- Prostate MRI currently lacks the accuracy to replace biopsies in active surveillance for prostate cancer.
- The negative predictive value of MRI remains below 90%, the general goal for safely avoiding biopsy.
- Routine confirmatory biopsy should continue in active surveillance until MRI accuracy improves.
Despite technologic advances and guideline recommendations, prostate MRI still lacks the accuracy to replace biopsies in active surveillance, data on almost 2,000 patients suggested.
MRI had a 75% negative predictive value (NPV) for predicting grade group 2 or higher disease at confirmatory biopsy and 77% at surveillance biopsy. Among patients with grade group 1 at diagnostic biopsy, MRI had an NPV of 79% for both confirmatory and surveillance biopsies. NPV was lower for patients with grade group 2 at diagnostic biopsy, for Black patients across all subgroups, and for ability to reclassify cancers as high or low risk by Prostate Imaging Reporting and Data System (PI-RADS).
Until improvements under development can be broadly implemented, routine confirmatory testing in active surveillance for prostate cancer should continue to include biopsy, reported Matthew B. Cooperberg, MD, of the University of California San Francisco, and colleagues in JAMA Oncology.
The level of accuracy that would allow MRI without biopsy during follow-up remains open for discussion.
"This is a matter of opinion and will depend a bit on specific clinical situations and patient preferences -- but generally speaking if we're going to use a test to rule out further investigation, we will usually look for an NPV at or above 90%," Cooperberg told MedPage Today.
"There are lots of potential improvements in the MRI sequence, for example augmenting a bp [biparametric] MRI exam with restricted spectrum imaging," he added. "But the problem in general seems to be less about MRI itself as it is about the inconsistent application and interpretation of PI-RADS scoring. In this arena AI [artificial intelligence] is moving very quickly, with a number of commercial applications and perhaps most notably the PI-CAI open-source initiative which is already outperforming most radiologists (for free) at version 1."
A randomized trial would be ideal for obtaining a definitive answer, Cooperberg acknowledged, but such a trial would "take many years" to show a difference in clinically meaningful outcomes.
"It will likely be more efficient to continue to develop new imaging tools under a rigorous biomarker framework, like REMARK," said Cooperberg. "For now, that means biopsy still has to be the gold standard to evaluate the tool's accuracy, but my bet is this space will evolve substantially in a very finite number of years."
Active surveillance has evolved into the preferred management strategy for favorable-risk early prostate cancer, but the strategy includes confirmatory biopsy to address undersampling at diagnosis. Guidelines from the National Comprehensive Cancer Network and the VA National Oncology Program suggest that MRI may replace confirmatory biopsy, but limited evidence exists to support the accuracy. For example, one single-center study showed that at confirmatory biopsy MRI had a NPV of 74% for grade group 2 or higher, decreasing to 57% in patients with a PSA density (PSAD) >0.15 ng/mL2.
To test the findings from the single-center study, Cooperberg and colleagues examined records for 1,901 U.S. veterans with grade group 1 to 2 at diagnostic biopsy from 2013 to 2023. All the patients underwent MRI (with PI-RADS score) within 180 days prior to confirmatory and/or subsequent surveillance biopsy. The primary outcome was the NPV of MRI (PI-RADS ≤2) on confirmatory or surveillance biopsy. PI-RADS ≥3 was considered positive.
The proportion of pre-biopsy MRI scans increased from 0.7% to 25.9% over the study period. The results showed an NPV <80% for MRI before confirmatory or surveillance biopsy. Men with grade group 1 disease accounted for 80% of the study population, and NPV was only slightly higher than for men with grade group 2 disease but still below 80%.
For men with grade group 2 disease at diagnostic biopsy, MRI's NPV dipped to 44%. Analysis by PSAD showed an NPV of 70% overall for men with PSAD ≥0.15 ng/mL2 and 80% for PSAD <0.15 ng/mL2.
A sensitivity analysis that considered PI-RADS 1 to 3 as negative and PI-RADS 4 to 5 as positive produced an NPV of 68% at confirmatory biopsy and 72% at surveillance biopsy, even though the number of negative MRI results increased over time. When the goal was to rule out grade group 3 or higher, multiparametric MRI had an NPV >95% in many subgroups when considering PI-RADS 3 to 5 to be positive and was still greater than 90% when defining positive as PI-RADS 4 to 5. Positive predictive value, however, dropped to 20% to 30% under those conditions.
"While quality multiparametric MRI clearly can help guide prostate biopsies and improve diagnosis, our data confirm that relying on MRI as a surrogate for biopsy in AS [active surveillance] would result in underdiagnosis of clinically significant cancers -- though we acknowledge few of these would likely be imminently life threatening," the authors concluded.
Disclosures
The study was supported by the Department of Veterans Affairs.
Cooperberg reported relationships with Bayer, Janssen, Astellas, AstraZeneca, Veracyte, Exosome Dx, Tempus, Merck, LynxDx, Pfizer, ConcertAI, and Verana Health. Co-authors reported multiple relationships with industry.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.