More than a third of the new indications for oncology drugs that became available for patient care during the past 25 years entered clinical practice as a result of the FDA’s accelerated approval program.
Julia A. Beaver, MD
More than a third of the new indications for oncology drugs that became available for patient care during the past 25 years entered clinical practice as a result of the FDA’s accelerated approval (AA) program, an expedited review process that relies on surrogate endpoints to predict benefit before clinical trials are fully conducted. The use of the program has grown over the years, so much so that more new cancer drugs and indications gained approval through the AA pathway than through the regular review process during the first 6 months of 2017.
That is the picture that emerges from an analysis that FDA officials wrote in March for JAMA Oncology1 after a spate of criticism in academic journals about the agency’s cancer drug approval practices. The report comes at a time when the FDA is seeking to streamline regulatory processes and pick up the pace of new drug approvals.
Supporters of the program, including the FDA itself, deem it a major success. Countless patients have been able to take what confirmation trials usually demonstrate to be life-extending therapies several years before those medications likely could have reached the market via the regular approval process.
However, critics say the benefits of AA in particular and surrogate endpoints in general have come at a high price, with many patients exposed to costly medications that turn out to be inadequate alternatives to drugs with proven efficacy. They say the pace of confirmatory trials to convert AAs into regular approval status has been too slow.
The FDA analysis shows that the AA pathway accounts for 35% of the indications approved for new agents (those classified as new molecular entities) and existing drugs from December 11, 1992, when the program began, through May 31, 2017. Notably, more than half of the monoclonal antibodies approved through the AA program are checkpoint blockade agents targeting the PD-1/PD-L1 pathway that have been credited with ushering in the new era of anticancer immunotherapy (Figure).Much of the controversy about the AA program stems from a broader debate about appropriate endpoints for oncology clinical trials. Since the FDA began requiring clinical trials for new drug approvals in 1962, the gold standard for any new drug approval (not just cancer drugs) has been a randomized clinical trial that demonstrates an improvement in survival or important clinical symptoms. However, surrogate endpoints have been used, such as improvements in blood pressure or serum cholesterol for cardiovascular disease studies.2
For oncology drugs, the FDA usually issued approvals in the 1970s based on the objective response rate (ORR) determined by tumor assessments from radiological tests or physical examinations. In the early 1980s, the agency shifted to what it considered “more direct evidence of clinical benefit,” such as improvement in survival, quality of life (QOL), or tumor-related symptoms.2 In the 1990s, surrogate endpoints were established, including disease-free survival, durable complete response in leukemia, response duration, and relief of tumor-related symptoms.
Drug approval benchmarks came under heavy fire during the AIDS epidemic in the 1980s, when desperate patients and their supporters demanded access to experimental medications as soon as they demonstrated efficacy in any measure that might prove a reasonable proxy for extended survival. The FDA responded in 1992 by writing regulations that created the AA pathway, allowing the FDA to approve drugs based on a surrogate endpoint that is “reasonably likely to predict clinical benefit” for patients with a serious or life-threatening condition who do not have other options or if the experimental therapy demonstrates an improvement over existing treatments.2 (It is 1 of 4 programs the FDA has crafted to expedite drug development; the others are the fast track, breakthrough therapy, and priority review programs.)
The most frequently used surrogate endpoints in cancer drug trials in the AA program have been response rates, including pathologic complete response (pCR) and hematologic endpoints such as CR or major cytogenic response.1 Over the years, the FDA’s use of the AA pathway for oncology drugs has been highly controversial. “Many have noted the risks associated with making toxic drugs available rapidly on the basis of surrogate endpoints that may not translate into real clinical benefit. Others deplore the delays in completing the required studies to demonstrate a survival advantage or other established clinical benefit following an accelerated approval....Still others are concerned that the process has been insufficiently accelerated,” wrote Susan S. Ellenberg, PhD, a professor of biostatistics and medical ethics and health policy at the Perelman School of Medicine at the University of Pennsylvania in Philadelphia, in a 2011 commentary.3 “The FDA is routinely castigated for proceeding too cautiously and slowly, and at the same time, for rushing drugs to market too rapidly without adequate study.”
