The Utility Of Magnetic Resonance Imaging And Spectroscopy For Predicting Insignificant Prostate Cancer: An Initial Analysis

Up to one-half of patients undergoing active surveillance for prostate cancer (CaP) will convert to active treatment. Indeed, many in this subgroup actually do not have insignificant CaP (defined as organ confined, <0.5cm in total volume, and no Gleason pattern 4 or 5). While rising PSA levels or pathologic up-grading on serial biopsies will identify these patients, Dr. Shuka-Dave and colleagues at Memorial Sloan-Kettering Cancer Center developed a nomogram incorporating MRI and MR spectroscopic imaging (MRSI) and compared it to other nomograms. Their report in the online version of the BJU International suggests that the new MR models perform better than standard clinical models to detect clinically insignificant CaP.

A total of 220 patients with CaP had MRI/MRSI prior to radical prostatectomy performed between 2000 and 2004. Inclusion criteria included a PSA level <20ng/ml, clinical stage T1c or T2a, <50% of biopsy cores positive and Gleason pattern 1-3 only. MRI was performed using an expandable endorectal coil with T1- and T2-weighted images. Images were evaluated for the presence of CaP as the group has previously reported. The probability of CaP on MRI was classified on a scale of 0-3; 0 definitely insignificant CaP, 1 probably insignificant CaP, 2 indeterminate, 3 definitely significant CaP. This helped to account for confounding factors such as hemorrhage secondary to biopsy or prostatitis. The nomogram models incorporating the MR data was compared to two clinical models; one a base model using clinical variables and the other a medium model adding percentage of biopsy cores positive and prostate volume.

In the cohort, 129 patients were found on radical prostatectomy to have significant CaP; 71 because the tumor was >0.5cm, 16 due to a Gleason score in the RP specimen >6 and 42 due to both. Both the MR and MR/MRSI models were more accurate than the base clinical model (p<0.001) and medium model (p<0.018) for identifying insignificant CaP. The MRI model identified 30 or 32 men with an MRI score of 0 or 1. Most of the misclassified patients were due to Gleason grade. Also, 57 of 67 patients with an MRI score of 3 had significant CaP. In the 42 patients who were assigned to the indeterminate group, the addition of MRSI to MR correctly reclassified 22 of 24 from indeterminate to insignificant.

This new model might be useful in predicting the probability of insignificant CaP in tumors of <0.5cm3 from those >0.5cm3. The unique expertise of the radiologists in this study would limit immediate translation of these models to most other institutions.

Full Story >>