Oncology surrogates under fresh scrutiny
Experts are warning that oncology approvals based on surrogate endpoints like response rates or progression‑free survival often don’t predict overall survival, prompting reversals and complicating post‑market benefit‑risk assessment. For safety teams, that means greater emphasis on monitoring whether early approval surrogates translate to clinical outcomes and on documenting the uncertainty in ongoing surveillance. (x.com)
Cancer drug approvals are being pushed back toward a simple question: do patients live longer, or live better? For years, the FDA has often let cancer drugs onto the market earlier by accepting substitute measures instead of waiting for overall survival data. Those measures are called surrogate endpoints. In oncology, the common ones are response rate, which asks whether tumors shrink, and progression-free survival, which asks whether scans stay stable for longer. The logic is obvious. Survival can take years to measure. Patients with advanced cancer do not have years to wait. The problem is that these early signals often do not tell the truth about what happens later. (fda.gov) That gap is no longer a side debate. It is now shaping how regulators think about cancer trials. In August 2025, the FDA issued draft guidance on overall survival in oncology trials that makes the agency’s position unusually plain. Overall survival is still the clearest patient-centered endpoint, and even when it is not the primary endpoint, the FDA wants it analyzed in advance as a safety signal. The agency’s own summary says the guidance is aimed at randomized oncology trials, with special emphasis on prespecified analysis of overall survival when some earlier endpoint is doing the work for approval. (fda.gov) That shift did not come from nowhere. A growing body of research has shown that many oncology surrogates have weak or inconsistent links to survival. A 2024 FAQ-style review in *eClinicalMedicine* stated the problem bluntly: across settings, most surrogate measures have weak correlation with outcomes that matter to patients. A separate review in the *European Journal of Cancer* reached the same place from another angle, finding that objective response rate and progression-free survival are widely used as surrogates for overall survival even though their validity remains unclear. (thelancet.com) Once you see that, a second problem comes into focus. Drugs approved on shaky surrogates do not always get rescued by later evidence. A 2024 *JAMA* study of cancer drugs granted accelerated approval from 2013 through 2017 found that after more than five years, 41 percent still had not shown an improvement in overall survival or quality of life in confirmatory trials, and results were still unavailable for another 15 percent. Many of those drugs were converted to regular approval anyway on the basis of progression-free survival or response-based measures rather than survival. (jamanetwork.com) That is why withdrawals have become part of the story, not an exception to it. The FDA’s Project Confirm was built to make outcomes from oncology accelerated approvals easier to track in public. Reviews of withdrawn cancer indications show the same pattern again and again: most were approved on response rate, a smaller number on progression-free survival, and the median time from approval to withdrawal was measured in years, not months. By the time the market corrects itself, patients and clinicians may have already built treatment decisions around a benefit that never materialized. (fda.gov) That lag is exactly why the FDA has also tightened the machinery around accelerated approval itself. In January 2025, the agency released draft guidance on what it means for a confirmatory trial to be “underway” before an accelerated approval is granted. Congress had already pushed the FDA in that direction, after years of concern that sponsors were too slow to produce the follow-up evidence needed to verify clinical benefit. The message is straightforward: if approval rests on a surrogate, the real test cannot be an afterthought. (fda.gov) For drug safety teams, this changes the job. It is not enough to log adverse events while assuming the efficacy side of the ledger will sort itself out later. If a drug reached patients because it improved scans or delayed progression, then post-market surveillance has to keep asking whether those early gains are translating into longer survival, better quality of life, or neither. The uncertainty belongs in the record from the start. The FDA’s recent work on overall survival says as much. When approval leans on early endpoints, survival still matters because it can reveal harm, missed benefit, or both. (fda.gov)
