Eighteen RCTs and one pooled analysis of 3 trials (n=14,595) and 473 observational studies were included.
No intervention was supported by consistent results from two high-quality, adequately powered RCTs with adequate follow-up.
Limitations in the identified evidence included a paucity of studies directly comparing major treatments, underpowered RCTs, failure to report long-term survival rates, RCTs conducted in the era before PSA testing, wide variation in the reporting and definitions of outcomes, and failure to report the results stratified by patient and tumour characteristics.
No RCTs evaluated cryotherapy, laparoscopic- or robotic-assisted radical prostatectomy, primary androgen deprivation, high-intensity focused ultrasonography, proton beam radiation or intensity-modulated radiation therapy.
There were 2 RCTs of radical prostatectomy versus watchful waiting. The larger RCT (n=695 men, most without PSA-detected disease) reported that radical prostatectomy reduced all-cause deaths (24% versus 30%, p=0.04) and disease-specific deaths (10% versus 15%, p=0.01) at 10 years, but increased urinary and sexual dysfunction compared with watchful waiting. The smaller, older RCT (n=142) reported no significant difference between treatments in overall survival.
One small, older RCT (n=106) reported that radical prostatectomy was associated with reduced disease recurrence compared with EBRT (14% versus 39%, p=0.04).
The addition of neoadjuvant androgen deprivation therapy to radical prostatectomy was not associated with any improvement in survival or cancer recurrence rates, but was associated with an increase in adverse events (3 RCTs).
None of the EBRT regimens appeared to be better than any other at reducing overall or disease-specific mortality (5 RCTs).
Androgen deprivation treatment plus EBRT may reduce overall and disease-specific mortality, but it increased adverse events compared with EBRT alone in high-risk patients (3 RCTs).
The non-randomised studies were predominantly case series. Definitions of adverse events varied considerably. It was not possible to compare treatments in terms of their efficacy or adverse events.
Many other results were also reported.