How to read a modern phase 3 oncology readout: hazard ratios, crossover and what regulators care about

What a phase 3 oncology readout is, and what it has to deliver

A phase 3 oncology trial is a prospective, typically randomised comparison of an investigational therapy against a comparator (standard of care or placebo, depending on the indication and ethical setting). The trial has primary endpoints (the questions it is designed to answer), secondary endpoints (supporting questions), and a set of prespecified analyses defined in the statistical analysis plan before the data are unblinded.

The endpoints that matter most for approval and reimbursement in oncology are typically:

• Overall survival (OS): time from randomisation to death from any cause
• Progression-free survival (PFS): time from randomisation to disease progression or death
• Objective response rate (ORR): the proportion of patients achieving a defined response on imaging
• Duration of response (DoR): how long responses last in the patients who achieve them

OS is the gold-standard clinical benefit endpoint. PFS, ORR and DoR are surrogate endpoints that can support approval depending on the indication and pathway.

The numbers behind the headline

A headline like "the trial met its primary endpoint of progression-free survival, hazard ratio 0.65, p less than 0.001" is reporting that, in the trial population, patients on the investigational arm progressed or died at 65 percent of the rate of patients on the comparator arm, and that this difference is unlikely to be due to chance. The hazard ratio is the most-watched number in a phase 3 oncology readout.

But the hazard ratio is not the whole story:

• Event maturity. The hazard ratio is more reliable when a high proportion of expected events have occurred. Early hazard ratios from less mature data can shift as the trial matures.
• Crossover. If patients on the comparator arm crossed over to the investigational arm at progression, the OS comparison may understate the investigational arm's benefit. Trials handle this with prespecified statistical adjustment, but the adjustment itself becomes part of the regulatory conversation.
• Subgroup analysis. Prespecified subgroups (by biomarker, line of therapy, geography) inform the label but require careful framing. Post-hoc subgroup analysis is a flag for regulators.

What regulators and payers weigh

Regulators are looking for a clear, prespecified primary endpoint result, supported by a coherent secondary endpoint picture and an acceptable safety profile. Payers and HTA bodies are looking at the same plus the magnitude of benefit relative to existing standard of care, the quality-of-life impact, and the budget shape over the indication's eligible population.

The trial readout that lands cleanly with both audiences is the one where the headline is supported by mature event data, the secondary endpoints are coherent, and the subgroup analysis does not introduce a complication the label has to navigate.