Companion diagnostics regulatory landscape is complex, and developers must understand the strategies to mitigate risk in this constantly changing industry.
In July 2011, FDA issued draft guidance for the development of companion diagnostics (CDx). This added a significant new aspect to drug development with which pharmaceutical companies must contend, and expanded the reach of the molecular diagnostics industry by opening a new category of diagnostic test. A growing list of therapeutics is now being launched with a CDx, with nine premarket authorizations (PMAs) for five indications in 2012 (see Table I). The PMAs for the treatment of HER2-positive breast cancer with trastuzumab are marketed by five companies, in part because of CDx's complexity. It is difficult to develop the best long-term assay the first time, and there may be more than one assay format with commercial value. The fact that five companies entered the market shows that the opportunity for each CDx can be large.
|Table I. FDA approvals of therapeutics and companion diagnostics in 2012.|
When CDx are mentioned, attention immediately turns to the pharmaceutical industry, because the therapeutic indication is the raison d’etre for the CDx. Roche is arguably the industry leader, with the first CDx and the greatest number of therapeutics accompanied by a CDx (three). In 2012, Roche stated that more than 60% of its therapeutic projects are coupled with a CDx to make treatments more effective and more than 50% of its approvals will be accompanied by a CDx by 2014.
The industry has also seen notable disappointments in which a CDx has not been adequately developed, is not needed for drug approval after all, or faces commercial hurdles. These hurdles include FDA approval of a test that is not reimbursed6 and the substitution of the FDA-approved CDx with another test (see Table I).
In this emerging paradigm, diagnostic platform companies do not have clear guidelines to determine whether to develop a CDx. This article discusses key points to help ensure that the right CDx test is developed, at the right time, within the right business model. Because reimbursement has been written about extensively recently, this article does not cover reimbursement.
FDA states that approval of therapeutics requiring molecular stratification for safe or effective use must be accompanied by an FDA approved or cleared CDx. Given the technical, regulatory, and commercial risks, it is critical to understand what a CDx is and, equally important, what it is not. A test is a CDx if it is essential for any of the following:
The most common misconception is that a test that selects a treatment population with a critical, unmet need—without a concomitant requirement for selection to see therapeutic effectiveness—is a CDx. Such tests, while of clinical value, often do not fall into the CDx category. This important distinction resulted in one failed New Drug Application (NDA) and one voluntary withdrawal of a PMA submission, both associated with therapies for chronic myelogenous leukemia (CML).
Tyrosine-kinase inhibitors (TKIs), beginning with imatinib, which received FDA approval in 2001, have changed CML from a fatal disease into a manageable, chronic condition for most patients. However, about 7% of patients develop resistance to imatinib. The most common cause of resistance is a mutation in the BCR_ABL kinase domain. The T315I mutation was particularly troublesome because it reduced the effectiveness of all therapies available before 2012. Two CML drugs approved by FDA in fall 2012, omacetazine (created by Synribo) and ponatinib (released by Iclusig), were brought through clinical development particularly for patients carrying the T315I mutation.
A CDx for selecting patients with the T315I mutation was planned for each therapeutic. However, in both cases, the CDx was not needed because the drug met its primary end point in clinical trials that included all patients, not just the T315I-carrying patient subpopulation. Because all patients could potentially benefit from the therapeutic, restricting the label to those with the T315I mutation would not be good medical practice. It would not be good business practice either. Restricting the indication to patients with the T315I poor-prognosis marker would have unnecessarily reduced the number of patients treated with the drug by three-fold (Table II).
|Table II. Economic analysis quantifies the trade-offs of targeting different subpopulations for the indication, clarifying the impact of biomarker selection based on clinical need or biology.|
The stories of the two drug-CDx combinations demonstrate just how difficult is to the right test.4,5 Even if a drug is developed with a poor prognosis marker in mind, the drug may perform well in a broader population. The poor prognosis marker is not, in all cases, a CDx.
Before completing the clinical trial, it can be difficult to guess whether the drug performs well in the broader population. In the case of ponatinib, Phase I results suggested that the drug might be more effective in patients with the T315I mutation, prompting the development of a CDx. Although Phase II results confirmed better performance in the T315I-mutation patient subpopulation, no stratification was required for the clinical trial results to meet the primary end point for all patients. The PMA submission for coapproval with ponatinib was voluntarily withdrawn. The CDx was justified, from the perspective of therapeutic development, for risk mitigation.
It's also important to acknowledge the uncertainty in regulatory policy. FDA intends to publish the final CDx guidance in 2013. However, given the continual advances in pharmaceutical development and our understanding of disease, FDA guidance evolves and is updated periodically. FDA encourages sponsor communication about CDx strategy early and often.
This practice of frequent communication goes hand in hand with the use of the modular PMA. In a modular PMA, each module of the PMA is submitted to FDA for review and approval as it is completed, removing regulatory risk along the development path. The modular PMA was used successfully for both the ALK test for crizotinib (Zelboraf) and the BRAF test for vemurafenib (Xalkori) (Table l).
The difference in the development time line for therapeutics and diagnostics is well-known. Under today’s practices, therapeutic development begins many years before biomarker characterization of its impact. The realization that a biomarker may be needed for patient selection, and may be elevated to become the basis of a CDx, comes later.
