Issue: July 25, 2017
July 26, 2017
16 min read

FDA’s ‘breakneck pace’ accelerates drug access, but concerns persist about lack of complete data

Issue: July 25, 2017
You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact

The FDA has utilized accelerated drug approval pathways for a quarter century to expedite access to therapies that fulfill unmet needs for individuals with serious medical conditions.

The agency approved 16 cancer drugs last year, five of them through its accelerated program. The FDA already approved 18 cancer drugs in the first half of this year, another five of which received accelerated approval.

“The oncology drug products division is approving more drugs than any other sector. They’re on a breakneck pace,” Vinay K. Prasad, MD, MPH, assistant professor of medicine at Oregon Health Sciences University, told HemOnc Today.

Surrogate endpoints that serve as the basis for accelerated approvals can be arbitrary, according to Vinay K. Prasad, MD, MPH.
Surrogate endpoints that serve as the basis for accelerated approvals can be arbitrary, according to Vinay K. Prasad, MD, MPH. “Response rate doesn’t mean the tumor has shrunk to nothing,” he said. “It just means the sum of the product of the diameter has shrunk more than 30%.”

Photo credit: Kristyna Wentz-Graff, courtesy of OHSU.

Still, some members of the oncology community speculated whether safety and efficacy have taken a back seat to speed.

Many drugs that receive accelerated approval based on surrogate endpoints ultimately fail to show a survival benefit, yet often remain on the market. Other agents turn out to be more toxic than anticipated.

“The political pressure to approve at this rate in this climate is so staggering that, if the oncology drug products division was a bit more of a purist about what is best for patients, I think they would incur the wrath of industry, congress and industry-sponsored ‘patient’ advocacy groups,” Prasad said.

In response, the FDA issued a statement to HemOnc Today indicating it approves drugs based only on whether they have been shown to be safe and effective for their intended use.

“We strive to approve these medications as quickly as possible once we have adequately and appropriately weighed a drug’s benefits and risks based on the data received in a drug application,” the statement read. “Standards for demonstrating safety and effectiveness have remained constant over the past several decades and are recognized as the gold standard worldwide.”

HemOnc Today spoke with oncologists and policymakers about the advantages and drawbacks of accelerated approvals, the increased reliance on surrogate endpoints, the impact that delays in postmarketing studies can have on patients and providers, and the proliferation of so-called “me-too” therapies.

Earlier access, less information

Accelerated approvals — instituted by the FDA in 1992 — allow for faster regulatory clearance of therapies based on surrogate endpoints, such as objective response rate and PFS.


The FDA Safety and Innovation Act — signed into law in 2012 — expanded the FDA’s authority to grant accelerated approvals of drugs based on an intermediate clinical endpoint that predicts short-term benefit or a surrogate endpoint that is “reasonably likely to predict clinical benefit.” Such decisions must account for the severity, rarity or prevalence of the condition, as well as the availability of alternative treatments.

“The pro of accelerated approval is clear: A promising drug becomes more widely available to patients more quickly,” Richard L. Schilsky, MD, FASCO, senior vice president and chief medical officer of ASCO, told HemOnc Today. “The con is that the earlier you introduce the drug, the less information you have about it.”

It is possible that patients will be exposed to a drug — approved based on a surrogate endpoint — that ultimately will not provide clinical benefit.

“After a drug is introduced based upon a surrogate endpoint, it is going to take time to develop the survival endpoint that everyone is the most interested in, and it is possible to never get reliable survival data for various reasons,” Schilsky said. “The risk is that the drug gets out there with incomplete information about how it performs. It is possible the initial risk–benefit does not hold up well, and it turns out to be more toxic or less efficacious than what was expected.”

The FDA has pulled only a handful of cancer therapies off the market following accelerated approvals: gefitinib (Iressa, AstraZeneca), gemtuzumab ozogamicin (Mylotarg, Pfizer), bevacizumab (Avastin, Genentech) and ponatinib (Iclusig, Ariad). All but gemtuzumab ozogamicin were reintroduced under narrower indications.

In 2011, the FDA revoked the accelerated approval it had granted 3 years earlier to bevacizumab in combination with paclitaxel as first-line treatment for advanced HER-2–negative breast cancer. The agency kept the drug on the market for certain types of colon, lung, kidney and brain cancers.

“[Bevacizumab] was initially thought of as almost a probable cure for many cancers,” Lowell E. Schnipper, MD, chief of hematology/oncology at Beth Israel Deaconess Medical Center and a HemOnc Today Editorial Board member, told HemOnc Today. “It looked like there was improved PFS but, in fact, OS was not impacted, and oncologists wondered if it was justifiable with a treatment cost of $9,000 a month.”

