When the first personalized cancer therapies came to market more than 20 years ago, some experts predicted that the field’s growing understanding of the molecular features driving tumors would alter the way drugs are developed.
The oft-cited statistics about drug development is that it takes more than a decade and well over a billion dollars for a therapy to go from discovery to market. But many industry observers have been eagerly anticipating the time when the integration of genomics would transform drug development, shortening that timeline and reducing costs.
For some, those predictions are becoming reality. “We are seeing an increasing number of drugs getting fast-track approvals or accelerated approvals with a biomarker,” said Jordan Clark, chief technical officer of diagnostics data analytics firm Diaceutics. “Drugs are more targeted and responses are better in those particular cohorts and subsets, which ultimately means less time for a clinical trial because you need less people in it and it [takes less time] to recruit. With precision therapies that equals a cheaper and quicker clinical trial.”
He cited the example of the TRK inhibitor larotrectinib (Loxo Oncology’s Vitrakvi), which had one of the fastest development timelines: five years between when it was first discovered and when it garnered FDA approval.
Another example is osimertinib (AstraZeneca’s Tagrisso), which inhibits a mutation within EGFR. It went from the first patient dosed to FDA approval in less than three years, according to Bob Holt, director of knowledge and insights at Diaceutics.
Diaceutics predicts that in the next few years the US Food and Drug Administration will approve more molecularly-informed personalized drugs than therapies indicated for general disease populations. This shift, if it happens, will be the result of changes that drugmakers have made to development strategies and increasing flexibility on the part of regulators to approve therapies based on smaller, single-arm trials in the era of precision oncology.
Meanwhile, there are those who believe that these fast approvals based on emerging data from small cohorts may not be in the best interest of cancer patients in the long run. Some have accused the FDA of lowering regulatory standards under industry influence and political pressure. The agency has stood by its decisions to approve molecularly targeted cancer drugs based on single-arm studies with high response rates. It is also willing to accept drug safety and efficacy evidence from sources other than interventional clinical trials, including real-world data.
“We’ve had our critics who say the FDA is lowering standards because we aren’t demanding two trials, randomized, showing a survival advantage,” Richard Pazdur, director of the agency’s oncology division, said during a meeting of its Oncologic Drugs Advisory Committee in December. During the meeting, several committee members had recommended against approving olaparib (AstraZeneca’s Lynparza) as maintenance therapy for BRCA-mutated, metastatic pancreatic cancer because the drug had failed to show an overall survival advantage over placebo.
“People have to be realistic,” Pazdur told the committee. “Simply, it can’t be done in all situations.”
“Molecular oncology is changing the entire process of running clinical trials, our expectations, regulations, and approvals,” said Howard West, a medical oncologist at City of Hope. “Ten or 15 years ago, we routinely saw lung cancer and many other cancers as a giant monolithic pool of patients that were pretty heterogeneous in how they responded to various treatments.”
But over the last five to seven years, oncologists have started recognizing the potential in dividing up this pool of patients into discrete subpopulations who had their cancers characterized by specific biological or genetic features, and subpopulation-specific treatments followed. “We don’t have everybody getting amazing outcomes, but we have more and more patients being treated as distinct subgroups with the optimal key for their particular lock,” said West.
Traditionally, before a medication becomes widely available, it passes through stages of clinical research, from early, smaller, Phase I trials to larger, late-stage, Phase III trials. FDA’s cancer division used to require that sponsors show an overall survival advantage in two randomized, late-stage trials. However, in recent years, the regulatory and drug development paradigm has been shifting away from that traditional three-phase trials framework.
Drugmakers attribute this to improved understanding of disease biology and more precise drugs to interrogate those mechanisms. “We started making more selective small molecules, whether they’re ACE inhibitors or epigenetic modifiers or even biologics against immunology targets in the context of cancer,” said Jake Van Naarden, COO of Loxo Oncology, which was acquired by Lilly last year. “We started [doing this] in a world where most good drugs showed themselves in Phase I, whether or not it’s a biomarker-defined population.”
