Can biochemical failure (ASTRO definition) be used as a surrogate endpoint for prostate cancer survival in phase III localized prostate cancer clinical trials? Analysis of RTOG protocol 92-02

Reviewer: Ryan Smith, MD
Last Modified: May 31, 2003

Presenter: H.M. Sandler
Presenter's Affiliation: RTOG
Type of Session: Scientific


  • Prostate cancer is a disease with a long natural history, so the use of survival as an endpoint requires a very long follow-up to reach conclusions
  • Therefore, PSA-defined failure has been used as a surrogate endpoint, shortening the time needed for reporting results.
  • This study eveluated the use of PSA in this respect to determine if it should continue to be used as a surrogate endpoint in studies regarding prostate cancer

Materials and Methods

  • Data obtained was from RTOG 92-02, which treated patients with T2c-4 prostate cancer with goserelin and flutamide 2 months prior to beginning radiation therapy and during radiation therapy. These patients were then randomized to no further therapy (short term androgen ablation (STAD)) vs. 24 additional months of goserelin (Long term androgen ablation (LTAD)).
  • 1554 patients were entered, with a median follow up of 5.8 years
  • The ASTRO definition of PSA failure was used, which is defined as three consecutive rises in PSA, with the time of failure backdated to halfway between the first rise and the last nonrising PSA
  • Four criteria were to be met to accept PSA failure as an acceptable surrogate endpoint for survival: 1) Treatment must be prognostic for the true endpoint of prostate cancer survival 2) Treatment must be prognostic for the surrogate endpoint of PSA failure 3) The surrogate endpoint of PSA failure must be prognostic (associated with) the true endpoint of survival 4) The full effect of the treatment (and the differences of the treatment arms) on the true endpoint of survival must be explained by the surrogate endpoint of PSA failure


  • There have been only 142 prostate cancer deaths but 542 PSA failures
  • By criterion: 1) was satisfied, as treatment was prognostic for overall survival (LTAD patients had better prostate cancer survival) 2) was satisfied, as treatment was prognostic for PSA failure (LTAD patients had less PSA failure, p<.0001) 3) was satisfied, as those who developed PSA failure had a worse survival. The relative risk of patients with biochemical failure to die of prostate cancer was 9.49. In those who developed biochemical failure, 70% died of prostate cancer and 30% died of other causes. In those who did not develop PSA failure, 7-12% died of prostate cancer and 88-93% died of other causes 4) was not satisfied, as those who failed biochemically had different survivals depending on their treatment arm. Those treated with LTAD who had PSA failure had a relative risk of prostate cancer death of 1.5 compared to those treated with STAD

Author's Conclusions

  • RTOG 92-02 demontrated an advantage to LTAD
  • PSA failure is associated with an increased risk of prostate cancer deaths
  • The risk of death after biochemical failure is increased in the LTAD arm
  • PSA failure may not mean the same thing in different arms of a protocol
  • Therefore, it may not be a good surrogate marker in trials where treatments differ by the duration of androgen deprivation

Clinical/Scientific Implications

    Overall survival is undebatably still the best endpoint in any study looking at patient outcome. This study proves the use of PSA as a surrogate endpoint must be used with caution. However, as also proved in this study, PSA failure does accurately predict death by prostate cancer. Other studies, in validating risk group stratification, have also shown that PSA failure predicts for death by prostate cancer. Therefore, it can be used and should be used as an surrogate endpoint in prostate cancer studies, as data using survival as an endpoint may not mature for 15 or 20 years. However, their conclusion that PSA should be used with caution is well-taken, although this conclusion is not surprising. In this study, as in many other studies, when patients fail after more aggressive therapy, they tend to do worse in terms of survival. This is true in (for example) lymphoma when patients fail after combined modality treatment compared to radiation alone, breast cancer when comparing patients who fail after lumpectomy vs. after mastectomy, and in numerous pediatric cancers, to name a few. In this case, it is likely that longer androgen ablation allows for the selective growth and subsequent failure of hormone insensitive cells. Therefore, as PSA will almost certainly continue to be used, data such as that presented here, should be kept in mind.

Oncolink's ASCO Coverage made possible by an unrestricted Educational Grant from Bristol-Myers Squibb Oncology.