As practicing physicians, approval by the U.S. Food and Drug Administration (FDA) reassures us that the treatments we prescribe our patients have been rigorously vetted as effective and safe. For a growing number of drugs, however, FDA approval comes with lingering uncertainty of their benefit. Under the accelerated approval pathway, FDA provisionally approves drugs based on surrogate endpoints, such as laboratory or imaging results that are proxies for disease progression, with the requirement that manufacturers conduct clinical trials following approval to confirm clinical benefit. This arrangement is intended to speed access to promising treatments, with the expectation that manufacturers will confirm that the drugs improve how patients feel, function, or survive.
But manufacturers have largely failed to hold up their end of the bargain — nearly half of drugs granted accelerated approval have not been confirmed to be clinically beneficial. Deadlines to complete postapproval trials are often generous, and the trials that are completed frequently confirm benefit using surrogate endpoints, using those same endpoints that supported the initial accelerated approval. In some cases, drugs without evidence of clinical benefit can linger on the market for years. FDA recently withdrew romidepsin’s approval for peripheral T-cell lymphoma — ten years after accelerated approval, and more than a year after its confirmatory trial failed to show benefit. By the time ineffective uses are withdrawn, drugs can be anchored in the market by other indications — leaving physicians with an unclear picture of whether these treatments offer benefit over alternatives.
In the meantime, patients can be exposed to serious risks. Duvelisib, which was granted accelerated approval for a type of lymphoma, carries multiple black-box warnings, including potentially fatal infections in nearly one-third of patients. It was approved based on the assumption that the benefits would potentially outweigh these risks; nevertheless, its manufacturer never started the required trial to confirm duvelisib’s predicted benefits in the two years before FDA withdrew it. Safety concerns are more common for drugs granted accelerated approval, and there is no efficient process for the FDA to withdraw drugs, instead relying on manufacturers to do so voluntarily. Patients and physicians trying to weigh the potential risks of new drugs may wait years for answers — if they come at all.
Accelerated approval drugs also have massive financial implications, including for Medicare and Medicaid, which together provide coverage for 135 million Americans. In our recently published study in JAMA Health Forum, we found that Medicare and Medicaid spent $68 billion through 2020 on new drugs granted accelerated approval between 2012 and 2017, three-fifths of which was on drugs with postapproval trials evaluating surrogate endpoints, for which clinical benefit was not confirmed. As Medicare and Medicaid are mandated to cover certain therapeutic classes, including cancer drugs that accounted for nearly all spending in our study sample, they are on the hook for these drugs, regardless of whether they are ultimately, or ever, proven to work.
In light of recent scrutiny of accelerated approval, catalyzed by FDA’s controversial approval of aducanumab for Alzheimer’s disease, Congress is considering reforms to the pathway, in hopes of resetting the balance for patients between timely access to treatments and timely certainty of their benefit. These reforms are being considered as part of user fee legislation that authorizes FDA to collect and use fees paid by pharmaceutical companies to support new drug reviews.
Proposed legislative reforms to accelerated approval are intended to support FDA in holding manufacturers accountable for completing their postapproval studies, by requiring that confirmatory trials are initiated by the time of accelerated approval, increasing FDA oversight of trial design, and mandating manufacturers provide routine progress reports for studies. These are good first steps, but Congress should also include previously introduced provisions that would facilitate FDA enforcement of approval standards. These include automatic expiration of accelerated approvals without proof of benefit, either one year after study completion or five years after approval, and streamlining the process for withdrawing indications. Removing the proposed additional requirements to offer manufacturers meetings with advisory committees or the FDA commissioner prior to taking action would also minimize delays in withdrawing indications that pose ongoing clinical and financial risk.
Legislation should also clarify that FDA “shall,” not “may,” require postapproval studies for all accelerated approval drugs, as well as require that drugs be evaluated by advisory committees prior to approval, reversing a recent trend towards fewer independent, expert voices in the approval of new drugs. These reforms are essential to provide patients with certainty of drugs’ benefits and prevent the use of unvalidated or clinically inconsequential outcomes in granting accelerated approval or fulfilling postapproval study requirements.
Finally, legislation should ensure real-world evidence augments, rather than replaces, clinical trials fulfilling postapproval study requirements. FDA-funded research suggests only half of studies could be conducted using sources such as electronic health records or medical claims data in place of traditional clinical trials. In another published study, we found that none of the required postapproval studies for accelerated approval drugs could be emulated using such sources of evidence. Although real-world evidence can provide new insights, especially for patients underrepresented in clinical trials, our current health systems are not yet ready to be used as the sole source of data for the rigorous and reliable evaluations necessary to confirm drugs’ benefit.
In the clinic and on the wards, we are frequently called upon to interpret the relevance of clinical trials for our patients. This essential act of translation, from abstract to individual, is made significantly more challenging by the absence of data on the outcomes most relevant to patients — on months lived, hospitalizations shortened, and adverse outcomes avoided. Strong standards for FDA approval are necessary to ensure new drugs reach market in service of these goals.
Reshma Ramachandran, MD, MPP, is a family medicine physician and fellow in the Yale National Clinician Scholars Program. She leads efforts on drug pricing and FDA policy for the national non-profit organization, Doctors for America. Besides having published in the highest-impact peer-reviewed medical journals, we have also published editorials in The New York Times, Washington Post, Bloomberg Law, Baltimore Sun, The Nation, STAT News, The Hill, and CNN Opinion. Joshua Skydel, MD, is a resident physician in internal medicine at Dartmouth-Hitchcock Medical Center and led the study described in the op-ed. Joseph S. Ross, MD, MHS, is a Professor of Medicine and Public Health at Yale University with over two decades of research experience in pharmaceutical regulation, pricing, and reimbursement policy.
