Accelerated approval and what happens when the confirmatory trial fails

Accelerated approval lets FDA clear a drug or biologic for a serious or life-threatening condition on the strength of an early read-out, before the long-term clinical benefit has actually been demonstrated. The trade-off is a binding promise: the sponsor runs a post-approval confirmatory trial to prove the benefit is real. This piece explains the standard FDA applies, the confirmatory-trial obligation, and, most importantly, what has actually happened in practice when those trials came back negative.

What accelerated approval is

Accelerated approval is one of FDA's expedited pathways for products that treat a serious or life-threatening disease. It rests on section 506(c) of the Federal Food, Drug, and Cosmetic Act and the implementing regulations at 21 CFR part 314 subpart H (drugs) and 21 CFR part 601 subpart E (biologics) ¹⁵¹⁸. Rather than requiring proof of clinical benefit up front, the statute allows approval "upon a determination that the product has an effect on a surrogate endpoint that is reasonably likely to predict clinical benefit, or on a clinical endpoint that can be measured earlier than irreversible morbidity or mortality, that is reasonably likely to predict an effect on irreversible morbidity or mortality or other clinical benefit, taking into account the severity, rarity, or prevalence of the condition and the availability or lack of alternative treatments" ²⁵.

Two endpoint types can support it:

A surrogate endpoint is "a marker, such as a laboratory measurement, radiographic image, physical sign, or other measure, that is thought to predict clinical benefit, but is not itself a measure of clinical benefit" (in oncology, tumor response rate is the classic example) ⁴⁶.
An intermediate clinical endpoint is "a measurement of a therapeutic effect that can be measured earlier than an effect on IMM and is considered reasonably likely to predict the drug's effect on IMM or other clinical benefit" ¹²²².

The key word in both is predict. Pharmacologic activity alone is not enough: FDA expects "[c]linical data ... to support a conclusion that an effect on the surrogate endpoint or intermediate clinical endpoint is reasonably likely to predict the intended clinical benefit," and whether an endpoint clears that bar is a judgment based on biological plausibility and empirical evidence ³¹⁷. That "reasonably likely" standard is deliberately lower than the "established" benefit standard for traditional approval, which is why the confirmatory trial matters so much.

The bargain: confirmatory trials

Accelerated approval is conditional. For products cleared this way, "postmarketing confirmatory trials have been required to verify and describe the anticipated effect on IMM or other clinical benefit," and "[t]hese trials must be completed with due diligence" ¹⁴. Sponsors submit progress reports to FDA roughly every 180 days ¹⁰. If the trial verifies benefit, FDA generally terminates the requirement and the approval effectively converts to a regular approval ¹⁴.

FDA's oncology-specific guidance is more prescriptive about how to run these trials so the verification period is short and interpretable. It recommends randomized controlled confirmatory trials, structured either as two separate trials (one on an early endpoint for the accelerated approval, one powered for a longer-term endpoint) or as a single randomized "one-trial" approach powered for the longer-term endpoint and followed to verify benefit ²²⁶. It "strongly recommends" the confirmatory trial be well underway, if not fully enrolled, by the time of the accelerated-approval action, because enrollment often becomes harder once the drug is on the market ²²⁶. FDA's stated aim is to minimize the time a drug stays on the market without verified benefit; in oncology, the median time from accelerated approval to verification has run at roughly three years ²²⁸²²⁹.

What happens when the confirmatory trial fails

This is the crux. Failure to verify benefit is one of the enumerated statutory grounds for pulling the product. Under the expedited withdrawal procedures in section 506(c)(3)(B), FDA may withdraw an accelerated approval when the confirmatory study fails to verify and describe the predicted clinical benefit, when the required study is not conducted with due diligence, when other evidence shows the product is not safe or effective under its labeled conditions, or when promotional materials are false or misleading ²⁷. In plain terms, a negative confirmatory trial does not automatically or instantly remove the drug: it triggers a defined administrative process, and in many cases a negotiation, before the indication comes off the market.

In practice, two routes have dominated.

Voluntary withdrawal by the sponsor. A sponsor may ask to withdraw the indication itself. When it does, FDA can either suspend the expedited process and proceed under the ordinary withdrawal/revocation provisions or continue on the expedited track ³³. Much of the 2021-2023 clean-up of "dangling" oncology indications went this way, with companies electing to drop specific accelerated-approval indications after confirmatory data disappointed rather than contest a withdrawal.

Contested withdrawal. If the sponsor does not agree, FDA runs the formal process: it gives the sponsor due notice and an explanation, provides an opportunity for public comment and publishes a summary of and response to those comments, and lets the sponsor file a written appeal and request a meeting with the Commissioner or designee ²⁷³¹³⁴. If the sponsor requests it, FDA must provide an opportunity for an advisory committee meeting on the withdrawal, unless a committee has already advised FDA on the same issues ³¹³⁴. FDA says it generally intends to convene the advisory committee before proposing withdrawal, and that a single such meeting can satisfy the statutory requirement ³⁴. On appeal, FDA decides on the docket record, the application file, and any advisory committee record; where the studies do not verify benefit or the evidence does not show safety and effectiveness, "FDA generally intends to withdraw accelerated approval absent unusual circumstances" ³⁴. The Oncologic Drugs Advisory Committee (ODAC) has been the recurring venue for these deliberations.

