Three Problems with Accelerated Access: Will They Be Overcome?

This is the fourth in a series of posts on reforming the U.S. drug approval process. Part 1, Part 2, Part 3.

Cite this Article
Roger Bate, Three Problems with Accelerated Access: Will They Be Overcome?, Truth on the Market (January 10, 2024),

This post discusses three important problems with the Food and Drug Administration’s (FDA) accelerated-approval process. The first is that regulatory authorities and patient groups maintain that, legally, the standards of accelerated approval are the same as standard approval. Yet from a risk perspective, the standards are quite different; by shifting risk taking from regulator to patient, physician, and payer, this creates problems. The second problem is more practical and is generally considered the most significant problem with accelerated approval. Some companies that have received accelerated approval for their products have not done confirmatory studies, as required by their agreement with FDA. This leads to distrust of these companies and their products, and threatens to undermine the accelerated-approval program. The third problem is the issue of approving medicines with marginal benefits.

Surrogate Endpoints and Clinical Benefit

Improved T-cell count was determined to reliably predict fewer infections in AIDS patients and was accepted as a surrogate endpoint that could be used to demonstrate the efficacy of HIV/AIDS drugs. AZT, the first medicine approved to combat HIV, improved T-cell counts and was provisionally approved on March 20, 1987, based on the surrogate-endpoint data. 

The AZT approval convinced many scientists, physicians, and regulators that a successful drug could be predicated on the use of surrogate endpoints. As consensus grew about their utility in clinical-trial design, the FDA embraced drug-approval reform and promulgated regulations formalizing the accelerated-approval pathway in 1992.

Political Decisionmaking and Risk Tradeoffs

Weighing risks and benefits underlies all drug approvals. Even when a clinical benefit is shown in a trial, it doesn’t mean that it will sustain for all—or even most—patients. But it is obvious that relying on a surrogate endpoint instead of a clinical benefit lowers the odds of a successful clinical outcome. As a result, the FDA has demanded that any company promoting a drug for accelerated approval based on surrogate-endpoint data must conduct a post-marketing confirmatory study if the drug was approved. If that study does not confirm a significant clinical benefit, then the drug should probably be withdrawn.

The FDA has discussed whether accelerated approval was creating a different standard, which might require changes in regulatory oversight and even the law, but it has to date rejected the notion of a separate legal standard. On this, the agency is unequivocal: “[t]he evidence available at the time of approval under this rule will meet the statutory standard, in that there must be evidence from adequate and well-controlled studies showing that the drug will have the effect it is represented to have in its labeling.” Accelerated approval did not represent a “lower standard,” nor one “inconsistent with section 505(d) of the Act,” but rather an approval based on assessment of a different type of data demonstrating “that the same statutory standard has been met.”[1]

In 2012, Congress codified the accelerated-approval pathway by enacting the Food and Drug Administration Safety and Innovation Act (FDASIA), which amended the Food Drug and Cosmetic Act. In codifying the accelerated-approval pathway, Congress acknowledged the vital role it served for patients with poorly served diseases, and expressed their hope that it would bring lifesaving drugs to the market expeditiously. Congress also affirmed the FDA’s conclusion that accelerated approval did not create a different standard for drug approval, stating that accelerated approval “may result in fewer, smaller, or shorter clinical trials… without compromising or altering the high standards of the FDA for the approval of drugs.[2]

That Congress and the FDA would consider accelerated approval to be the same standard as normal approval is both understandable, and also problematic. It was understandable because the FDA was approving a medicine based on some positive change (a surrogate endpoint) that was linked to a clinical benefit. They claimed that this positive change meant that the legal standard had been established and that this change was reflected in what was claimed on the label. Approval based on such changes allows both the agency and its overseers in Congress to not have to acknowledge true risk analysis, and not change any laws or loosen their grip on the power of approval.

It is problematic for the same reason. The surrogate-endpoint change may turn out to be clinically irrelevant for many patients. Without a real assessment of risk and benefit, and of how the speed of approval based on surrogate-endpoint data changes that calculus, accelerated approval is a limited change in policy. After all, decisionmaking on more speculative data means more risk is assumed by the patient and physician (and also the insurer or other payer). The increased importance of these other players in making decisions is not acknowledged by either the FDA or Congress.

After all, the ramifications are significant. Will physicians be sued by patients or patients’ relatives when there is a bad outcome? And perhaps more significantly, are all payers duty bound to cover accelerated-approval medicines in the same way as normally approved medicines? Notably, the Centers for Medicare & Medicaid Services (CMS) and some U.S. states (notably, Oregon) want to limit payment for some accelerated-approval medications, or remove such medications from key formularies until they pass regular FDA approval. The concept of risk taking and the more practical payment mechanisms are not directly linked, but a greater appreciation of the former could inform discussions about who pays for drugs and under what circumstances. These issues will be discussed in future posts, and also later in this post in the section about approving drugs with marginal efficacy.

