Dr. Matthew Cooperberg is an Associate Professor of Urology and the Helen Diller Family Chair in Urology at the University of California, San Francisco. He is keenly interested in risk-stratifying prostate cancer to better match treatments to those most likely to benefit.
Prostatepedia spoke with Dr. Cooperberg recently about how advances in imaging have impacted how we diagnose and stage prostate cancer.
How have advances in imaging improved our ability to accurately diagnose and stage prostate cancer?
Dr. Cooperberg: Imaging plays an emerging role. We have to be careful in this field—and specifically in this place in history—not to confuse advances in research with advances in clinical care. There are lots of exciting things going on with imaging research in prostate cancer that are presented to patients as being ready for prime time, but they are not ready yet in most community settings.
Two major imaging stories are evolving right now. One is better local assessment of the prostate itself, mostly based on MRI. The other is better staging with advanced imaging modalities centering on PET/CT, specifically prostate-specific membrane antigen (PSMA) PET/CT imaging. MRI imaging of the prostate has been a little bit of a challenge. Prostate cancers are notoriously not visible on CT scan and are only marginally visible on the historical MRIs that would be done in single phase, primarily to image the lymph nodes. Ultrasound, which is what we use to guide prostate biopsies, can identify a reasonably high proportion of prostate cancers, especially high-risk prostate cancers. However, doing ultrasound well takes a lot of experience and expertise. There are urologists who use ultrasound to identify the different regions of the prostate and don’t spend that much time looking for the cancers. With practice, it is possible to see at least a significant proportion of cancers with ultrasound.
In the last few years, the goal with MRI is to do a better job with local assessment of prostate cancers. Can we see the cancers? Can we accurately stage them using a multiparametric MRI exam? That means looking at T2-weighted imaging, which is essentially anatomic imaging. We look at diffusion weighting, which is intended to give an indication of cellular density. And we look at dynamic contrast, which gives us a sense of how much angiogenesis is going on. That combination of imaging modalities all within a given MRI exam definitely gives us a lot more information than before.
The goal is to see the tumor within the prostate and to get a measure of its stage. However, this all depends on accuracy, which remains highly variable. The UK has been a major proponent of MRI for a long time. Most reports suggest that there is improved accuracy when you do an MRI biopsy versus an ultrasound-guided biopsy. This means that we miss fewer high-grade cancers. The problem with a traditional biopsy has always been both under-sampling and over-sampling.
You find small Gleason 3+3 prostate cancers that you do not need to find, and there is a chance of missing higher-grade cancers. With MRI, both the over-diagnosis problem and the under-diagnosis problem tend to improve. Most agree that an MRI can identify cancers that would be missed on the ultrasound. Whether MRI can replace the standard mapped-out biopsy is more controversial.
In places like the UK, they are on the cusp of implementing a policy in which patients with an elevated PSA get an MRI-targeted biopsy if the MRI shows something. If the MRI doesn’t show anything, you don’t get a biopsy. But I don’t think the world is nearly ready for that. Even in the best series, there is quite a substantial risk of under-sampling with MRI-targeted biopsy alone. It is common to find higher-grade cancer with the non-MRI-guided biopsy.
The bigger problem, moreover, which is barely discussed, is with observer variability in reading the multiparametric MRI. A CT scan of the kidneys, for example, is a very consistent, straightforward test. Pretty much any radiology center can push “start” on the CT scan and generate similar-looking pictures. Any half-decent radiologist should be able to read the CT scan of the kidneys.
A multiparametric MRI of the prostate is a completely different story. There’s a lot of subtlety and a lot of expertise required in programming the machine, protocoling the exam, and interpreting the results.
We see patients who had a prostate MRI at an outside radiology center that is completely unreadable: it hasn’t been performed correctly, let alone interpreted correctly. Even if it is done right, there is an observer variability problem. There have been studies in places like the National Cancer Institute (NCI), who are major experts at MRI, but even at the NCI, there is major interobserver variation because there are shades of gray in terms of how we identify these lesions. (Observer variability is the failure to read the test accurately; interobserver variability refers to two or more clinicians interpreting different results.)
Would you say that MRI-guided biopsy plus the traditional biopsy would be a better approach?
Dr. Cooperberg: Because of the interobserver variability problem, I don’t think we should be anywhere close to getting rid of the traditional biopsy—not even in a center of excellence like NCI, and certainly not in a community radiology setting. Also, the sensitivity of MRI for high-grade cancer is still not as high as we’d like it to be.
There are lots of other questions. How do you do an MRI-targeted biopsy? Do you do what is called cognitive fusion, meaning you just review the MRI to guide your ultrasound-guided biopsy? Do you use fusion systems where we overlay the MRI images onto the ultrasound picture at the time of biopsy? Or are we doing an in-bore MRI-guided biopsy where the biopsy is done under direct MRI guidance in real time? These latter options increase complexity and cost. There is minimal evidence that one is particularly better than another.
Plus, the better you are at identifying lesions by ultrasound, the less commonly you are going to miss something on ultrasound that you would have seen on MRI. We do a lot of MRI at UCSF. We’re a big MRI center, and I order MRIs all the time because we have subspecialty radiology expertise available. A community radiologist who looks at one prostate MRI a month is not going interpret as well as a radiologist who reads them frequently.