approvals, with the likelihood this trend will not only continue, but progress from its current focus on oncology to other disease areas.
It is not the intention of this white paper to look in detail at all the issues arising from the draft guidance, or to acknowledge individual comments, but to put forward and examine some of the major recommendations and challenges identified by the 15 companies and associations* (stakeholders) that responded to the guidance to ensure this crucial FDA document serves as a truly practical guide to sponsors working in this space. Some of the key challenges highlighted include: distinguishing between a companion and complementary diagnostic; the difficulties inherent in contemporaneous approval of an IVD companion diagnostic and a therapeutic product; performing analytical validation of an assay prior to its use in a clinical trial; prescreening for eligibility in a clinical trial using local tests; and what is required from a bridging study.
Companion vs complementary diagnostic
Overall, the draft guidance is felt to be a thoughtful, progressive document that offers flexible advice that is generally consistent with current practice. Nevertheless, a number of important areas have been singled out for improvement and further clarification by the 15 stakeholders. Notable among these is the omission of a clear definition of a complementary diagnostic. Given that FDA may not determine whether a diagnostic will be awarded companion or complementary diagnostic claims until the pivotal clinical trial data are reviewed, an omission of the principles and criteria necessary for distinguishing between the two diagnostics could put diagnostic manufacturers at a distinct disadvantage in codevelopment planning with their therapeutic partners.
Diaceutics sees this gap in clear understanding of complementary diagnostics. In a recent survey conducted by Diaceutics in this regards, only 51% of oncologists had clear understanding of difference between companion and complementary diagnostics1. 34% didn’t understand the difference in labelling and 15% understood it incorrectly. This is a problem that needs to be addressed. Diaceutics works diligently with its pharmaceutical clients in understanding the regulatory and compliance environment around these two test classifications and helps them successfully navigate the process of development and commercialization of targeted therapy along with diagnostic test.
Conflicts in contemporaneous approval
Interestingly also, the general premise of the guidance – that the only appropriate test is the one codeveloped with the drug, or developed with studies that use specimens from patients being treated with that drug (which in reality may not always be obtainable) – is also called into question by stakeholders, who suggest this regulatory approach does not allow FDA to keep pace with the speed at which the science is progressing
Indeed, the guidance states that a companion diagnostic should be granted a de novo request or cleared by FDA contemporaneously with the approval of a corresponding therapeutic product2. It continues: “If an IVD companion diagnostic is essential to assuring safety or effectiveness of the therapeutic product, FDA generally will not approve the therapeutic product or new indication for a therapeutic product if the IVD companion diagnostic does not already have marketing authorization or will not receive contemporaneous marketing authorization for use with that therapeutic product for that indication”3.
In our experience, pharmaceutical companies can significantly reduce the risk of delays to FDA therapeutic drug review and approval through early diagnostic partnering, early engagement with FDA alongside that partner, and transparent communication between pharma and diagnostic partner through all the phases of codevelopment.
But while early and effective communication is key to successful codevelopment, there are nevertheless two major issues with this aspect of the guidance. First, the guidance is inconsistent with the Agency’s 2014 In Vitro Companion Diagnostic Devices guidance, which, by comparison, states that it may decide to approve a therapeutic product even if an IVD companion diagnostic is not approved or cleared contemporaneously. In the second issue, the Association for Molecular Pathology contends contemporaneous approval would hinder the application of new technologies and improvements to current tests over the decades the drug is in use, preventing some patients from receiving the most appropriate care. Diaceutics suggests that by proper crafting of a commercialization agreement between pharma and IVD partner, where diagnostic product life cycle management (LCM) related risks are identified and mitigation strategies are addressed, these gaps in patient care can be addressed.
Early engagement is essential
The importance of early engagement is recognised in the guidance, which recommends both therapeutic and diagnostic product sponsors meet with the appropriate FDA review centers during planning and prior to launching a clinical trial designed for regulatory purposes, and that information is shared between the different FDA centers and the two sponsors4.
