RE: FDA's accelerated approval pathway needs stronger transparency, evidence standards: ICER19 Apr 2026 12:17
In other instances, treatment with oncology products with initially positive signals on PFS has led to reductions in OS. This occurs most often when the magnitude of effect in tumor response or PFS is relatively small or treatment entails significant toxicity.93 The latter is often related to the long-standing industry practice of using maximum tolerated dose (MTD) in oncology trials. Dosing to MTD became an established norm in oncology during the development of cytotoxic chemotherapies, in which efficacy was inevitably linked with its adverse effects. The targeted therapies under development today break with that paradigm. Higher doses do not necessarily increase efficacy but do increase toxicity, which limits adherence to oral treatments delivered over longer timeframes, undermining clinical outcomes.93,94 For example, sotorasib (Lumakras®, Amgen), which received AA in 2021 for NSCLC, later became the subject of FDA concern regarding high-dose toxicity, leading the agency to request investigation of lower-dose alternatives. While survival on the higher dose was nominally longer than on the lower dose, side effects were also significantly more severe. Amgen nevertheless pushed ahead with developing the higher-dose version.95
Despite their frequent use, the correlation between surrogate endpoints such as PFS, objective response rate (ORR) and event-free survival (EFS) and OS varies widely, both within and across cancers and treatment types.96 While AA is designed to accept some degree of uncertainty, this raises questions about when and how to validate surrogate endpoints in oncology (and to varying degrees the other therapeutic areas of interest in AA as well). The FDA accepts validated surrogate endpoints, such as cholesterol lowering as a surrogate for cardiac events, in due course of regular approval, without the need for additional confirmatory trials. Although not rising to the level of validation, PFS and other intermediate endpoints may be more reliable predictors of OS in some types of cancers than others. Better understanding of these differences, which could be developed through meta-analyses of trials and/or observational studies, could allow FDA to manage AA more adaptively for cancer indications.
2 of 2