Adding multiparametric magnetic resonance imaging (MRI) findings to conventional clinical predictors in a prostate cancer risk stratification model increases diagnostic accuracy and may reduce the number of unnecessary biopsies while maintaining a high rate of diagnosis of clinically significant prostate tumors, according to a new study.
In a validation cohort, the area under the curve (AUC) increased from 64% with a baseline model to 84% with a model that incorporated multiparametric MRI-derived prostate volume and PI-RADSv2 category as well as the clinical predictors in the baseline model (age, race, prior biopsy findings, results of a digital rectal examination [DRE] and PSA level).
At a risk threshold of 20%, the MRI-based model had a lower false-positive rate than the baseline model (46% vs 92%), with only a small reduction in the true-positive rate (89% vs 99%), a team led by by Baris Turkbey, MD, of the National Cancer Institute (NCI) in Bethesda, Maryland, reported online ahead of print in JAMA Oncology. At a 20% risk cutoff, 38% of biopsies could have been avoided with the MRI model compared with 6% of biopsies avoided by the baseline model.
“The net reduction in the number of false-positives based on the MRI model, compared with having to perform a biopsy in all patients with positive MRI results, was equivalent to performing 18 fewer unnecessary biopsies per 100 men, with no increase in the number of clinically significant prostate cancer left undiagnosed,” the investigators stated.
Dr Turkbey and his colleagues created the MRI-based model based on the findings from a development cohort of 400 patients enrolled at NCI and a validation cohort of 251 patients enrolled at the University of Chicago Medical Center and the University of Alabama at Birmingham. All participants underwent MRI, MRI-transrectal ultrasound (MRI-TRUS)-guided prostate biopsy, and 12-core systematic biopsy.
The development cohort included patients with elevated PSA levels or abnormal findings on a digital rectal examination and at least 1 lesion detected on multiparametric MRI scans. All detected lesions were evaluated and assigned a category based on the PI-RADSv2 guideline. Patients with category 3 or higher lesions routinely underwent MRI-TRUS fusion-guided biopsy, whereas those with category 1 or 2 lesions were targeted only under certain circumstance or based on patient preference. Only the category of the index lesion was considered in this study and was defined by gthe highest PI-RADSv2 category in the prostate.
For the validation cohort, investigators used the same enrollment criteria as used for the development cohort. All patients underwent multiparametric MRI, and lesions were assigned PI-RADSv2 categories. Investigators also applied the same definitions for index lesions and biopsy decision rules.
Mehralivand S, Shih JH, Rais-Bahrami S, et al. A magnetic resonance imaging-based prediction model for prostate biopsy risk stratification. JAMA Oncol 2018; doi 10.1001/jamaoncol.2017.5667