THE PROGNOSTIC VALUE OF PSA LEVELS IN RADIATION-THERAPY OF PATIENTS WITH CARCINOMA OF THE PROSTATE - THE UCLA EXPERIENCE 1988-1992

Background and objectives: Pretreatment prostate-specific antigen (PSA ) levels may be of prognostic significance for patients with prostate cancer. Posttreatment PSA data are more limited. This study was undert aken to examine the prognostic role of pretreatment and posttreatment PSA levels in the radiation treatment of patients with carcinoma of th e prostate. Methods: One hundred one patients who received primary rad iation therapy at UCLA between 1988 and 1992 for clinical stage A to D 1 prostate cancer were analyzed. Included were 4 patients with stage A , 77 with stage B, 16 with stage C, and 4 with stage D. All patients h ad pretherapy and posttherapy PSA values. Patients received definitive radiation therapy with photons (81), neutrons (13), or interstitial i mplant (7). Correlations were made with other prognostic factors and t reatment outcome. Results: Median follow-up was 28 months. At last fol low-up, 64% were without evidence of disease, 17% had rising PSA profi les or failure of PSA to normalize (chemical failure), and 19% had loc al recurrence and/or distant metastases. The 4-year overall survival w as 85%, whereas actuarial survival free of chemical or clinical failur e was only 32%. Pretreatment PSA levels and posttreatment PSA level no rmalization at 6 months correlated significantly with disease-free sur vival. On univariate analysis, pretreatment PSA levels correlated sign ificantly with stage, high versus low Gleason score, and outcome. Post treatment PSA level normalization at 6 and 12 months correlated with s tage, pretreatment PSA level, and outcome, but not with Gleason score. Only PSA level normalization at 6 months and age were independent var iables using multivariate analysis. PSA nadir values differed signific antly between patients free of disease and those who failed. Conclusio ns: In our analysis, posttreatment PSA levels were independently predi ctive of outcome, whereas pretreatment PSA levels, while correlating w ith other prognostic factors, were not independently predictive. Given the prognostic value of posttreatment PSA levels, it is appropriate t hat chemical failures be included in outcome analyses, although this w ill lower disease-free survival.