Purpose: We examined 4 postulates: 1) radical perineal prostatectomy provides a substantial disease control benefit in men with clinically confined prostate cancer, 2) postoperative prostate specific antigen (PSA) levels are an excellent surrogate end point for defining disease control, 3) the biology of primary malignancy defines the interval to death after recurrence and 4) the interval from intervention to death from recurrence is so long that current series of alternative curative therapies have insufficient duration of observation to permit a comparison with the results of surgery.
Materials and methods: A total of 1,242 men with a median age of 65.2 years who had stage cT1 to 2 N0M0 disease underwent radical perineal prostatectomy. The final pathology specimen was characterized in regard to disease extent, and Gleason grade and score. Patients were followed at 2 weeks, at 2 months and then at 6-month intervals for biochemical, physical and radiographic evidence of disease recurrence. Outcome was evaluated by determining time to biochemical failure (PSA 0.5 ng./ml. or greater) and cancer associated death.
Results: Median time to noncancer death was 19.3 years. Median cancer associated death end point was not reached by patients with organ and specimen confined disease, while it was 12.7 years for margin positive disease. At 5 years 8, 35 and 65% of the patients with organ confined, specimen confined and margin positive disease, respectively, had PSA failure. This served as an excellent surrogate end point, preceding cancer associated death by 5 to 12 years depending on the biological aggressiveness predicted by Gleason grade or score. Biologically aggressive organ confined disease that had been surgically removed was associated with a high percentage of disease-free survival.
Conclusions: Our study confirms our postulates. It also provides guidelines for comparing therapies among institutions and emphasizes that enthusiasm for new treatments may be based on insufficient followup. Patient selection may severely bias outcome independent of treatment when death is used as the end point. Our study establishes the value of PSA as a surrogate end point.