With earlier insights, Loxo made record time in its development program for the RET inhibitor selpercatinib. According to Van Naarden, the firm knew the drug was active in the first half of 2017, after which Loxo moved the drug into human studies. Last fall, the company unveiled data from a Phase I/II trial, in which 68 percent of 105 heavily pretreated non-small cell lung cancer patients with RET fusions saw their tumors shrink on selpercatinib, and said it would submit the data to the FDA by the end of 2019 as part of a new drug application. Based on data from the same study, the company is taking a similar strategy for selpercatinib in heavily pretreated RET-altered thyroid cancer.
Robert Doebele, a thoracic oncologist at the University of Colorado School of Medicine, anticipated this paradigm shift several years ago and argued in Nature Reviews Clinical Oncology that molecularly targeted cancer therapies could forgo randomized Phase III trials if they could show in early Phase I/II trials that they were “superior to chemotherapy with respect to several quality-of-life metrics including physical and emotional function and global quality of life.”
More specifically, for tumors such as lung cancer, which can be driven by a number of rare, targetable mutations, he asserted that “the burden of proof for approval should be adjusted for oncogene-targeted therapy in biomarker-selected patients.” New clinical trial formats are needed, he wrote, that can “accommodate this new paradigm and allow the expedient study and rapid approval of these therapies for patients with oncogene-positive lung cancer.”
Since Doebele’s piece was published, several molecularly targeted cancer drugs have been granted expedited approval based on data from relatively small, single-arm trials in a variety of tumor types. For example, the tissue-agnostic approval of pembrolizumab (Merck’s Keytruda) for advanced solid tumors with microsatellite instability or mismatch repair deficiency was based on data from five uncontrolled, single-arm trials involving 149 patients. The two other tissue-agnostic drugs approved by the agency, larotrectinib (Loxo’s Vitrakvi) and entrectinib (Genentech’s Rozlytrek), were similarly based on data from small, single-arm cohorts.
“Precision-guided medicines can demonstrate strong efficacy signals in early clinical trials, including in trials where small groups of patients are selected based on biomarkers or other criteria suggesting they’re likely to benefit. These trials can potentially allow earlier regulatory assessment of benefit and risk,” FDA said in a statement last year outlining new strategies to modernize clinical trials in the era of precision medicine.
The agency has also been more flexible in terms of the efficacy endpoints it has been willing to consider. “Through discussions about endpoints with oncology professional societies, patient advocates, and the FDA’s Oncologic Drugs Advisory Committee, the FDA received advice that endpoints including objective response rate of sufficient duration and progression-free survival can be clinically meaningful to patients and may be acceptable for either accelerated or regular approval,” an FDA spokesperson said in an emailed statement. “This has led to approvals of drugs that have provided meaningful treatments to patients who previously had few therapeutic options”
The FDA also expects that adaptive design, umbrella, and basket trials will increasingly be used in drug development, which allow researchers and drugmakers the flexibility to start and end molecularly defined treatment arms and prioritize indications with promising activity. Diaceutics’ Clark has noted a rise in basket trials exploring pan-cancer or tissue agnostic indications. Loxo’s registrational study for selpercatinib is a basket trial, which tests the drug in patients with different types of cancers that share the same type of mutation in RET.
Real-world evidence is also making its way into drug applications submitted to the FDA. The 21st Century Cures Act, passed in 2016, required the FDA to establish a program, which evaluates real world evidence (RWE) to support a new indication for an approved drug. This may include data from electronic health records and billing statements or claims.
Since the agency published a framework for its program last year, the use of real-world data in precision oncology drug applications has increased. In 2019, Roche used Flatiron Health’s electronic health record-derived database to show that its investigational drug entrectinib for 1 percent of NSCLC patients with ROS1 fusions fared better than Pfizer’s crizotinib (Xalkori).