Advances in science leading to concepts such as genetics-driven therapeutic development, are changing the flow. Biomarker studies are beginning earlier, setting the stage for the greatest medical and commercial success with the least risk. This changing paradigm of therapeutic development is driving remarkable changes in cooperation between therapeutic and diagnostic developers, with the goal of reaching FDA approval with a fully developed CDx.
Such concurrent approval brings with it an additional timing challenge, because of mismatched commitments on the part of FDA. For therapeutics reviewed under the Prescription Drug User Fee Act (PDUFA V), FDA must adhere to the eight-month review goal. For diagnostics, reviewed under the Medical Device User Fee and Modernization Act (MDUFA III), the review process must be 90–180 days, with clock stops for certain events. Thus the CDx developer must have a contingency plan for a longer-than-expected review period.
As an alternative to the modular PMA discussed earlier, using a laboratory-developed test (LDT) → PMA development model can reduce regulatory timing. In the LDT → PMA model, the therapeutic is approved for commercialization accompanied by an LDT, which has a shorter validation process. Only after the therapeutic has been approved is the CDx development completed and submitted for PMA. The LDT → PMA model is attractive because it reduces the business risk for the CDx developer. If the CDx is not needed, as in the case of ponatinib and the T315I CDx mentioned earlier, or if the drug is not approvable, the business finds out relatively quickly and can adjust plans accordingly.
The concept for the LDT → PMA model arose from cases in which an already-approved therapeutic was relabeled by FDA. In 2009, cetuximab (Erbitux) and panitumumab (Vectibix) were relabeled postlaunch to restrict their use to colorectal cancer patients lacking a resistance-conferring KRAS mutation. At that time, the only KRAS-mutation tests were LDTs. The FDA approved the test for KRAS mutation in July 2012 (Table I). In January 2012, natalizumab (Tysabri, NASDAQ:BIIB, NYSE:ELN) was relabeled upon FDA clearance of an LDT for the presence of anti-JCV antibodies. A PMA is yet to be approved for this test. These LDT-based CDx are allowed when the FDA deems it unethical to withhold a therapy until a PMA-approved test has been achieved. FDA has required that a PMA (or 510K) CDx must be approved or cleared within a certain time.
Contemporaneous therapeutic and CDx development is only part of the timing requirement for a successful CDx. Commercialization timing is also critical. The typical diagnostic reaches peak market access four to five years after FDA approval or clearance (Figure l), compared with one to two years for new pharmaceuticals. During the long post-launch ramp-up period, the diagnostic sponsor is expanding market access through reimbursement channels, including being granted a charge code by CMS and coverage by other insurers. By comparison, market access for pharmaceuticals is a streamlined process.
|Figure l. Timing of patient access for the CDx must match or exceed the typical timing for the pharmaceutical.|
Global commercial reach also is important. Pharmaceutical companies planning to launch a new therapeutic globally will need global distribution of the CDx, combined with rapid global patient access. The CDx developer must be prepared with global distribution and customer training more rapidly than usual.
The need for global commercial reach of a CDx is the biggest challenge to the LDT → PMA model, wherein individual laboratories develop and validate the test to support therapeutic launch. A global network of such laboratories in keeping with each national regulation would be required.
Classical pharmaceutical company deal structures keep most of the sales royalties within the company. This is accepted and expected, because the therapeutic developer has shouldered the majority of the risk. Deal structures for CDx development do not fit this model. As alluded to earlier, the CDx developer is also exposed to the therapeutic development risk, especially the risk that the therapeutic will not be approvable. The risk associated with therapeutic approval dwarfs that of the CDx, and association with a therapeutic increases risk by an order of magnitude above that which diagnostic companies usually face.
Within the context of current policy, the CDx is a tool for achieving FDA approval for the therapeutic. The combined dependence of the therapeutic on the CDx and the exposure of the CDx developer to this risk introduces new expectations into traditional business partnerships.
Further, there is no guarantee that the CDx test will be commercially successful; the FDA encourages the development of competitive tests. In 2012, for example, FDA approved six additional CDx tests for trastuzumab for marketing by five different companies. Thus, while a few to several diagnostic test developers might share in the success of a therapeutic, the developer of the CDx accompanying the therapeutic through FDA approval bears the risk that the therapeutic might not be approvable.
Given this distribution of risk, traditional partnership models are extremely challenging in the current CDx environment. Three basic pillars are needed for successful partnerships, as follows:
Relationships between pharmaceutical companies and CDx developers are changing. Proposed strategies include combination products, integrated organizations, and alterative development paths such as the LDT → PMA model. With the rapidly changing science and business realities, continued evolution can be expected.
Successful development of CDx requires selecting the right test to develop; meeting development, regulatory, and commercialization timelines; and doing so within a business model that properly incentivizes all parties.
Ensuring commercial success depends on patient access and especially clinical adoption. Physician education is critical, and genomic medicine initiatives are gaining traction. Ultimately, the physician must be convinced by the quality of the evidence, have an actionable alternative for a rule-out result, and see a compelling enough reason to change practice.
Carol Preisig is CEO of Preisig Consulting LLC. She is a business leader with twenty years of experience building value by combining educational, scientific, technical, business and real-world experience to deliver exemplary results in life science tools, pharmaceuticals, diagnostic and biotechnology industry sectors. Preisig builds and leads multifunctional high performance teams in mixed large cap, small cap, for-profit, non profit, private, public and entrepreneurial environments. Preisig holds an MBA from the Rady School of Business, University of California, San Diego and received her PhD in Plant Physiology at the University of Kentucky.