Predicting non-organ-confined prostate cancer in men diagnosed after 2000

Abstract

The objective of this study was to preoperatively predict non-organ-confined disease in patients considering radical prostatectomy. To account for the stage migration seen in prostate cancer, we included only those patients who underwent prostatectomy after the year 2000. Information on a cohort of 1895 patients who underwent radical prostatectomy from 2000 to 2008 was retrieved from the Duke Prostate Center database. Race (African American, non-African American), body mass index, age at surgery, PSA, biopsy Gleason sum (<7, 7 and >7) and clinical tumor stage (cT1, cT2/3) were analyzed by univariate analysis followed by logistic regression analysis. The Duke Interactive Clinical Equation for staging (DICE-S score) was calculated from the logistic regression model. The model was then internally validated using a bootstrapping technique. Biopsy Gleason sums 7 and >7 were more likely to have non-organ-confined disease compared with <7 (OR=2.97, Gleason sum=7; OR=3.25, Gleason sum>7). Clinical tumor stage, cT2/3, predicted non-organ-confined disease (OR=1.58). Older age was associated with non-organ-confined disease (OR=1.02), as was greater PSA (OR=1.12). DICE-S equation x=ln (p/1-p)=-3.627+0.019 (age)+0.109 (PSA)+1.087 (bGleason=7)+1.180 (bGleason >7)+0.459 (clinical T stage >T1), where p=(e(x))/(1+e(x)). A concordance index (prediction accuracy) of 0.73 was reached on internal validation. Using the DICE-S score, age, PSA, biopsy Gleason sum and clinical tumor stage, we can predict non-organ-confined disease in radical prostatectomy at an acceptable accuracy. Preoperative information on disease stage may aid in treatment decisions and surgical approach.