In 2013, FDA pulled ponatinib off the market — 10 months after it received accelerated approval for chronic myeloid leukemia — because of a substantial increase in arterial and venous thrombosis. At the time of its approval, results of a phase 2 trial had not yet been published.


In May, a confirmatory phase 3 trial that evaluated atezolizumab (Tecentriq, Genentech) for second-line treatment of locally advanced or metastatic urothelial cancer failed to meet its primary endpoint.

Although complete data have not yet been released, the results raised questions about the FDA’s accelerated approval of the drug a year earlier. At the time of reporting, the drug remained on the market for this indication.

Prasad said he believes the phase 3 failure of atezolizumab is “just the tip of the iceberg” when it comes to cancer drugs that do not meet their primary endpoints.

“If atezolizumab is pulled, that would be a good decision. We have another drug, pembrolizumab [Keytruda, Merck], that won in a very similar randomized trial,” Prasad said. “That’s the system working the correct way, but it hasn’t been pulled yet.”

A Genentech spokeswoman told HemOnc Today in June that company officials believe atezolizumab will continue to play an important role in the treatment of advanced bladder cancer.

FDA officials declined to comment specifically on atezolizumab. However, the statement it provided to HemOnc Today noted that — during the postmarketing process — a pharmaceutical company submits full data from a confirmatory trial to the FDA, which then discusses the findings in detail with members of a review division.

“If, after a review of the data and consideration of the totality of evidence (eg, from all confirmatory trials), a clinical benefit is not demonstrated (or if the applicant fails to perform required postmarketing studies with due diligence), the FDA may withdraw approval following an open public hearing,” the agency’s statement read.

‘Lesson in humility’

The withdrawal of bevacizumab and the failure of the confirmatory atezolizumab trial has sparked debate in the oncology community about the use of ORR and PFS as clinically meaningful endpoints.

The FDA defines ORR as the sum of partial responses plus complete responses, and PFS is defined as the time from randomization until objective tumor progression or death.

“People miss the fact that all of these surrogates are very arbitrary,” Prasad said. “Response rate doesn’t mean the tumor has shrunk to nothing. It just means the sum of the product of the diameter has shrunk more than 30%.”

Likewise, PFS does not mean the tumor has not grown, Prasad added.

“It can grow quite a bit, but it can’t cross an arbitrary threshold of progression, which is 144% in a cross-sectional area and 173% in a 3-D area,” Prasad said. “[That] is a lot more tumor than people might think.”


The challenge with surrogate endpoints is compounded by the fact patients with refractory diseases often have few or no therapeutic options, the FDA said in its statement.

“We take that into account when examining the risk–benefit profile of these drugs,” the statement read. “We have had multiple discussions over a number of years within the global scientific and patient community — including with the Oncologic Drugs Advisory Committee — regarding the use of PFS, response rate and other endpoints to support approval of drugs that treat cancer. It has been widely accepted that benefit can be demonstrated by a number of endpoints, not just OS.”

OS may be impractical for a variety of reasons, the FDA added. For example, some cancers grow slowly, so it may take years for a new drug to show a survival benefit. Other barriers include patients’ need for additional therapies during disease progression, and the emergence of data that show new drugs demonstrate overwhelming benefit compared with other available therapies.

If pressure exists for quicker approval through surrogate endpoints, it is in response to Congress’ policies and is not dictated by the FDA, according to Derek Raghavan, MD, PhD,HemOnc Today’s Chief Medical Editor for Oncology and president of Levine Cancer Institute at Carolinas HealthCare System.

“When the bar is lowered to allow drugs through more quickly, that reflects the desires of our Congress, as the FDA does not set policy,” Raghavan said.

Surrogate endpoints in hematologic cancers may be more meaningful than in solid tumors, for which tumor shrinkage is easily measured.

“Among the community of leukemia specialists, there is a growing consensus that maybe we should be shifting to PFS,” Eytan M. Stein, MD, internist and hematologic oncologist at Memorial Sloan Kettering Cancer Center, told HemOnc Today. “It may not be the best surrogate, but there is value in PFS in and of itself, and that’s something a lot of us are talking about at conferences. PFS may be a clinically meaningful endpoint.”

Earlier this year, the FDA received a new drug application for enasidenib (AG-221/CC-90007; Agios, Celgene), an experimental inhibitor of IDH2 that nearly tripled OS compared with historical data among patients with relapsed or treatment-resistant acute myeloid leukemia. An FDA ruling on enasidenib is expected in August.