The document does not, however, cover all bases here. Procedural details are not included in at least three key areas. For example:
The Biotechnology Innovation Organization (BIO) recommends the Agency considers implementing a process for integrating engagement with different centres within FDA to help increase efficiency of the codevelopment process for both the diagnostic and therapeutic sponsor. Additionally, as the Personalized Medicine Coalition (PMC) points out, the inclusion of a sample timeline in the guidance would allow therapeutic and diagnostic partners to work together in a timeline consistent with FDA’s expectations. In addition to recommendations made by BIO and PMC, we suggest that alignment and transparency between pharma and the diagnostic partner’s regulatory teams is very important. Effective communication with FDA, related to codevelopment regulatory strategies, and representation from both parties (pharma and IVD partner) during different phases of development (as early as PreSub) is critical for success.
Investigational device exemption and risk determination
One of the early considerations for codevelopment is when an investigational device exemption (IDE) application must be submitted for an investigational IVD companion diagnostic. FDA employs a risk-based approach to decide whether the study of a particular IVD companion diagnostic is exempt from IDE, requires a non-significant risk IDE, or a significant-risk IDE. However, as risk is evaluated in the context of a specific therapeutic clinical trial design, such evaluations can be subjective.
Sponsors need to be fully appraised of the criteria to assess risk determination, so the guidance requires clarification of when an IDE application must be submitted, and what type.
Further guidance is also deemed useful on what the likelihood is of non-significant risk when using an approved test with a new indication; whether annual IDE reports are reviewed by the drug center if an IDE application is embedded within the investigational new drug (IND); and when an IDE should be closed – as Dako North
America notes, the guidance states six months after completion or termination of the investigation but the patient follow-up period can extend beyond a year in some investigations.
Genentech suggests an IDE may not be necessary for a prospective-retrospective analysis, but requests FDA and external stakeholders develop guidelines for this type of evolving challenge.
Analytical validation rigour
The guidance is clear on the importance of conducting analytical validation studies that evaluate critical performance parameters of an assay prior to its use in a trial. But while it provides recommendations on resources sponsors can use to help design these studies, and on the performance characteristics that should be tested – such as accuracy, precision, analytical sensitivity and specificity, and reproducibility5 – it lacks information on analytical performance criteria. Examples of such criteria in different technologies would be extremely beneficial to the industry, notably analytical validation statistical standards, or the level of analytical validation that should be applied to assays for a given risk categorisation.
In order to help determine which studies are needed, and the degree of rigour that should be applied to each study, the guidance suggests sponsors (referred to as Dx company) use the Pre-Submission Program guidance issued on February 18, 2014, which recommends sponsors propose a protocol for a Pre-Submission meeting, with a rationale for the chosen approach. To ensure the protocols submitted by sponsors meet FDA’s expectations, Flagship Biosciences suggest that FDA provides a framework of statistical standards for determining the validation performance needs for an assay in each risk category. Such an approach, the company argues, could increase the quality of communication between sponsors and FDA, and enhance collaboration. Diaceutics echoes this suggestion and believes that a tight collaboration between the diagnostic company and FDA in early phase stages, is most important for designing validation studies and understanding performance standards.
Problems with prescreening
The practice of prescreening for eligibility is another key area where further guidance is sought. The draft document suggests that prescreening – using a wide assortment of ‘local’ tests where there “is often no assurance they are standardised or interchangeable” – “may result in a biased clinical trial population that does not represent the population that would be selected by the IVD companion diagnostic in real-world testing. Thus, planning to enrol subjects into a trial based on confirmation of a local test result is strongly discouraged”6. One way for sponsors to avoid potential bias from prescreening, the draft guidance recommends, is to send specimens for testing with the trial test from all potential enrollees, rather than just those identified from a prescreening test.
These proposals throw up several problems, however. The increased application of personalized medicine in oncology means local tests are often used to determine
treatment pathways for patients. Indeed, as several stakeholders point out, prescreening has not only become the new baseline for clinical care and the basis on which many patient-enrolled studies are performed, but many patients consider enrolling in a clinical trial only after receiving these results.
