Influence of Varying Assessment Parameters on the Diagnostic Accuracy of Magnetic Resonance Imaging in the Local Staging of Prostate Cancer.

Introduction: There is a broad variability in the accuracy levels of MRI with regard to the local staging of prostate cancer (PCa). Methods: A prospective analysis was conducted in patients with localized PCa with MRI of the prostate before radical prostatectomy. MRI and pathology findings were independently reviewed and reported based on a standardized map of the prostate with 16 regions of interest (ROIs). Diagnostic accuracy analysis of the MRI was performed using varying prostate-subpart sizes and varying cutoffs for the radiological probability for PCa presence. Results: Seventy four patients were included. Using varying cutoff probabilities and varying sizes of prostate-subparts resulted in a broad range of sensitivity (6-88%) and specificity (38-100%). Lower probabilities of PCa presence and larger prostate-subparts resulted in higher sensitivity but lower specificity and vice versa. Best diagnostic performance was achieved by using prostate sextants and at least moderate probabilities for PCa presence; mean sensitivity and specificity were 38% (95% CI 13-75) and 95% (95% CI 88-98). Conclusion: The use of varying assessment parameters strongly affects the diagnostic accuracy of MRI in the local staging of PCa. Hence, precise and standardized reporting regarding these parameters is important. In our study, using at least moderate probabilities for PCa presence on MRI and prostatic sextants as ROI size was associated with best diagnostic performance. (C) 2015 S. Karger AG, Basel