Because patients with relapsed or refractory AML have dismal prognoses, the need for treatments in that patient population is extraordinarily high. So is the need for accelerated approval, according to Stein, who has been a researcher on trials of enasidenib.


“For drugs like enasidenib that target relatively small patient populations, the review of the accelerated approval packet must be robust,” Stein said. “It would take too long to get a lot of these drugs to patients if we didn’t have an accelerated approval pathway.

“The fear, obviously, is that a confirmatory trial shows that a drug does not actually help people or may even hurt people,” Stein added. “Therefore, both drug companies and investigators need to do a really good job of having the data submitted to the FDA without flaws.”

Beyond surrogate endpoints and whether a drug confers a survival benefit, quality of life is another factor to consider.

The 71 cancer therapies the FDA approved between 2002 and 2014 provided patients a median 2.1 months of additional survival compared with standard therapies, according to a special communication by Fojo and colleagues, published in JAMA Otolaryngology-Head & Neck Surgery.

The quality of those additional months matters most to oncologists and their patients.

“If the drug is also associated with significant toxicity, even though you might have a delay in the time it takes for the cancer to start progressing again, the patient is likely to be affected by negative consequences of the drug,” Schnipper said. “That would at least cause many patients to wonder if they are really benefitting from a medication that is causing them to feel ill. If people won’t be assured that they’re going to live longer, that tradeoff becomes less valuable.”

The example of epoetin alfa (Epogen, Amgen) — designed to improve anemia in patients with chronic kidney disease and later approved to alleviate cancer fatigue — shows that survival and quality-of-life endpoints sometimes are at odds.

“It became a multibillion-dollar drug for Amgen that was advertised widely on television to doctors,” Schnipper said. “However, people with advanced cancer who received this drug to treat their anemia had a higher death rate from the cancer, which wasn’t known until many years after the drug’s approval. It’s a lesson in humility because we don’t know everything there is to know about these drugs.”

Ruxolitinib (Jakafi, Incyte) — which inhibits the growth-promoting enzymes active in cells prevalent in myeloproliferative disorders — is an example of a drug that improved patients’ quality of life without showing a survival benefit, Schnipper said.

“Patients’ spleens got smaller, they could eat more and there was a general sense of well-being,” he said. “That drug was approved even though there was no survival benefit because the patients who were suffering from the disease felt a lot better. Our patients need to weigh in on this because some may say they really don’t care if they live another 2 months; they want to be comfortable.”


‘Shaky ground’

Between 2005 and 2012, the FDA approved 117 novel drugs for 123 indications on the basis of a single pivotal trial, multiple pivotal trials that used surrogate markers of disease, or both, according to a study by Pease and colleagues published this spring in The BMJ.

The FDA approved 54 cancer drugs between 2008 and 2012, according to a research letter by Kim and Prasad published in 2015 in JAMA Internal Medicine. Thirty-six were approved based on surrogate endpoints, and 15 of those received accelerated approval.

Follow-up of these drugs showed 18 of those approved based on surrogate endpoints — including six that received accelerated approval — ultimately failed to improve OS.

“The FDA is getting into shaky ground,” Schnipper said. “Although [it is] trying to respond to the hue and cry of people with chronic intractable diseases who beg for solutions, and the industries that are developing potential solutions, approving something so quickly may mean you overlook some real problems.”

Lowell E. Schnipper, MD
Lowell E. Schnipper

The FDA approved marketing applications for 55 indications based on surrogate endpoints between 2009 and 2014, according to a study by Kim and Prasad published last year in Mayo Clinic Proceedings. These included 25 accelerated approvals and 30 traditional approvals.

Only four (16%) of the accelerated approvals included level-one analyses, the most robust way to validate a surrogate endpoint.

“It’s OK to use those surrogates in certain times, as long as you follow up after the fact to ensure the drug is doing what it should be doing in the real world,” Prasad said. “We found in many studies that the FDA is using surrogates quite a bit — two-thirds of the drug approvals are based on surrogates — and [it is] doing a very poor job of following up on these drugs.

“To me, the deepest problem is when drugs receive full regulatory approval based on a surrogate,” Prasad added. “Those are drugs that will never be tested again. This is the only information we will ever have about their efficacy, and they cannot be revoked for lack of efficacy later. Drug manufacturers get the benefit of a surrogate approval but have no postmarketing obligations for efficacy.”

Asked how the predictive value of PFS is tested, the FDA stated that — for accelerated approvals — surrogate endpoints must be reasonably likely to predict clinical benefit, as determined by epidemiological, pathophysiological, therapeutic, pharmacologic or other biomarkers.

