UCSF

Prostate cancer is currently the most prevalent noncutaneous cancer in males. An increase in the number of men diagnosed with indolent, organ-confined disease has lead to the increasing numbers of patients and their physicians selecting active surveillance or "delayed definitive treatment" as a viable approach for managing prostate cancer. However, the selection of appropriate patients for active surveillance has been confounded by sampling errors associated with transrectal ultrasound guided biopsies as well as accurate clinical and imaging biomarkers that predict for aggressive, progressive disease at diagnosis. Multiparametric magnetic resonance (MR) imaging may assist in overcoming this problem; it allows for the identification of the whole gland and a combination of T2-weighted MR imaging, proton MR spectroscopic imaging (1H MRSI), and diffusion-weighted imaging (DWI) can be used to better characterize the aggressiveness of intraglandular cancer at diagnosis. In this retrospective study, we quantitatively analyzed multiparametric MR images from a cohort of active surveillance patients (N=119) to determine the combination of MR imaging techniques that best predicted disease progression on active surveillance. Fifty-nine of 119 patients progressed within 43 ± 32 months on active surveillance. Receiver-operator characteristic (ROC) curve analysis indicated that all three techniques (T2 MRI, MRSI, DWI) demonstrated similar modest accuracies in predicting progression on AS (AUC of 0.59, 0.63 and 0.61, respectively). The best prediction of prostate cancer progression in AS patients was when all three techniques were positive for cancer presence, yielding an odds ratio for progression of 2.91 (95% CI 1.19 - 7.08) as compared to all other findings. Whereas a negative finding for all 3 tests for patients that were appropriate for AS yielded an odds ratio for no progression of 2.84 (95% CI = 1.26 - 6.37) as compared to all other findings. In conclusion, multiparametric MR imaging could play a valuable role in better selecting patients for active surveillance.