Enabling patient access
As genomic insights carve up histology-specific indications into smaller, molecularly defined patient subsets, some experts believe these changes in drug development and regulation are necessary in order to get much needed treatments to patients, particularly in settings where there are limited therapeutic options. Razelle Kurzrock, director of the Center for Personalized Cancer Therapy at University of California San Diego, remembers a time when the FDA was not so flexible in its regulatory requirements for cancer drugs, even those for rare patient populations.
She recently recalled a drug development project she worked on more than 10 years ago. The drug was “extremely active” in a rare indication, but regulators still wanted a randomized-controlled trial. Kurzrock knew that a randomized trial would take a long time and be expensive. Ultimately, the sponsor discontinued the drug’s development.
Kurzrock and West both said that when there is a strong sign of therapeutic benefit in rare disease settings, mandating randomized trials may be unethical if the new treatment has already demonstrated a high level of activity. West argued that it’s a waste of time to ask for randomized trials for drugs that have “eye popping efficacy” in single-arm trials. In the single-arm studies that led to larotrectinib’s approval, 75 percent of patients saw their tumors shrink and nearly 40 percent of patients kept responding to the drug for a year or longer.
“There are some cases that are more marginal where the response rate may be 30 or 40 percent,” he said. “When you have a response rate of 60 percent or higher in patients with a progression-free survival of six, eight, or 12 months, that’s essentially incontrovertible that this is a beneficial therapy.” In such cases, the role of subsequent trials is to try to decide whether these treatments should be moved into the first-line setting or are reserved for second-line or later, he said.
Experts also pointed out that patients who seek out or are selected for these kinds of early-phase targeted therapy trials become eligible because they have a fatal cancer, have failed to respond to approved therapies, or have no treatment options for their condition. Additionally, if a drug’s toxicity profile is already well understood, it gives regulators more confidence on the risk/benefit profile of the drug based on early-phase, single-arm studies. In the case of pembrolizumab, before it was approved in a tissue agnostic setting, it was widely investigated and approved in multiple tumor types, providing insights into the expected adverse events.
Larotrectinib entered the market first with a tissue agnostic indication because it demonstrated extremely high response rates that were durable and had acceptable toxicity profiles in a biomarker-defined population during the early studies. “The reason that 10 years ago, 15 years ago, drugs were approved on randomized, large Phase III trials is because their single agent response rates were still low … and maybe not that durable,” said Van Naarden. “When you put a drug into human beings and it has a 70 percent response rate and those responses last for a long time and the drug is well tolerated, it actually flips the entire equation on its head.”
Focus on postmarket setting
The reliance on early-phase data to garner approvals has also made it increasingly important to conduct post-marketing evaluations to ensure that these drugs are indeed benefiting patients, extending lives, and not posing unexpected toxicities. Van Naarden noted that the initial approval of a drug is often just the first step in an overall development program. “Usually there are patients that are still on the drug, on the study. They’re continuing to get their medicine, continuing to go to their study visits, and you’re continuing to gather that data,” he said. “FDA demands that you continue to give them updates on that data because they ultimately want to characterize how long the responses last.”
However, confirmation studies can prove challenging to design and conduct in some cases, for example in late-line disease settings where there aren’t many options. “What the FDA has historically done is look in an earlier line of therapy or in a related tumor type in order to try and confirm the activity of those drugs,” according to Jeff Allen, CEO of the public policy organization Friends of Cancer Research, which has been instrumental in advancing the breakthrough therapy designation within the FDA, through which many precision oncology drugs have been expedited to market.
In Allen’s perspective, many of the changes taking place at the FDA in the era of precision medicine, including the use of real-world data, is not necessarily an effort to replace clinical trials, but to conduct “active surveillance research” and learn what cannot be gleaned from a traditional study. “A clinical trial can not be run to answer every single question that might arise. So, to be able to do this with more health system type data holds a lot of potential,” he said. “In the future, this type of real-world evidence assessment could play a role in identifying additional populations that could benefit from a drug and characterizing differences based on different demographic factors or genetic factors that could influence outcomes.”