Moreover, the guidance doesn’t allow for:
A possible solution, proposed by PhARMA, is to allow sponsors to enrol subjects into a trial based on confirmation of a local test result on a case-by-case basis.
At the same time, stakeholders feel the final FDA document should recommend approaches to adequately address potential selection bias, as well as provide guidance on handling prescreening, and examples of how to manage, design, populate and conduct a clinical trial where prescreening is a factor.
Interestingly, the Association for Molecular Pathology takes the argument in favour of prescreening one step further by refuting FDA’s claims that assurances cannot be made about the performance of local tests. “The CLIA (Clinical Laboratory Improvement Amendments) program at the Centers for Medicare & Medicaid Services, in addition to state level requirements and third party reviewers, are readily verifying that LDTs (local tests) are both accurate and precise”, it states. The Association goes on to suggest that “standards developed through collaborative efforts between professional organizations and various government entities, which could include FDA, would provide laboratories with the ability to ensure that agreed upon thresholds for analytical validity were met without the need for burdensome and costly FDA premarket review”.
We recognize the importance of performance standards and utilization of local tests in the prescreening process. Our laboratory operations team works with pharma clients and helps them design testing strategies related to prescreening that minimize the bias during clinical trial enrolment. This is an important issue which needs further consideration as clinical trial design and the access to innovative therapies is destined to become more challenging as we further stratify diseases by pathogenic genetic abnormalities.
Biomarkers and best practice
As with prescreening, FDA’s guidance on biomarkers also raises a number of concerns. Acknowledging the growing use of multiple markers, FDA advises that: “analytical validation of each reported marker may be required regardless of each marker’s prevalence. When it is not possible for sponsors to obtain specimens containing a particular marker, validation studies with contrived samples may be permitted. Analytical validation studies may also be complicated for IVDs that have the potential to detect a very large number of markers, in which case it may be necessary for the study to use a representative sampling of markers”7.
However, Genentech points out that with high-throughput technology-based testing where markers for different uses might be represented as part of a single test or panel, it is unclear whether all markers have to undergo analytic validation or just the predictive markers used to make therapeutic choices. The company suggests it should be the latter.
Indeed, Genentech makes a number of thoughtful recommendations in this area. An important goal of early stage clinical development is to identify the population unlikely to benefit from the therapeutic product [although this is not always the case, especially in biomarker positive enrolment design]. It would be helpful, the company suggests, if FDA included best practice guidance to determine the marker negative population, specifically the number of marker-negative patients to be enrolled in order to establish the lack of benefit in early studies. Indeed, where lack of benefit is established Genentech feels late stage development should focus solely on recruiting marker positive patients.
FDA states, “the results of the clinical trial can also be used to establish the clinical validity of the IVD companion diagnostic”8, but it is unclear how sponsors can distinguish between analytical and clinical validity in the exploratory phase. FDA goes on to state that knowledge of the prevalence of the marker of interest in the population to be treated is critical in a prospective-retrospective design – Genentech recommends a “good-faith” effort at enrolling adequate numbers of marker positive patients should be sufficient.
It would also be helpful if the guidance offered specific advice on how to deal with issues caused by the availability of multiple devices for the same biomarker. In our view, the guidelines on defining biomarker negative population will be very helpful in reflecting the true performance in terms of specificity of diagnostic test for evaluating safety and effectiveness of the test.
Cutoff values for biomarkers
It is in the area of setting a priori cutoff values for biomarkers, however, that sponsors may encounter some of the biggest practical challenges. According to the guidance: “The cutoff for an IVD companion diagnostic is the test value above (or below) which the clinical decision changes” and patients are considered to be positive or negative for the marker(s) of interest, and those that distinguish relevant trial populations “should usually be established for the investigational IVD prior to use in clinical trials intended to be submitted to support a therapeutic product’s approval”9.