Illustrative cases

Bevacizumab (Avastin), metastatic breast cancer. This is the textbook example of the surrogate that did not hold up. Avastin received accelerated approval in first-line metastatic breast cancer, in combination with taxane-based chemotherapy, based on the surrogate of progression-free survival (PFS) ³⁷. The confirmatory evidence was not persuasive: of the two randomized trials submitted, AVF2119g failed to show an effect on PFS or overall survival, and while E2100 was positive for PFS, FDA judged the estimate uncertain because substantial loss to follow-up before progression confirmation weakened the independent review ³⁷. FDA ultimately revoked the breast cancer indication, which was "later withdrawn due to lack of confirmation of clinical benefit" ³⁶. The drug itself remained on the market for its other, separately supported indications; only the unverified indication was removed.

Pembrolizumab (Keytruda), esophageal and gastric cancer. The PD-1 inhibitor picked up several accelerated approvals on single-arm response data, and not all of the confirmatory trials delivered. For esophageal squamous cell carcinoma, the confirmatory KEYNOTE-181 trial did not meet its prespecified threshold for a statistically significant overall-survival improvement in any of its three co-primary populations (esophageal SCC HR 0.77, 95% CI 0.63-0.96; PD-L1 CPS greater than or equal to 10 HR 0.70, 95% CI 0.52-0.94; all randomized HR 0.89, 95% CI 0.75-1.05) ⁹⁴⁹⁵. For third-line gastric/GEJ adenocarcinoma, FDA had granted accelerated approval on 22 September 2017 based on the single-arm KEYNOTE-059 cohort (objective response rate and duration of response), with KEYNOTE-061 designated as the confirmatory trial ¹⁸³¹⁸⁴¹⁹²²⁰². These indications were among those reassessed once the confirmatory data were in hand, and the gastric accelerated approval illustrates the standard structure: an early-endpoint approval tied to a specific randomized confirmatory study whose result governs whether the indication stays.

Duvelisib (Copiktra), relapsed/refractory follicular lymphoma, and the PI3K-inhibitor class. Copiktra was granted accelerated approval on 24 September 2018 for relapsed or refractory follicular lymphoma after at least two prior systemic therapies, based on a single-arm phase 2 study (IPI-145-06) showing an objective response rate of about 42% ¹⁶⁷¹⁶⁸¹⁶⁹¹⁷⁷. As a condition of that accelerated approval, FDA required a postmarketing randomized phase 3 confirmatory trial to verify and isolate clinical benefit, with defined protocol, interim, and final-report milestones ¹⁷⁶. The review flagged a class-wide safety signal that later proved decisive across PI3K inhibitors: as early as March 2016, several phase 3 trials of the related agent idelalisib had shown increased serious adverse events and decreased overall survival, and were terminated for increased toxicity and reduced survival ¹⁷². This case shows that confirmatory-trial failure is not only about efficacy that does not pan out; a confirmatory trial (or class data) that surfaces an overall-survival detriment is an even stronger basis for removing an accelerated-approval indication.

The 2023 FDORA reforms

Congress tightened the pathway in the Food and Drug Omnibus Reform Act (FDORA), enacted as part of the Consolidated Appropriations Act, 2023, amending section 506(c) ²⁷²⁹. Three changes matter most for how confirmatory-trial failure now plays out:

Confirmatory trials underway before approval. FDA may require that the confirmatory study or studies be underway prior to accelerated approval; if FDA makes that determination and the trial is not underway, it does not intend to grant accelerated approval until the deficiency is addressed ²⁸²⁹. In practice FDA generally intends to require this for programs seeking accelerated approval, and in some cases may require enrollment to be complete at approval ²⁸. This directly attacks the old failure mode where a confirmatory trial had barely started (or had not started) years after approval.
Specified conditions at approval. The amendments require FDA to specify confirmatory-study conditions at the time of approval, which can include enrollment targets, the protocol, milestones, and a target completion date; failing to conduct the trial with due diligence, including meeting those conditions, is itself a ground for expedited withdrawal ²⁸²⁹.
Streamlined withdrawal. FDORA added expedited withdrawal procedures, giving FDA a faster, codified route to remove an indication when the confirmatory obligation is not met, while preserving the notice, public-comment, appeal, and advisory-committee rights described above ²⁷²⁹.

What this means for regulatory teams

Accelerated approval is best understood as a conditional, revocable authorization, not a lighter-weight version of full approval. The confirmatory trial is the deliverable that discharges the condition, and a negative result puts the indication squarely into the withdrawal framework. The historical record shows the failure does not erase the whole product: FDA and sponsors typically remove the specific unverified indication while other, adequately supported indications remain, most often by voluntary withdrawal, sometimes after an ODAC discussion, and occasionally through a contested proceeding. Post-FDORA, the practical center of gravity has shifted upstream: designing a credible, already-enrolling confirmatory trial (ideally randomized and powered for a clinical endpoint) at the time of the marketing application is now the single most important way to protect an accelerated-approval indication from later withdrawal.

If you want to go deeper, natural follow-ups to ask Rhizome include the full list of oncology accelerated-approval indications withdrawn since 2021 and their ODAC votes, the confirmatory-trial status of a specific product in your portfolio, or how EMA's conditional marketing authorization compares with FDA's pathway on the same molecules.