The prognostic value of different forms of prostate specific antigen and their ratios in patients with prostate cancer

Objective To assess the prognostic value for patient survival of different forms of PSA and ratios thereof, before treatment for prostate cancer, by c onsidering the forms and ratios both as independent markers and by comparin g them with other commonly used prognostic markers, e.g. tumour grade, loca l stage (T-stage) and absence or presence of skeletal metastases (M-stage). Patients and methods Blood samples were collected consecutively from men di agnosed with prostate cancer at our department in 1988. From this group, 66 men were followed until death, or for greater than or equal to 9 years. Tw enty-five patients died from their prostate cancer and 21 from other causes during the follow-up period. Forty-eight patients received hormonal treatm ent, whereas 18 remained untreated or received no treatment for their cance r before they died from other causes. Assays measuring the serum levels of free prostate specific antigen (fPSA), PSA complexed to alpha 1-antichymotr ypsin (PSA-ACT), and total PSA (tPSA) were used to calculate the percentage of free to total PSA (f/tPSA) fPSA/ACT and ACT/tPSA at diagnosis. Based on the initial levels or ratios of the PSA forms, the patients were divided i nto three numerically comparable groups (tertiles) for survival analysis. P rognostic factors predicting patient survival were evaluated using univaria te (Kaplan-Meier life-tables with the log-rank test) and multivariate techn iques (Cox proportional hazards regression model). Results Univariate analysis using the log-rank test showed that the serum l evel of each molecular form of PSA, i.e. tPSA (P=0.001), PSA-ACT (P < 0.001 ) and fPSA (P < 0.001), as well as grade (P < 0.001), T-stage (P=0.00355) a nd M-stage (P < 0.001), provided statistically significant prognostic infor mation. Log-rank tests showed that none of the ratios, i.e. f/tPSA, fPSA/AC T and ACT/tPSA, were informative of prognosis (P > 0.05). However, in a mul tivariate analysis regression model, not only M-stage (risk ratio 4.2; P=0. 026) and grade (risk ratio 2.6; P=0.022), but also f/tPSA (risk ratio 1.8; P=0.037), provided significant prognostic information. Conclusion The values of tPSA, fPSA and PSA-ACT, as well as grade and T- an d M-stage, are all independent prognostic factors of prostate cancer surviv al. In a multivariate analysis, not only M-stage and grade but also f/tPSA provided significant prognostic information.