In 1 snapshot of the FDA’s policies, researchers found that between 2008 and 2012, the FDA granted 36 indications to cancer drugs based upon surrogate endpoint performance, but that after a median follow-up of 4.4 years, just 5 (14%) demonstrated an improvement in overall survival (OS). Just 1 of 15 indications obtained through the AA program reached the OS milestone; the survival benefit was not reached for more than 50% of the indications and was unknown in the others.4
In a 2017 analysis, investigators examined AA approvals that the FDA issued for 24 indications, including 19 for cancer treatment, between 2009 and 2013.5 Efficacy was confirmed in postmarketing studies within a minimum of 3 years for 42% of the indications in confirmatory trials but these studies also used surrogate endpoints as the benchmark. Findings from a study of meta-analyses of surrogate endpoints used in oncology trials questioned the validity of such markers; just 23% of 65 surrogate markers used were highly correlated with survival.6
Bishal Gyawali, MD, PhD, whose areas of study include the use of surrogate endpoints in cancer trials, contends that clinical trial data gleaned from studies that use surrogate endpoints such as progression-free survival are not as straightforward or consistent as outcomes that use OS as a standard.7
“It’s understandable that patients with incurable cancers want faster access to drugs that may save their lives, and it’s understandable to expect drugs that dramatically shrink tumors to end up extending lives, but our desire to be compassionate and use logical shortcuts has led us to give patients a large number of very expensive and significantly toxic medications that have done them no good whatsoever,” Gyawali, MD, PhD, a research fellow at Brigham and Women’s Hospital in Boston, Massachusetts, said in an interview with OncologyLive®.
“This isn’t to say that we shouldn’t use alternative endpoints for accelerated approvals, but we need to revoke those indications for medications that do not support them with timely follow-up evidence of either extended survival or improved QOL,” he said. “Instead, some drugmakers are pretty openly defying the requirement of timely postmarket trials, and others continue to sell drugs after trials have proven they provide no real benefit.” Although some researchers are calling for more studies to use OS as a primary outcome, others argue that the randomized trials needed to establish such a benefit may be ethically questionable, time-consuming and inflexible.7,8 Others note that it would be almost impossible to recruit patients for a randomized trial that would deprive patients in the control arm of a drug that already won an AA in order to establish an OS benefit in a confirmatory trial.
“Some targeted therapies and immunotherapies demonstrate large durable objective response rates with a different toxicity profile compared with standard therapy [when they win AA]. These early results make it unlikely that patients would participate in a subsequent trial in which they could be randomized to a standard therapy control arm with low response rates and high toxicity rates without the opportunity to cross over to investigational therapy on disease progression,” Julia Beaver, MD, lead author of the FDA study on the AA pathway1 and director of the agency’s Division of Oncology Products 1, said in an email interview with OncologyLive®.
“For most advanced cancers, progression of disease is important clinically and often triggers changes in clinical management,” she said. “Progression-free survival is considered a clinically relevant endpoint for postmarketing trials and occurs prior to crossover, mitigating the challenges with crossover seen when assessing OS. In addition, durable tumor shrinkage (response rate) has been used in confirmatory trials where randomization may not be feasible [as when] a drug shows unprecedented activity early in clinical development. For these reasons, OS is often an impractical endpoint to verify clinical benefit for many of the drugs approved under the AA program.”
Another major problem with using OS is the cost in time and dollars, and it’s a cost that goes up with the value of the research being done.
“It simply isn’t realistic to use OS as the primary endpoint on many of our most important trials—the ones that extend the lives of curable patients by years rather than extending the lives of end-stage patients by months. There simply is too much time or money required to use OS in such studies,” said Laura Esserman, MD, MBA, coleader of the Breast Oncology Program at the University of California, San Francisco, Helen Diller Family Comprehensive Cancer Center.
To illustrate her point, Esserman discussed the ongoing I-SPY 2 trial that she helps lead. The study’s goal is to test the efficacy of a wide range of entirely novel treatments in patients with stage II/III breast cancer. These patients typically have many years to live, so a traditional trial aimed at testing just 1 novel treatment’s effect on OS would likely last more than a decade and cost many millions of dollars. I-SPY 2 is using a number of strategies to save time and money—and still generate accurate efficacy data—among the most important of which uses complete disappearance of the tumor or pCR as the primary endpoint.
“Cardiology trials don’t track patients all the way to death before they conclude that a blood pressure medicine can save lives,” Esserman said. “They use the surrogate endpoint of reduced blood pressure because they know from previous studies that lower blood pressure will eventually lead to improved OS. We’re at the same point in cancer care. We know how the disease works well enough that we can pick surrogate endpoints that predict a better survival.”
Confirmation trials that have followed AAs predicated on alternative endpoints have created a wealth of data about the correlation of various endpoints and survival in particular patient populations. In some cases, the data seem to have fully validated certain surrogates for certain populations. There’s now substantial evidence that pCR is a strong proxy for OS in women with stage II/III breast cancer, Esserman said.
Similarly, Robert M. Califf, MD, a former FDA commissioner, believes that the ability to leverage a growing trove of data about drug safety will help pave the way to a new paradigm for investigating oncology drugs. “As electronic records, information storage, and analytic methods continue to improve, there are no technical limitations that would prevent tracking a marketed drug or device throughout its life cycle,” wrote Califf, a professor of cardiology and a member of the Duke Clinical Research Institute at Duke University School of Medicine in Durham, North Carolina, in a commentary about the AA program.8 “As technological improvements and nationally and globally connected networks of health systems make it feasible to conduct high-quality, low-cost [randomized clinical trials] and to continuously monitor product performance, the impediments to progress are mostly those built into the culture of medicine and healthcare.”