Understanding Progression-Free Survival and Overall Survival
Overall survival (OS) is the time from treatment initiation until death from any cause, while progression-free survival (PFS) is the time from treatment initiation until disease progression or death, whichever occurs first 1.
Overall Survival (OS)
OS represents the gold standard endpoint in cancer clinical trials because it is:
- The most error-free measurement - death is an unambiguous endpoint
- The most clinically relevant outcome - it directly measures how long patients live
- An important safety endpoint - it captures all therapy-driven toxicities and their impact on mortality 2
OS measures the ultimate outcome that matters most to patients: survival time. It is measured from the moment of treatment randomization (or initiation) until death from any cause 1.
Progression-Free Survival (PFS)
PFS measures the time from treatment start until either the cancer worsens or the patient dies, whichever happens first 1.
PFS is considered a "softer" endpoint than OS but remains scientifically and clinically relevant 1. It captures:
- Disease control - how long treatment keeps cancer from growing
- Direct treatment effect - the immediate impact of therapy without confounding from subsequent treatments
- Earlier signal of benefit - PFS events typically occur much sooner than death
Why PFS Matters
In many cancer settings, PFS may be the only feasible endpoint to demonstrate treatment benefit. For example, in metastatic breast cancer where median OS exceeds 2 years but median PFS is only 6 months, multiple effective salvage therapies after progression can obscure any OS benefit from first-line treatment 1. Even an effective experimental treatment may not show OS improvement with reasonable sample sizes because subsequent therapies dilute the signal 3.
Critical Relationship and Discordance
Important Caveat: PFS Does Not Always Predict OS
Recent evidence demonstrates that PFS improvement does not guarantee OS benefit and can even occur alongside OS harm 2. Notable examples include:
- PI3K inhibitors in hematologic malignancies showed PFS benefit but potential OS detriment
- PARP inhibitors in ovarian cancer demonstrated similar discordance
- Venetoclax plus bortezomib in multiple myeloma showed favorable PFS but patients were twice as likely to die compared to bortezomib alone 2
Why This Discordance Occurs
The relationship between PFS and OS depends heavily on:
- Survival post-progression (SPP) - the time patients live after disease worsens
- Treatment toxicity - significant toxicities can offset modest efficacy gains
- Subsequent therapies - effective post-progression treatments dilute OS differences
- Crossover rates - when >50% of control patients receive the experimental drug after progression, detecting OS benefit becomes substantially harder 4
When median SPP is long (>12 months), OS becomes too high a bar for detecting benefit because SPP variability dilutes the OS comparison, causing loss of statistical significance even when real OS benefit exists 3.
Clinical Implications
For treatment decisions, OS remains the definitive measure of net clinical benefit or harm 2. When evaluating cancer therapies:
- Prioritize OS data - even when PFS is the primary endpoint, OS must be collected and analyzed to assess for potential harm 2
- Be cautious with PFS-only benefits - improvements in PFS without OS data or with concerning OS trends should not be assumed to represent meaningful clinical benefit
- Consider the context - in diseases with short SPP (<12 months), PFS and OS are more likely to correlate; in diseases with long SPP, they may diverge significantly 3