The ability of prostate-specific antigen (PSA) density to predict an upgrade in Gleason score between initial prostate biopsy and prostatectomy diminishes with increasing tumour grade due to reduced PSA secretion per unit tumour volume

OBJECTIVES To analyse the performance of PSA density (PSAD) as a predictor of Gleason score upgrade in a large cohort stratified by Gleason score. We and others have shown that an upgrade in Gleason score between initial prostate biopsy and final radical prostatectomy (RP) pathology is a significant risk factor for recurrence after local therapy. PATIENTS AND METHODS Patients undergoing RP with matching biopsy information were identified from two prospective databases. Patients were analysed according to the concordance between biopsy and final pathology Gleason score in three paired groups: 6/>6, 3 + 4/>3 + 4, 7/>7. Receiver-operating characteristic (ROC) curves were generated stratified by Gleason score, and the area under the curve (AUC) calculated. Logistic regression models were fitted to identify significant predictors of tumour upgrade. RESULTS From 1516 patients, 435 (29%) had an upgrade in Gleason score. ROC analysis showed a decline in AUC with increasing biopsy Gleason score, from 0.64 for biopsy Gleason score 6, to 0.57 for Gleason score 7. In logistic regression models containing pretreatment variables, e. g. clinical stage and number of positive cores, for Gleason score 6 and 3 + 4, PSAD was the strongest predictor of subsequent tumour upgrade (odds ratio [OR] 1.46, 95% confidence interval [95% CI] 1.18-1.83, P = 0.001 and OR 1.37, 95% CI 1.14-1.67, P = 0.002, respectively). Surprisingly, in tumours upgraded from Gleason score 7 to >7, PSAD was not predictive even on univariable analysis, whereas clinical stage and number of positive cores were significant independent predictors. To explore the relationship between serum PSA and Gleason score, tumour volume was calculated in 669 patients. There was a strong association between Gleason score and tumour volume, with the median volume of Gleason score 7 and Gleason score >7 tumours being approximately twice and four-times that of Gleason score 6 tumours, respectively (P < 0.001). In contrast, the median serum PSA level per millilitre tumour volume decreased significantly with increasing grade, from 5.4 ng/mL for Gleason score 6 to 2.1 ng/mL for >7 ( P < 0.001). CONCLUSIONS There is a strong correlation between Gleason score and tumour volume in well/intermediate differentiated tumours, and as they produce relatively high amounts of PSA per unit volume of cancer, high PSAD is the strongest single predictor of tumour undergrading. However, as higher grade tumours produce less PSA per unit volume, PSAD loses its predictive ability, and other clinical markers of tumour volume such as palpable disease and numbers of positive cores become more predictive.