Accuracy of Risk Assessment Tools To Predict EPE in Patients 45 Years Old or Younger with Prostate Cancer Treated by Radical Prostatectomy
Sara M Falzarano, Esteban Walker, Karen Streator Smith, Eric A Klein, Ming Zhou, Cristina Magi-Galluzzi. Cleveland Clinic, Cleveland; Cleveland Clinic, Cleveland, OH
Background: Controversy exists regarding the outcome associated with a prostate cancer (PCA) diagnosis in young patients. We aimed to evaluate existing calculators for prediction of extraprostatic extension (EPE) in men 45 years old or younger who underwent radical prostatectomy (RP) at our institution.
Design: We queried our RP database for all men diagnosed with PCA from 2000 to 2011 and selected patients 45 years old or younger. Preoperative and postoperative characteristics of the patients were collected in an IRB approved database. The association between D'Amico risk group stratification and presence of EPE was explored; the Memorial Sloan-Kettering Cancer Center (MSKCC) Pre-treatment Prostate Cancer Nomogram (http://nomograms.mskcc.org/Prostate/PreTreatment.aspx) and the CaP Calculator (http://www.capcalculator.org/) that includes Roach formula, Gancarczyk prediction model, and 2007 Partin tables, were used to predict the risk of EPE based on the preoperative features. Predicted values were compared to the real incidence of EPE.
Results: One hundred and twenty-two patients aged 45 years or younger were included. Patients' median age and preoperative PSA were 44 years (range 36-45), and 4.18 ng/ml (range 0.70-90.80), respectively. Eighty-nine (74%) men met D'Amico criteria for low-risk, 26 (22%) for intermediate-risk and 5 (4%) for high-risk disease. Ten (11%) low-risk patients, 8 (31%) intermediate-risk, and 4 (80%) high-risk had EPE at RP. D'Amico criteria showed a strong association with the presence of EPE at RP (p=0.0006). The areas under the curve (AUC) for the CaP calculator tools and the MSKCC nomogram were 0.82, 0.80, 0.86, and 0.82 for the Roach formula, Gancarczyk prediction model, 2007 Partin tables and the MSKCC nomogram, respectively. The cutoffs at which the sensitivity and specificity were optimized were 39% for the Roach formula, 30% for Gancarczyk prediction model, 23% for the 2007 Partin tables and 11% for the MSKCC nomogram.
Conclusions: All the risk assessment tools examined had high discrimination power. However, the predicted risks were, in general, low with optimized cutoffs for detection ranging from 11% to 39%. The lack of calibration could limit their use in younger populations.
Category: Genitourinary (including renal tumors)
Wednesday, March 21, 2012 1:00 PM
Poster Session VI # 162, Wednesday Afternoon