“When the FDA approves a drug using the regular approval pathway, an improvement in established surrogate endpoints or a direct clinical benefit must be demonstrated,” the FDA stated. “A large magnitude of PFS benefit or a durable ORR may be considered a clinical benefit in and of itself in certain contexts and, therefore, may be suitable for regular approval.”


The FDA should provide data that prove response rate or PFS is meaningful and warrants regular approval, Prasad said.

“Although I understand the convenience of granting regular approval — it waives postmarking efficacy requirements — I am not sure that it is truly what is best for the public, as there is tremendous uncertainty about these drugs that will never be resolved.”

Postmarketing studies

Once the FDA accepts a sponsor’s application for a new drug — which requires a fee of more than $2 million — it generally takes between 6 months and 8 months to receive accelerated approval.

Following full approval, a drug’s sponsor is required to make two annual payments — a $512,200 establishment fee and a $97,750 product fee — and comply with confirmatory trials that generally are noted in the FDA’s approval letter.

In its statement to HemOnc Today, the FDA indicated confirmatory trials must be completed “with due diligence” and conducted “promptly to facilitate determination, as soon as possible, of whether clinical benefit has been verified.”

The FDA requested or required about 1,400 postmarketing studies between March 2008 and September 2013, according to a 2015 report by the U.S. Government Accountability Office. More than half of FDA reviews of those postmarketing studies were delayed or overdue, the report’s authors found.

The FDA “lacks reliable, readily accessible data on tracked safety issues and postmarket studies needed to meet certain postmarket safety reporting responsibilities,” the report read.

In their research letter, Kim and Prasad reported that the 2009 U.S. Government Accountability Office report also showed 30% of 400 postmarketing studies requested by the FDA were pending, ongoing, delayed or terminated years later.

“In 2009 and again in 2015, the Government Accountability Office came to the same conclusion that the FDA has done a lousy job of enforcing those commitments and, in many cases, they are long overdue,” Prasad said.

Thirty-two percent of new drugs approved by the FDA between 2001 and 2010 were flagged for a safety issue after approval, according to a study by Downing and colleagues, published in May in JAMA. Most of those safety concerns were addressed with black box warnings, researchers wrote.

Postmarket safety events occurred more commonly with therapies that received accelerated approval (incidence rate ratio = 2.2; 95% CI, 1.15-4.21).

Experts with whom HemOnc Today spoke said poor comparison arms used in clinical trials could be the reason for safety issues following approval.

For instance, Robert and colleagues compared nivolumab (Opdivo, Bristol-Myers Squibb) with dacarbazine — an alkylating agent that had been used to treat melanoma and Hodgkin lymphoma — in 518 patients with melanoma. Enrollment occurred from January 2013 through February 2014.


Results, published in 2015 in The New England Journal of Medicine, showed nivolumab improved median OS (not reached vs. 10.8 months; HR = 0.42; 99.79% CI, 0.25-0.73), median PFS (5.1 months vs. 2.2 months; HR = 0.43; 95% CI, 0.34-0.56) and ORR (40% vs. 13.9%; OR = 4.06; P < .001).

“By 2013, no one in America was using dacarbazine,” Prasad said. “We were using dabrafenib [Tafinlar, Novartis] vemurafenib [Zelboraf, Genentech] and ipilimumab [Yervoy, Bristol-Myers Squibb]. So, of course nivolumab would beat dacarbazine.”

FDA regulations do not contain a comparative efficacy requirement for trials that support approval, the agency said in its statement.

“Sponsors may choose to use comparators as long as they do not pose a specific safety concern to patients,” the statement read. “Clinical trial enrollment may be compromised if sponsors use suboptimal comparators, so the FDA advises sponsors to treat patients in trials according to U.S. standard of care. In addition, if a trial lacks clinical equipoise, it should not be conducted.”

‘Me-too’ drugs

Another issue facing the oncology community is the prevalence of “me-too” drugs, a term often used to describe a drug that contains an active pharmaceutical ingredient that is chemically related and often structurally similar to an existing active pharmaceutical ingredient.

The phenomenon is particularly apparent with PD-1 and PD-L1 inhibitors, many of which have moved through accelerated approval pathways.

For example, since the accelerated approval of atezolizumab for second-line urothelial cancer, four additional checkpoint inhibitors have been approved for advanced bladder cancer: the PD-1 inhibitors nivolumab and pembrolizumab, and PD-L1 inhibitors durvalumab (Imfinzi, AstraZeneca) and avelumab (Bavencio, Pfizer).