Toward this end, FDA is building on its current technological capabilities to improve and centralize long-term surveillance of products once they’re on the market. As part of planned updates to its electronic safety monitoring system, the Sentinel Initiative, the agency will help set data quality standards “which will be really important in helping to guide the use of this data in the most appropriate and scientifically based way possible.”
A different risk/benefit calculation
The FDA, however, has its critics who view the changing drug development and regulatory environment with more concern. Where supporters see the agency’s new “flexibility” as a benefit for patients with limited options, detractors see the potential for patient harm in rushing drugs too quickly to market. The New York Times Editorial Board recently wrote an OpEd expressing some of these worries, and accused the agency of approving drugs too fast, influenced by industry lobbying, under political pressure, and without fully considering the scientific evidence.
FOCR’s Allen doesn’t subscribe to the idea that the FDA has consciously lowered standards in precision oncology. He pointed out that because the latest molecularly targeted drugs tend to be for subsets of patients who have not responded well to available therapies, it warrants a new risk/benefit calculation.
“These are not general population-type drugs or pediatric cold medicines. These are often times for patients who have run out of therapeutic options,” he said. “You’re willing to entertain a different level of certainty in these instances than perhaps in larger population-based diseases.”
The discussion at the December meeting of the FDA’s Oncologic Drugs Advisory Committee provided a window into the risk/benefit calculation that the agency must make in the face of limited data about drugs intended for small, molecularly defined patient populations.
At that meeting, committee members voted seven-to-five in favor of FDA approving olaparib as a first-line maintenance drug for metastatic pancreatic cancer with germline BRCA1/2 mutations. Interestingly, this was not one of the instances where a sponsor submitted data from an early-phase, single-arm trial. In fact, AstraZeneca conducted a randomized, double-blind, placebo-controlled Phase III study. The trial, however, showed a median 3.6-month difference in progression-free survival between the placebo and olaparib arms, and there was no improvement in overall survival.
The committee members who voted against approval or were more hesitant about recommending approval highlighted the lack of data showing olaparib improved quality of life in light of a modest PFS advantage. Some experts wanted to see an overall survival impact. Patients in this trial were enrolled based on a predictive biomarker since individuals with BRCA mutations are more likely to respond to a PARP inhibitor. But after seeing the efficacy data, some ODAC members felt that while a small proportion of patients may be deriving benefit from this drug, the trial didn’t identify which subset within the BRCA-mutated population that was.
In response to these concerns at the meeting, Pazdur acknowledged that the agency had also struggled with these questions. He added that the agency brought this application before the ODAC to discuss in a public forum the difficulty of developing and approving drugs targeted to a niche population with a small number of patients. “[In these settings,] just saying, ‘Well, overall survival is the gold standard and we’re not going to take a look at anything else is probably to the detriment of patients,” he said. “We wanted to address at a public meeting that we are looking at alternative endpoints here.”
Pazdur acknowledged that while AstraZeneca attempted a randomized trial and looked at overall survival as an endpoint, this is not always possible in the era of molecular targeted drugs, due to small cohort sizes, the natural progression of the disease, or the availability of other less-active therapies on the market. “For some of the targeted therapies in other diseases, we have given full approval based on a single-arm trial with very high response rates,” Pazdur said. “There has to be some degree of flexibility here in looking at these therapies.”
In the end, the FDA decided to be flexible in this case. Five-year survival rates in pancreatic cancer are between 2 percent and 9 percent. Current treatments include surgery, chemotherapy, and radiotherapy. Although more recently, some targeted and immunotherapy treatments have become available, there is still a need for new options. Before 2019 came to a close, the FDA approved olaparib as an option for BRCA-mutated metastatic pancreatic cancer patients.