Androgen suppression therapy and radiotherapy are used to treat locally advanced prostate cancer. 3 years of androgen suppression confers a small survival benefit compared with 6 months of therapy in this setting, but is associated with more toxic effects. Early identification of men in whom radiotherapy and 6 months of androgen suppression is insufficient for cure is important. Thus, we assessed whether prostate-specific antigen (PSA) values can act as an early surrogate for prostate cancer-specific mortality (PCSM).

We systematically reviewed randomised controlled trials that showed improved overall and prostate cancer-specific survival with radiotherapy and 6 months of androgen suppression compared with radiotherapy alone and measured lowest PSA concentrations (PSA nadir) and those immediately after treatment (PSA end). We assessed a cohort of 734 men with localised or locally advanced prostate cancer from two eligible trials in the USA and Australasia that randomly allocated participants between Feb 2, 1996, and Dec 27, 2001. We used Prentice criteria to assess whether reported PSA nadir or PSA end concentrations of more than 0·5 ng/mL were surrogates for PCSM.

Men treated with radiotherapy and 6 months of androgen suppression in both trials were significantly less likely to have PSA end and PSA nadir values of more than 0·5 ng/mL than were those treated with radiotherapy alone (p<0·0001). Presence of candidate surrogates (ie, PSA end and PSA nadir values >0·5 ng/mL) alone and when assessed in conjunction with the randomised treatment group increased risk of PCSM in the US trial (PSA nadir p=0·0016; PSA end p=0·017) and Australasian trial (PSA nadir p<0·0001; PSA end p=0·0012). In both trials, the randomised treatment group was no longer associated with PCSM (p ≥ 0·20) when the candidate surrogates were included in the model. Therefore, both PSA metrics satisfied Prentice criteria for surrogacy.

After radiotherapy and 6 months of androgen suppression, men with PSA end values exceeding 0·5 ng/mL should be considered for long-term androgen suppression and those with localised or locally advanced prostate cancer with PSA nadir values exceeding 0·5 ng/mL should be considered for inclusion in randomised trials investigating the use of drugs that have extended survival in castration-resistant metastatic prostate cancer.

None.