Diaceutics commented that while establishing pre-specified cutoff values for the IVD is deemed essential before use in pivotal therapeutic product clinical trials, in practice the use of semi-qualitative values can be challenging. For example, in HER2 using IHC, use of the 0, 1+ 2+ 3+ scale posed a number of problems, not least the cutoff set at 3+, with 2+ deemed equivocal and requiring further testing using in situ hybridisation (ISH). This caused difficulties in the lab and led to 2%-7.4% of breast cancer patients not receiving the correct therapy before effective guidelines were implemented10,11.
More recently, PD-L1 expression determination by immunohistochemistry assay (IHC) has been used with many different cutoff values (1%, 5%, 10% and 50%) using different monoclonal antibodies without standardisation/harmonisation, leading to significant confusion for sponsors.
Genentech, in particular, argues that this part of the guidance may not be feasible. Moreover, with continuous biomarkers varying cutoffs can confer varying levels of benefit. Best practice guidance is needed to set cutoff values, in particular, the company suggests, where they cannot be determined prior to obtaining robust data on patient response to investigational medicine(s).
IVD bridging studies
A number of questions also arise in regards to FDA guidance on late therapeutic product development – most notably in the area of bridging studies.
FDA states that: “If a test other than the candidate IVD companion diagnostic is used for the major efficacy trial(s), the IVD sponsor should demonstrate that the candidate IVD companion diagnostic has performance characteristics that are very similar to those of the test that was used in the trial”12 – in other words, through a bridging study between the two tests.
But the Agency offers no insight into what a “very similar”, or indeed a different test might be. AstraZeneca asks if a RUO (locked protocol) is transferred into IUO (same locked protocol but now a physical product), does it represent a different test?
The ideal bridging study, according to the guidance, is “one in which all samples tested with the trial test are retested with the candidate IVD companion diagnostic and valid test results are obtained and used to assess comparative performance.”
Ideally, in designing pivotal clinical trials, pharma should plan to archive patient samples from the pivotal studies for use in the bridging studies. But what happens with older study samples where few or none of the samples are available, or the sample stability is past, or baseline sample versus study sample may be required? AstraZeneca asks whether a bridging/concordance study could be conducted between two tests on a different set of samples (i.e., a technical bridge), and if so, what would be the minimal requirements to satisfy FDA that the two assays are selecting the same patient population?
The company also questions whether there is an acceptable level of concordance if two different assays have been used for patient selection, and what would be the best/acceptable metric for FDA?; how imbalance, if identified, should be addressed; and whether FDA has a position on the minimal available sample set for bridging?
Diaceutics believes questions relating to bridging studies are important and should be addressed in the final guidance. Given the diverse range of technologies being used in IVD development, some of the bridging study strategies might vary on a per case basis. For example, the total number of samples to be retested or a technical bridging option might depend upon the prevalence of the disease, stability of stored samples and availability of additional clinical samples that are representative of study population. Concordance has been an important part of clinical lab validation of new assays for many years and CAP and AMP have published guidance on how laboratories should test for concordance. FDA should perhaps utilise the experience of these organizations in defining concordance criteria.
Publication of the final guidance will be a crucial step in helping the advancement of precision medicines. The thoughtful and proactive thinking already exhibited in the document will be raised further by the considered responses and suggestions of the 15 companies/associations that responded. But technologies are changing fast, and FDA needs to be nimble and disseminate information quickly if it is to keep up. While we await final regulatory guidance on the Principles for Codevelopment of an In Vitro Companion Diagnostic with a Therapeutic Product, attractive tools such as complementary diagnostics, and codevelopment using high-throughput technologies such as Next Generation Sequencing-based test panels, are already waiting in the wings.
* The 15 companies/organizations that responded to the guidance were: Advanced Medical Technology Association (AdvaMed), American Association for Cancer Research (AACR), Association for Molecular Pathology, AstraZeneca, Biotechnology Industry Organization (BIO), Dako North America, Diaceutics, Epstein Becker Green, Flagship Biosciences, Foundation Medicine Inc, Genentech,
Illumina, NanoString Technologies, Personalized Medicine Coalition (PMC), PhRMA.