“The system needs the equivalent of a health technology assessment group that is completely neutral, nonpartisan, and looks at a drug that has made its safety and efficacy hurdles and determines where it fits in with what we already have,” Schnipper said. “Is it a ‘me-too’ drug that adds just a little bit to what is already out there? Is it a blockbuster that is almost a game changer? Or is it somewhere in between? That kind of relative weighting would then translate into the degree to which Medicare and insurance companies might choose to cover it. There needs to be some rational way to value it.”

The FDA contends there are differences in PD-1 and PD-L1 inhibitors, such as the specific protein these antibodies target, their immunoglobulin subclass (IgG1 vs. IgG4), the recommended dose and schedule of administration, and the agents’ immunogenicity profiles.

“These differences may lead to variability in outcome in terms of both safety and efficacy,” the FDA said in its statement to HemOnc Today. “Although these drugs have not been directly compared, current data suggest that the approved PD-1 and PD-L1 inhibitors may have fairly similar safety and efficacy profiles in lung cancer. Differences in trial designs and patient populations studied may account for some differences in results reported to date.”


The FDA added that, in general, drugs in the same class may have different pharmacokinetic properties, toxicity profiles and efficacy profiles, and that individual patients may tolerate these drugs differently.

“For example, four ALK inhibitors are approved for the treatment of patients with ALK–positive NSCLC,” the agency’s statement said. “Each one has unique side effect profiles, pharmacokinetic properties and central nervous system penetration profiles, and each may be unique in how they overcome resistance to other ALK inhibitors. Therefore, it may be beneficial to have multiple drugs with the same mechanism of action for the same disease available to patients and their health care providers.”

Continual learning

Policymakers have had discussions about the drug development paradigm in oncology, and there appears to be growing support for a change in the way drugs are studied and tested.

In 2012, Schnipper — then chair of ASCO’s Cancer Research Committee — helped craft a guideline of clinically meaningful trial outcomes with the intent of raising the standard for FDA approvals.

For example, for previously untreated metastatic triple-negative breast cancer, the benchmarks include a 4.5- to 6-month improvement in OS from a baseline median OS of 18 months. The benchmark also calls for a 4-month improvement in PFS.

“We could be justifiably criticized that we didn’t raise it high enough,” Schnipper said.

In December 2016, Schilsky led a workshop for the National Academy of Medicine tasked with reviewing the standard approach of phase 1, phase 2 and phase 3 studies to determine a drug’s safety and efficacy.

“We have to continue to learn about a drug from the moment it enters human testing throughout its entire life cycle and the way it is deployed in clinical practice,” Schilsky said. “We never really fully understand how they work — and how they work in diverse populations — until the drug has actually been out there in clinical practice for a time.

“The real issue is, when does the FDA have enough confidence in the risk–benefit of the drug to allow its introduction into the marketplace knowing that is not the end of the learning about that drug?” he added. “That’s only the end of the premarket learning about that drug. FDA makes a judgment based upon the data that are available as to whether the risk–benefit of the drug in the intended use population justifies market entry, but with the expectation that we’re going to continue to learn about the drug, and with the possibility that they might adjust their stance on the drug at a future time.” – by Chuck Gormley

Click here to read the POINTCOUNTER, “Should an independent regulatory body oversee the FDA drug approval process?”


Downing NS, et al. JAMA. 2017;doi:10.1001/jama.2017.5150.

Fojo T, et al. JAMA Otolaryngol Head Neck Surg. 2014;doi:10.1001/jamaoto.2014.1570.

U.S. Government Accountability Office. FDA expedites many applications, but data for postapproval oversight need improvement. Available at: Accessed on July 6, 2017.

Kim C and Prasad V. Mayo Clin Proc. 2016;doi:10.1016/j.mayocp.2016.02.012.

Kim C and Prasad V. JAMA Intern Med. 2015;doi:10.1001/jamainternmed.2015.5868.

Pease AM, et al. BMJ. 2017;doi:10.1136/bmj.j1680.

Prasad V, et al. JAMA Intern Med. 2015;doi:10.1001/jamainternmed.2015.2829.

Robert C, et al. N Eng L Med. 2015;doi:10.1056/NEJMoa1412082.

Stein EM, et al. Blood. 2017;doi:10.1182/blood-2017-04-779405.

For more information:

FDA Office of Media Affairs can be reached at

Vinay K. Prasad, MD, MPH, can be reached at

Derek Raghavan, MD, PhD, FACP, FRACP, FASCO, can be reached at

Richard L. Schilsky, MD, FASCO, can be reached at

Lowell E. Schnipper, MD, can be reached at

Eytan M. Stein, MD, can be reached at

Disclosure: Stein reports grants and personal fees from Agios and Celgene. Prasad, Raghavan, Schilsky and Schnipper report no relevant financial disclosures.