The cancer treatment landscape is a crowded and complex one. Though not perfect, it has been able to prolong the median survival of patients drastically over the past decades. For instance, in the case of breast cancer, around 75% of patients will survive their cancer for 10 years or more after diagnosis (Cancer Research UK, 2024). This is a significant achievement that is rightfully celebrated. It does pose a new challenge: it complicates studying the effectiveness of new treatments that want to enter the market. Imagine a new drug with an innovative mechanism of action with a potential to significantly improve patient outcomes. It would take at least 10 years to determine its superiority over current treatments in terms of overall survival (OS), because the existing treatments have already succeeded in stretching survival that long. This indicates that OS is a challenging criterion for new medicine developers to prove if they want their new medicine to be approved by a drug approval body, such as the European Medicines Agency (EMA) or the US Food and Drug Administration (FDA).
To address this issue, surrogate endpoints like progression-free survival (PFS) are often used to indicate the effectiveness of drugs during the approval process. PFS can generally be proven in a much shorter time span, making it a practical alternative. Regulatory agencies such as the FDA and the EMA rely heavily on PFS when considering medicines for approval. This shift is logical and beneficial for patients needing timely access to new treatments without having to wait for over a decade.
However, there are significant problems with how accurately PFS predicts overall survival, as highlighted by a recent retrospective analysis by Josh Elbaz, published in Cancer Medicine (Elbaz, 2024). This study is pivotal, as it investigates the overall survival of patients that were treated with medicines that have been approved based on surrogate criteria.
The study looked at 392 drugs that were approved by the FDA in the period of 2006-2023. The most important conclusion: