Analyses have been performed every third year according to the protocol. The primary end points were death from any cause, death from prostate cancer (with death from other causes treated as a competing risk), and risk of metastases in bone, outside the pelvic area, or both. Secondary end points included initiation of androgen-deprivation therapy (with death from any cause treated as a competing risk).
We used Gray’s test to assess treatment effects.9 Effect sizes were quantified both by analyzing relative risks with 95% confidence intervals and by determining differences in cumulative incidence (with 95% confidence intervals). Relative risks were estimated with the use of Cox proportional-hazards models in cases in which proportionality was verified by means of visual inspection of the parallelisms of the logarithms of the estimated cumulative incidence. A cumulative incidence approach was used to account for competing risks among various causes of death.10 To assess the possible modification of the treatment effect, analyses were stratified according to the patient’s age at diagnosis (<65 years vs. ≥65 years) and tumor risk. The subgroup analyses were not included in the main protocol but were specified before the data were reviewed. Risk groups were defined with the use of Gleason scores from the pathological review as follows: low risk, PSA level less than 10 and either a Gleason score of less than 7 or WHO grade 1 (on a scale of 1 to 3, with higher grades indicating more aggressive disease) in tumors that were diagnosed only by means of cytologic assessment; high risk, PSA level of 20 or higher or a Gleason score greater than 7; and intermediate risk, all patients who did not fulfill the criteria for low or high risk. The modification of the effect of radical prostatectomy was tested in the Cox proportional-hazards model by including an interaction term between the subgroup category and randomization group.
The prevalence of the use of palliative treatment was calculated at every other year of follow-up, ending at 18 years after randomization. Palliative treatment was androgen-deprivation treatment (antiandrogen therapy or gonadotropin-releasing hormone analogues or orchiectomy) in patients with or without verified metastases and in patients with metastases who had received other palliative treatment (external or internal palliative radiation therapy, laminectomy, or chemotherapy drugs).
We used Gray’s test to assess treatment effects.9 Effect sizes were quantified both by analyzing relative risks with 95% confidence intervals and by determining differences in cumulative incidence (with 95% confidence intervals). Relative risks were estimated with the use of Cox proportional-hazards models in cases in which proportionality was verified by means of visual inspection of the parallelisms of the logarithms of the estimated cumulative incidence. A cumulative incidence approach was used to account for competing risks among various causes of death.10 To assess the possible modification of the treatment effect, analyses were stratified according to the patient’s age at diagnosis (<65 years vs. ≥65 years) and tumor risk. The subgroup analyses were not included in the main protocol but were specified before the data were reviewed. Risk groups were defined with the use of Gleason scores from the pathological review as follows: low risk, PSA level less than 10 and either a Gleason score of less than 7 or WHO grade 1 (on a scale of 1 to 3, with higher grades indicating more aggressive disease) in tumors that were diagnosed only by means of cytologic assessment; high risk, PSA level of 20 or higher or a Gleason score greater than 7; and intermediate risk, all patients who did not fulfill the criteria for low or high risk. The modification of the effect of radical prostatectomy was tested in the Cox proportional-hazards model by including an interaction term between the subgroup category and randomization group.
The prevalence of the use of palliative treatment was calculated at every other year of follow-up, ending at 18 years after randomization. Palliative treatment was androgen-deprivation treatment (antiandrogen therapy or gonadotropin-releasing hormone analogues or orchiectomy) in patients with or without verified metastases and in patients with metastases who had received other palliative treatment (external or internal palliative radiation therapy, laminectomy, or chemotherapy drugs).