Complaints About Accelerated Approval

Most public and media-reported complaints about accelerated approval appear to be driven by a distrust of pharmaceutical companies. Some of these complaints are entirely legitimate. A key failure of the existing system is that some companies are not fulfilling their obligations to do confirmatory studies on the products approved under accelerated approval. As National Public Radio recently reported, the manufacturer of one oncology drug (Clolar) approved 20 years ago has still not completed the required confirmatory study. When companies get accelerated approval, they promise to continue studying the medicines, complete existing trials and, if required, start and finish completely new ones. Continuing to sell drugs without doing the trials (and, potentially, finding out the drug doesn’t work) is financially appealing to companies, but it is suboptimal as a strategy. Over time, companies can lose the trust of regulators and potentially the benefits of the accelerated-approval system. There are good reasons why trials take a long time for rare diseases, since it is especially difficult to find patients and, hence, trial participants. But continuing to study drugs is the most important requirement for accelerated approval. If companies do not do them, products should have approval withdrawn. 

Completion of trials may not be a priority for some companies, but patients remain supportive of fast approval, nonetheless. As NPR reports, even parents of children who have died from cancer—and where the accelerated drug (Clolar and others) has failed or been potentially harmful—are supportive of faster access to drugs, even if those treatments lack confirmatory studies. They are not interested in suing the company, but they are also critical of the companies for not completing the trials and of the FDA for not making them do so. This finding is echoed in my own interviews with oncologists, the vast majority of whom said patients’ families had no interest in causing problems for companies making potential breakthrough drugs, even if they flouted regulations on study timelines.

Some medications probably aren’t withdrawn quickly enough. For example, some cancer medications stay on the market even after studies fail to show a benefit. This is arguably a far greater issue. If one has a drug approved early based on certain surrogate endpoints, but then fails to deliver the expected clinical benefit, it should be pulled immediately. The failure of drug companies to do legislatively required follow-up studies or withdraw ineffective medicines provides support for those who would slow approval processes. 

But as explained in the previous post, the vast majority of trials are completed, and drugs that were approved are either withdrawn (if ineffective) or moved to regular approval. In total, 26 cancer approvals have been withdrawn for failing to demonstrate clinical benefit after confirmatory studies, whereas 96 accelerated approvals to treat cancers have gone on to show undeniable clinical benefit. The remainder are still approved, with some clinical benefits shown, but to an uncertain extent and are still being investigated. 

Over the past dozen years—during which, oncology medicines dominated accelerated approvals—the median time from approval to withdrawal of approval was 3.5 years, and conversion from accelerated to regular approval took a median time of 2.3 years.

Drugs of Marginal Benefit

The third problem links the first two. Is the FDA approving medicines with no obvious benefits? 

Most of the complaints about underperforming drugs center on treatments for Alzheimer’s disease. The FDA approved Aduhelm (aducanumab) in 2022 for the treatment of AD with limited clinical data to support improvement in patient outcomes, but with evidence that it did reduce amyloid plaques, which are strongly associated with the disease. There was debate about whether this approval should have been given, and even more debate about whether insurers and Medicare should cover the drug. 

The Aduhelm decision may have hardened resolve against faster approvals. It certainly has among a subset of physicians and academics, if the remarks in the CMS-required public consultation are any guide. Of course, approving Aduhelm encouraged other researchers (and their venture capital backers) and drug companies to drive even more aggressively in trying to combat this terrible affliction, which prior to Aduhelm had no treatment options for the disease itself. 

At one level, with millions suffering from the condition and extremely limited treatment options (the last of which was approved in 2003), it seems like it was the right decision to allow patients who want to risk trying a new and marginally effective drug. But it was unlike most prior FDA decisions, partly due to the enormous potential market for the product. It’s a case almost custom-designed to create debate over accelerated approval. Some physicians and physician groups have criticized the FDA, and there is a congressional investigation into the decision to approve Aduhelm.

The drug had safety risks and marginal benefits, but it was left to physicians and patients to decide whether to take it. At least, that was the plan, but insurers, the CMS, and other payers refused to fund it outside of very restrictive clinical trials. 

It did, however, encourage other trial sponsors to file for approval. In 2023, lecanemab (Leqembi) was approved, with a better clinical profile than Aduhelm. And in March 2023, the U.S. Department of Veterans Affairs (VA) announced it would provide lecanemab to veterans who meet key criteria, something that had not happened with Aduhelm. This set the VA at odds with Medicare, which has still not approved lecanemab. 

Approving Aduhelm was arguably the correct decision, but it certainly shifted decisionmaking from FDA to physicians and payers, leading to much-heated arguments about the future of rapid approvals.


[1] 21 C.F.R. §§ 314.500 et seq. & 601.40 et seq.; 57 Fed. Reg. 58942, 58942 (Dec. 11, 1992)57 Fed. Reg. at 58944. 38 Id. at 58943–44 (emphasis added). 39 Id. at 58944

[2]  See 21 U.S.C. § 356(c). In the codified version, Congress made two notable modifications to the agency’s regulations. First, the original requirement that the new drug provide a meaningful therapeutic benefit to patients over existing treatments was replaced with the more flexible “tak[e] into account . . . the availability or lack of alternative treatments.” Second, Congress specified that accelerated approval “may” be subject to one or both of two requirements: (1) that appropriate post-approval studies be conducted to verify and describe the predicted effect on the surrogate or intermediate clinical endpoint; and (2) pre-dissemination review of promotional materials. Id. In practice, these changes have not altered the FDA’s practices. See H.R. Rep. 112-495, *35–36 (2012). 43 See 158 Cong. Rec. H3825-01, H3848 (2012)