Prostate-specific antigen cancer volume: A significant prognostic factor in prostate cancer patients at intermediate risk of failing radiotherapy

Purpose: Although the pretreatment serum prostate-specific antigen level (PSAL) is the single-most significant predictor of local and biochemical control in prostate cancer patients treated with radiotherapy, it is relatively insensitive for patients with a PSAL in the intermediate range (4-20 ng/ml). PSA density (PSAD) has been shown to be slightly more predictive of outcome than PSAL for this intermediate risk group; however, this improvement is small and of little use clinically. PSA cancer volume (PSACV), an estimate of cancer volume based on PSA, has recently been described and has been purported to be more significant than PSAL in predicting early biochemical failure after radiotherapy. We report a detailed comparison between this new prognostic factor, PSAL, and PSAD. Methods and Materials: The records of 356 patients treated with definitive external beam radiotherapy for regionally localized (T1-4,Nx,M0) adenocarcinoma of the prostate were reviewed. Each patient had a PSAL, biopsy Gleason score, and pretreatment prostate volume by transrectal ultrasonography. The median PSAL was 9.3 ng/ml and 66% had Gleason scores in the 2-6 range. The median radiation dose was 66.0 Gy and the median follow-up for those living was 27 months. PSACV was calculated using a formula which takes into account PSAL, pretreatment prostate ultrasound volume, and Gleason score. The median PSACV was 1.43 cc. Biochemical failure was defined as increases in two consecutive follow-up PSA levels, one increase by a factor > 1.5, or an absolute increase of > 1 ng/ml. Local failure was defined as a cancer-positive prostate biopsy, obtained for evidence of tumor progression. Results: The distributions of PSACV and PSAL were similar and, when normalized by log transformation, were highly correlated (p < 0.0001, linear regression). There was a statistically significant relationship between PSACV and several potential prognostic factors including PSAL, PSAD, stage, Gleason score, and pretreatment prostatic acid phosphatase (PAP). In univariate analyses, PSACV, PSAL, and PSAD proved to be the most significant predictors of both biochemical and local control. In multivariate analyses using Cox proportional hazards models with PSAL, PSAD, PSACV, and PAP as continuous variables, PSAL, PSACV, and Gleason score were significant in predicting biochemical control. Only PSAL was significantly correlated with local control. However, when these analyses were restricted to patients with intermediate PSALs (4-20 ng/ml), only PSACV was significant for predicting both biochemical and local control. Conclusion: PSACV was highly correlated with actuarial local and biochemical control and was superior to both PSAL and PSAD in predicting these outcomes in patients with PSALs between 4 and 20 ng/ml. (C) 1997 Elsevier Science Inc.