Last summer, the Food & Drug Administration published its final guidance on the development and approval path for companion diagnostics. These in vitro tests and devices put personalized medicine into practice by providing, FDA says, the “information essential for the safe and effective use of a corresponding drug.” In clinical trials and medical care, the diagnostics will help identify patients most likely to benefit from a particular therapy.
FDA approved the first companion diagnostic in 1998. Dako’s immunohistochemical assay detected overexpression of the HER2 gene in breast cancer tissue as a biomarker for determining which patients might benefit from treatment with Genentech’s Herceptin.
Other Herceptin-related tests emerged, but otherwise companion diagnostic approvals were slow for several years. Approvals picked up around 2011, the same year FDA published draft guidelines. Today, FDA has approved eight codeveloped drug-diagnostic combinations, and another 14 drugs have separately approved companion tests.
The path toward routine development and use of companion diagnostics is becoming clear, particularly for cancer. Scientists are learning to diagnose patients using samples of blood rather than tissue, and next-generation sequencing (NGS) technology is allowing multiple genetic analyses to be performed on each of many samples. But significant scientific, clinical, economic, and regulatory obstacles remain to a world in which medicines of all types are made personal by use of a companion diagnostic.
Indeed, FDA’s goal of contemporaneous development and approval of drugs and companion diagnostics remains “somewhat elusive,” says Joshua P. Cohen, an associate professor with the Tufts Center for the Study of Drug Development.
It’s not for lack of trying. The Herceptin approval “sparked enormous interest and also lots of resources being poured into biomarker identification and diagnostic development,” Cohen says. Pharma company investment in personalized medicine has risen nearly 90% over the past five years.
And certain successes “should not be forgotten in the story of personalized medicine,” Cohen says. They include the post hoc creation of diagnostics for already approved drugs and the inclusion of pharmacogenomic information on the labels of 137 FDA-approved drugs.
Moreover, codevelopment efforts between drug and diagnostic companies have been gaining traction. “Since 2002, VentanaMedical Systems has worked with more than 45 biopharmaceutical partners and is currently engaged in more than 180 collaborative projects,” says Doug Ward, vice president of Ventana’s companion diagnostics business.
In 2008, the Swiss drug giant Roche bought Ventana for $3.4 billion. Today, Roche systematically incorporates companion diagnostics into its own drug programs as part of a focus on personalized health care. “It is our highest priority now and into the future,” Ward adds.
Meanwhile, independent diagnostic developers, including Abbott Molecular, Illumina, Qiagen, Thermo Fisher Scientific, and Dako, which has been part of Agilent Technologies since 2012, are partnering with most of the major pharma companies. Some diagnostics firms get involved in biomarker discovery. Most can supply a range of technologies, and many already have FDA approvals for in vitro diagnostic devices.
The approval process for a companion diagnostic is “time-consuming and expensive,” says Rainer Metzger, vice president for pharma business development at Qiagen. It covers many linked components including sample prep, reagents, equipment, and software. And expectations for accuracy and reliability are high. The companion diagnostic has to be extremely robust yet simple to use, he adds.
By joining forces early, drug and diagnostic partners can identify a testing approach and present FDA with a combined clinical plan. The diagnostic will help stratify patients in drug trials according to relevant biomarkers to select those most likely to respond. Not only might patient outcomes improve, but the risk of side effects might also be reduced.
Simultaneously, the clinical significance and performance of the diagnostic are shown with actual data from the matched drug. If the stratified approach is done correctly, “FDA tends to be much more flexible in allowing you to move quickly, even with a smaller subset of patients,” Metzger says.
In contrast, partnering late can present problems, especially if the drug trial design has to be changed to accommodate the diagnostic, Metzger explains. Beyond adding time and cost, tissue biopsy samples may not be available. Diagnostic technology choices may be limited. And FDA might delay approval of a drug if the diagnostic required to use it safely is not cleared as well.
FDA officials realize that codevelopment is not always possible. Although 73% of cancer drugs in development rely on biomarkers, usually specific gene mutations, this wasn’t always the case. Many drugs were approved without knowledge of a related biomarker, according to Tufts’s Cohen. Finding a biomarker after the fact and developing a diagnostic can “resurrect a drug that can clearly benefit patients,” he says.
For example, AstraZeneca’s lung cancer drug Iressa has succeeded commercially only outside the U.S., where it is paired with a tissue-based epidermal growth factor receptor (EGFR) gene mutation diagnostic.
In January, European regulators approved use of a new blood-based version of Qiagen’s EGFR polymerase chain reaction (PCR) kit to analyze circulating DNA from blood when tumor samples are not an option. FDA, after previously restricting the drug, has now accepted AstraZeneca’s filing for using Iressa as a targeted first-line therapy for non-small-cell lung cancer patients diagnosed as EGFR-mutation positive.
Many marketed companion diagnostics have been designed to work with traditional tumor tissue samples. Qiagen’s Metzger says the new test is the first in vitro diagnostic product on the market that assesses EGFR mutation status from blood samples.
Such less invasive sampling is more patient friendly and can pave the way for cheaper diagnostics for routine monitoring. But tests using so-called liquid biopsies must be sensitive enough to detect extracted DNA or RNA while maintaining the genetic material in a state still representative of the disease, Metzger points out.
