Vital statistics following surgery or radiation for patients with clinically localized prostate cancer managed during the PSA era.

Reviewer: Neha Vapiwala, MD
Last Modified: May 31, 2003

Presenter: A. V. D'Amico
Presenter's Affiliation: Dana-Farber Cancer Institute
Type of Session: Scientific


    Various algorithms using pre-treatment risk groups to predict for PSA failure in prostate cacner patients currently exist, namely through single-institution studies by Roach et al and D'Amico, et al. The three main components of these algorithms are serum PSA level, Gleason score (GS) on biopsy, and clinical stage. However, PSA failure does not necessarily lead to mortality from prostate cancer, as many patients will die of non-cancer-related causes. This study was designed to determine whether these pre-treatment risk groups established at single institutions for PSA failure still retained validity in a multi-institutional setting for predicting prostate cancer-specific mortality (PCSM).

Materials and Methods

  • 7,316 pts treated at 44 centers in US between 1988-2002
    • 4,946 underwent surgery
    • 2,370 underwent radiation therapy
  • Clinical stage T1c-T2NxM0
  • Risk groups defined as follows:
    • Low risk
    • PSA
    • Intermediate risk
    • PSA > 10-20 OR GS 7 OR stage T2b
    • High risk
    • PSA > 20 OR GS 8-10 OR stage T2c
  • Cox regression analysis performed to determine accuracy of above pre-treatemnt groups in predicting time to PCSM
  • Relative risk of PCSM also calculated for each risk group, with low-risk pts serving as baseline risk of 1.0


  • Non-prostate cancer mortality rates at 8 yrs following PSA failure, by age:
    • < 70 yrs old 4% surgery vs. 15% radiation
    • >/= 70 yrs old 13% surgery vs. 18% radiation
  • Relative risk of PCSM, by primary treatment and risk group:
    surgery 14.2 vs.   4.9
    radiation 14.3 vs.   5.6

Author's Conclusions

  • Low-risk pts undergoing radical prostatectomy have virtually no cancer-related mortality and thus, if candidates, are well-managed with surgery alone.
  • Low risk pts undergoing rdaitaion therapy who are >/= 70 yrs of age also have minimal rates of prostate cancer mortality. However, the increased rates of non-cancer related mortality in irradiated pts who are < 70 yrs old likely reflects selection bias in this group. As pts under the critical age of 70 typically undergo surgery if they are medically cleared to do so, those who are irradiated but < 70 yrs old presumably have medical co-morbidities that both precluded surgery in the first place and probably led to the increased non-cancer mortality.
  • In high-risk pts, neither surgery nor radiation as monotherapies appear to be sufficient for reducing prostate cancer-related mortality.
  • Pre-therapy predictors of PSA failure appear to also predict for prostate cancer-specific mortality.

Clinical/Scientific Implications

    Significant rates of non-cancer-related mortality are seen in prostate cancer pts demonstrating PSA failure follwing management with either surgery or radiation therapy. The apparent prevalence of such deaths seen in this study among irradiated pts < 70 yrs old likely reflects the pt selection bias that often occurs, ie: non-surgical candidates often have serious medical co-morbidities and undergo radiation instead. With regards to predicting for prostate cancer-specific mortality, the utility of pre-treatment risk parameters for PSA failure (PSA, Gleason score and clinical stage) are validated by this study. Low risk pts who are > 70 yrs old or < 70 yrs old but surgical candidates appear to benefit from either prostatectomy or radiation. High risk groups have higher rates of death from disease if managed with surgery or radiation alone, making an argument for investigating post-operative radiation therapy/ adjuvant hormonal ablation. Finally, intermediate risk pts represent a heterogeneous group and demonstrate higher cancer mortality when more than one of the criteria (PSA, GS, stage) in that risk category are present. The authors suggest that in these cases, intermediate-risk pts would perhaps be more accurately placed in the high-risk group.

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

Frequently Asked Questions