Progression-free survival is simply a measure of a drug's effect while administered and is not a surrogate for overall survival

Cancer J. Sep-Oct 2009;15(5):379-85. doi: 10.1097/PPO.0b013e3181bef8cd.


Since 1992, the Food and Drug Administration has allowed for accelerated approval of cancer drugs in cases where a statistically significant and clinically meaningful improvement in survival or side effect over alternative therapies is clearly demonstrated in controlled randomized trials. To this effect, endpoints other than overall survival (OS) are accepted as surrogate markers for survival. A lack of consensus exists regarding the validity of progression-free survival (PFS) as a true measure of outcome due to treatment. To quantitatively evaluate the correlation between the magnitude of the difference in PFS and that of OS in randomized trials conducted in patients with metastatic solid tumors, we performed electronic searches for trials that reported a statistically significant improvement in PFS, OS, or both for 1 treatment arm over another. We cataloged variables of interest such as the hazard ratio (HR) for PFS and HR for OS from 66 studies that met inclusion criteria and calculated the difference (delta) in response rate (RR), PFS, and OS. We performed linear regressions between the differences in OS and the differences in PFS as well as between both those values and RR. We also examined the correlations of HR for PFS and HR for OS. Regression analysis of both delta PFS/delta OS (R(2) = 0.49; P < 0. 0001) and HR PFS/HR OS (R(2) = 0.62, P < 0.0001) showed strong statistical evidence that an increase in PFS is associated with an increase in OS of the same magnitude. Both PFS and OS showed strong association with RR. These results were reinforced when we looked at the 3 largest subgroups by cancer type. We found very little overlap in comparative analyses of genes annotated to terms of cell growth and death. We conclude that PFS is not a surrogate for OS; rather it is a straightforward measure of on-therapy benefit and is not predictive of tumor growth after the cessation of treatment. The magnitude of the improvement in PFS is the magnitude of the improvement in OS. PFS is simply a measure of a drug's effect on tumor growth while it is administered and is not a surrogate for OS. Although PFS should not replace OS in regulatory approval consideration should be given to studies that treat beyond current definitions of progressive disease as a strategy to augment OS.

Publication types

  • Review

MeSH terms

  • Antineoplastic Agents / therapeutic use*
  • Disease Progression
  • Disease-Free Survival
  • Humans
  • Neoplasm Metastasis / drug therapy
  • Neoplasms / drug therapy*
  • Neoplasms / pathology*
  • Randomized Controlled Trials as Topic
  • Regression Analysis
  • Treatment Outcome


  • Antineoplastic Agents