In the more than 20 projects it has under way, Qiagen has designed PCR-based companion diagnostics primarily for single-biomarker testing, Metzger says. But he sees more and more demand for multiplex testing, which the company is addressing by working with a consortium of drug companies. In 2014, Qiagen acquired PrimeraDx and its PCR/capillary electrophoresis system for simultaneously analyzing multiple samples and genetic markers.
Similar joint efforts between individual diagnostic firms and multiple pharma partners are in progress to develop multiplex NGS-based diagnostics. To simultaneously analyze many disease-related gene variants, Illumina, Qiagen, and Thermo Fisher have created cancer gene panels. These panels will first support translational and clinical research and later move toward approval as companion diagnostics.
For example, a Thermo Fisher panel has 52 genes relevant to a range of solid-tumor cancers. “They correlate to on-market therapies and therapies advancing in clinical trials,” says Mike Nolan, the company’s vice president for oncology. Unlike methods that use a lot of tissue and give information on only one biomarker, NGS can require just 10 ng of DNA but give a broad view of the genetic profiles of multiple tumors.
“It is probably not a coincidence that the emergence of NGS and of precision medicine have come hand in hand,” says John Leite, head of oncology at Illumina. “In the past 25 years, cancer has really shifted from a tissue- or morphology-defined disease to a genetically defined disease. And when you have such a close correlation between genotype, cancer subtype, and treatments, the clinical utility becomes far greater than purely morphological classification.”
Developers contend that NGS will facilitate identifying enough patients for clinical testing, which would be challenging with a one-sample, one-biomarker testing approach. Many cancer-associated mutations targeted by today’s drugs occur in less than 5% of patients. “The unfortunate reality of higher specificity, which gives you better responses and lower side effects, is that you are slicing the population into smaller and smaller sections,” Leite says.
The good news is that “the speed at which a pharma company can go to market with a strong correlation between patient response and drug is much higher than it used to be, and the attrition rates are lower,” Leite says. Regulatory agencies are aware of the potential development and approval bottleneck that could result and “are asking for more streamlined ways of identifying patients and more streamlined ways of clearing drugs,” he says.
Illumina and its pharma partners are looking first at standardizing the workflow from sample processing through data analysis and reporting, Leite says. The goal, he adds, is to create panels “representative of the drug and the companion diagnostic funnel for a lot of these pharma partners for the next few years and then submit them for clearance by FDA.”
NGS suppliers see efficiency and cost-effectiveness as their advantage with regulators, drug developers, and end users. Although NGS can support multiple tests for multiple therapies, FDA will have to review only one testing platform. And the platform will exist even if a drug fails in clinical trials, so that effort won’t have been wasted in building a single on-purpose diagnostic, Thermo Fisher’s Nolan explains.
Likewise, clinics won’t have to install and run many different methods and devices for separate tests but instead will be able to combine samples. From a physician and patient standpoint, this means access to testing shouldn’t be restricted by labs having insufficient sample volumes, Nolan says. And with faster testing turnaround times, therapies can be assigned more quickly.
Although momentum toward widespread use of companion diagnostics is building on the scientific front, regulatory and financial challenges remain.
In the NGS area, developers are consulting with regulators about the complexity of clinical trials that will involve multiple cancer types, markers, and drugs. In general, communications with regulators, along with clarity around approval and labeling requirements, are still being refined.
Meanwhile, companion diagnostics face reimbursement challenges across national health care systems. Payers often reimburse the cost of the drug but are “haphazard” in paying for the diagnostic, Tufts’s Cohen says. “They are not questioning whether the diagnostics are accurate but whether their use is improving patient outcomes.”
When the drug and diagnostic are codeveloped, the proof of clinical utility—or link between using the test and patient benefits—is “pretty much built into the system,” he says. That utility is less clear-cut when the test is developed separately, and showing it may have to come through postmarketing clinical trials.
Although many drugs with companion diagnostics are expensive, the combinations appear to be cost-effective, according to Cohen. “The numbers are quite good,” especially for those that were codeveloped. Savings come from giving patients the most effective therapies from the start and avoiding the expense of putting patients on therapies that don’t work.
Another barrier to adoption is doctors’ familiarity with genomic medicine, which was considered inadequate by 90% of drug and diagnostics companies surveyed by Tufts. Efforts are under way by companies and the Personalized Medicine Coalition to educate physicians about tests and their usefulness. When appropriate, FDA is adding information to labels, Cohen says.
Diagnostics firms see attractive opportunities around both new drugs and existing ones such as Iressa. In 2014, nine new drugs, or more than 20% of FDA approvals, were considered personalized medicines. Companion diagnostics are expected to grow by more than 20% per year and become a multi-billion-dollar market.
Although the development focus and regulatory experience so far in companion diagnostics have been largely in cancer, “we are going to see other categories follow suit,” Cohen says. “But it may take a while because the science takes time.”
About 40% of all compounds in preclinical and clinical development rely on biomarker data. But in the Tufts study drug and diagnostic firms ranked biomarker identification and test development highest among the challenges that remain to advance personalized medicine.
Beyond addressing different diseases and helping make treatment decisions, new biomarkers may also be used for measuring patient response and prognosis. In turn, they will drive the development of assays based on a range of diagnostic technologies and sampling methods.
To make all these things happen, drug companies are increasingly interested in partnering with diagnostic firms. And regulators are open in their thinking about how to evaluate new treatment approaches, according to Illumina’s Leite. To Leite, these factors add up to a robust scientific and business opportunity. “The reality,” he says, “is that targeted therapies and precision medicine are here to stay.”