In cancer research, nothing is more important than proving whether a treatment helps patients live longer. That’s why “overall survival” (OS) is considered the gold standard in oncology trials—it’s clear, objective, and clinically meaningful. The FDA’s new draft guidance sheds light on how sponsors should design, analyze, and report OS data to ensure that cancer therapies are both effective and safe.
Overall survival, measured as the time from randomization until death from any cause, remains the most direct way to evaluate whether a new therapy is truly benefiting patients. Unlike surrogate endpoints such as progression-free survival (PFS) or response rate, OS offers undeniable clinical value. A drug may shrink tumors or delay disease progression, but if those benefits come at the cost of higher toxicity or don’t translate into longer life, patients could ultimately be harmed.
The FDA emphasizes that OS should be prioritized as the primary endpoint in oncology trials when feasible, particularly in diseases with short survival times or where competing therapies already show survival benefits. However, in slow-progressing cancers or trials with highly effective treatments that extend life significantly, OS may not always be practical as the main endpoint. In these cases, sponsors are still encouraged to collect and analyze OS as a safety endpoint, ensuring new drugs are not inadvertently shortening lives.
The guidance makes some key recommendations for sponsors. All randomized oncology trials should include OS assessment. Whether used as a primary or secondary endpoint, studies must incorporate pre-specified plans for evaluating OS, including the possibility of interim analyses for futility or harm. The guidance stresses caution with crossover designs and unequal randomization, both of which can complicate OS interpretation.
The FDA recommends pre-specifying statistical methods in a trial’s protocol and statistical analysis plan (SAP). Standard approaches include hazard ratios (HRs) from Cox models, log-rank tests, and Kaplan-Meier survival curves. To account for uncertainty, sponsors should plan sensitivity and supplementary analyses—such as restricted mean survival time—to ensure results remain robust under different assumptions.
Even if OS is not the primary endpoint, it must be analyzed to rule out harm. The FDA suggests setting pre-specified thresholds to detect potential survival detriment. Interim analyses are encouraged in larger trials to minimize patient exposure to unsafe treatments. Subgroup analyses (e.g., by age, sex, race, or biomarker status) should be pre-specified when biologically plausible. While exploratory, these analyses can highlight populations that may benefit—or be harmed—differently from a treatment.
For accelerated approvals, surrogate endpoints may support early market entry, but sponsors will likely face post-marketing requirements to collect mature OS data. Ultimately, OS plays a central role in the FDA’s benefit-risk assessment and in decisions about whether accelerated approvals should be converted to traditional approvals.
By clarifying expectations around OS assessment, the FDA aims to strengthen the integrity of oncology trials, improve patient safety, and ensure that survival benefits are both real and reliable. For sponsors, this means more rigorous planning and transparency. For patients, it means a greater likelihood that approved therapies not only show promise on paper but also extend lives in practice.
