Negative Predictive Value of Multiparametric Magnetic Resonance Imaging in the Detection of Clinically Significant Prostate Cancer in the Prostate Imaging Reporting and Data System Era: A Systematic Review and Meta-analysis.
Sathianathen NJ., Omer A., Harriss E., Davies L., Kasivisvanathan V., Punwani S., Moore CM., Kastner C., Barrett T., Den Bergh RV., Eddy BA., Gleeson F., Macpherson R., Bryant RJ., Catto JWF., Murphy DG., Hamdy FC., Ahmed HU., Lamb AD.
CONTEXT: Prebiopsy multiparametric magnetic resonance imaging (mpMRI) is increasingly used in prostate cancer diagnosis. The reported negative predictive value (NPV) of mpMRI is used by some clinicians to aid in decision making about whether or not to proceed to biopsy. OBJECTIVE: We aim to perform a contemporary systematic review that reflects the latest literature on optimal mpMRI techniques and scoring systems to update the NPV of mpMRI for clinically significant prostate cancer (csPCa). EVIDENCE ACQUISITION: We conducted a systematic literature search and included studies from 2016 to September 4, 2019, which assessed the NPV of mpMRI for csPCa, using biopsy or clinical follow-up as the reference standard. To ensure that studies included in this analysis reflect contemporary practice, we only included studies in which mpMRI findings were interpreted according to the Prostate Imaging Reporting and Data System (PIRADS) or similar Likert grading system. We define negative mpMRI as either (1) PIRADS/Likert 1-2 or (2) PIRADS/Likert 1-3; csPCa was defined as either (1) Gleason grade group ≥2 or (2) Gleason grade group ≥3. We calculated NPV separately for each combination of negative mpMRI and csPCa. EVIDENCE SYNTHESIS: A total of 42 studies with 7321 patients met our inclusion criteria and were included for analysis. Using definition (1) for negative mpMRI and csPCa, the pooled NPV for biopsy-naïve men was 90.8% (95% confidence interval [CI] 88.1-93.1%). When defining csPCa using definition (2), the NPV for csPCa was 97.1% (95% CI 94.9-98.7%). Calculation of the pooled NPV using definition (2) for negative mpMRI and definition (1) for csPCa yielded the following: 86.8% (95% CI 80.1-92.4%). Using definition (2) for both negative mpMRI and csPCa, the pooled NPV from two studies was 96.1% (95% CI 93.4-98.2%). CONCLUSIONS: Multiparametric MRI of the prostate is generally an accurate test for ruling out csPCa. However, we observed heterogeneity in the NPV estimates, and local institutional data should form the basis of decision making if available. PATIENT SUMMARY: The negative predictive values should assist in decision making for clinicians considering not proceeding to biopsy in men with elevated age-specific prostate-specific antigen and multiparametric magnetic resonance imaging reported as negative (or equivocal) on Prostate Imaging Reporting and Data System/Likert scoring. Some 7-10% of men, depending on the setting, will miss a diagnosis of clinically significant cancer if they do not proceed to biopsy. Given the institutional variation in results, it is of upmost importance to base decision making on local data if available.