MRI Versus Ultrasound for Prostate Cancer - Samir Taneja

July 26, 2020

Samir Taneja joins a panel of urologists to debate the use of ultrasound when compared to magnetic resonance imaging (MRI) as the best diagnostic and guidance imaging option for prostate cancer patients. Prostate ultrasound, specifically, micro-ultrasound, has a high sensitivity for predicting clinically significant cancer however its diagnostic accuracy is dependent on the urologist performing the ultrasound as well as interpreting the results. Prostate MRI utilization has recently increased globally predominantly for biopsy guidance but also for significant cancer detection especially prediction of high-grade cancer. Pre- biopsy MRI can be a useful tool in selecting patients for biopsy and in guiding clinicians to therapeutic decisions post-biopsy.


Samir S. Taneja, MD, The James S. and Janet Riha Neissa Professor of Urologic Oncology, Professor of Urology, Radiology and Biomedical Engineering, Director, Division of Urologic Oncology, GU Program Leader, NYU Perlmutter Cancer Center, New York, New York, USA.

Read the Full Video Transcript

Samir Taneja: So I'm going to talk to you about MRI versus ultrasound. I think Neal Shore asked me to talk about this because of some emerging new technologies in ultrasound. These are a number of disclosures. None are really related to this talk with exception of the fact that recently I've been helping Exact Imaging the micro-ultrasound company to perhaps design a clinical trial.

So the goal of imaging now in the pre-biopsy space has evolved a bit. It's largely diagnostic. We hope to use it to help us identify clinically significant cancer and avoid indolent, and that should really be the goal of any biopsy. We like to use it now to map cancer and that plays into the focal therapy discussion, knowing where the clinically significant cancer resides is becoming more and more important. We, of course, use it to guide the biopsy but most interestingly, in recent years, we've been interested in knowing if imaging can serve as a biomarker and help us to assess risk in the patient who comes to us with an elevated PSA.

Prostate MRI in 2019 has grown tremendously in global utilization. In the United States, it's predominantly used for guiding a biopsy and less used for the purposes of risk stratification, which is not the case in Europe. Its quality varies widely, particularly inter-observer variability between readers and that's been the main kryptonite, not kryptonite, but main kink in the armor of the MRI and cost is, of course, a concern as well. What do we know about prostate MRI? Well, it helps us to detect more high-grade cancer and less low-grade. It allows prediction of the likelihood of high-grade cancer on a biopsy and the PI-RAD score, the suspicion score seems to correlate with outcomes on surveillance and surgery. This allows us to maintain clinically significant cancer detection, avoid indolent disease detection, individualize our biopsy approaches if we adopt it as a biomarker and ultimately help to guide our therapeutic decision making after the biopsy.

The critical study that's been done with MRI in the pre-biopsy setting is the PRECISION trial. This was a randomization of 500 men to undergo either prostate MRI or TRUS biopsy. The men who underwent prostate MRI then were not biopsied if they had a low suspicion MRI. The ones who were biopsied were those with PI-RAD scores of three to five, and among those about 40% had clinically significant cancer. In comparison, men who went straight to TRUS biopsy had clinically significant cancer 28% of the time. So the major findings of the study, 30% of biopsies could be avoided, more clinically significant cancer was detected and less indolent cancer was detected with the MRI targeted approach. The aftermath of PRECISION has very variable. In Europe it's been broadly accepted and in fact a growing number of regulatory bodies have endorsed pre-biopsy MRI both for risk stratification and for guidance of biopsies, some even requiring it before a biopsy can be done. The study has been heavily criticized in the United States due to failure to demonstrate the rate of cancers missed in the men who were not biopsied and in the abandoned systematic sample in the MRI-targeted group. And as such, pre-biopsy MRI has been largely rejected in the US as a tool for risk assessment and is widely utilized for the guidance of the biopsy.

The MRI-FIRST study addressed one of the concerns about PRECISION. This was a study in which men underwent both systematic and MRI-targeted biopsies for suspicion scores of two to five. They also had a biopsy if their suspicion score was one. The systematic biopsy and targeted biopsy in this study were comparable in their detection rate and when you added them together you found clinically significant cancer more often. So you see in the red box that the added value of a systematic biopsy was 5.2% in picking up more clinically significant cancer. So their conclusion was both are required to maximize clinically significant detection. But if you look at the footnotes, in fact by doing that you greatly escalate that the detection of indolent cancers. And so that's the trade-off. That if you were to do systematic biopsies in addition to the target, you quadruple the rate of indolent cancer detection and that perhaps defeats the purpose of the MRI target.

PROMIS has shown us that if we compare systematic biopsies to template biopsies, which was the intention of the study, the negative predictive value is poor. So a quarter of cancers are theoretically missed by the MRI. We've applied the PRECISION cohort data to our patients and found that if we do that we miss about 19% of clinically significant cancers. That's to say if we don't biopsy PI-RAD one and two, and if we also don't do systematics in the higher PI-RAD scores. If you dissect that most are often low-grade, low-risk and when they are high-grade they're typically small with a small amount of pattern four and often adjacent to the target reflecting targeting error. In fact, when we apply criteria to missed cancers, most are low-risk by either Epstein in the indolent category or CAPRA in the clinically significant category.

What about prostate ultrasound, historically and in 2019? Well, there's a long history that most urologists are aware of. It was described in the late 1980s and it really transformed the method of biopsy from digital to ultrasound-guided, first looking at hypoechoic lesions and then allowing systematic sampling with the inception of PSA in the 90s, and more recently it's been used to guide MRI-targeted biopsies using fusion platforms. More recently, micro-ultrasound has perhaps improved the resolution and detection characteristics of the ultrasound itself. It has a relatively high sensitivity for predicting clinically significant cancer, its diagnostic accuracy is highly operator dependent among urologists. It's equipment dependent. There's no standardized methodology. So the problem with ultrasound historically has been because urologists weren't comfortable reading it, because it had a lot of false positives, it was really abandoned as a diagnostic test and was largely used to guide systematic samples.

What we've observed in our data set that if we overlay fusion with ultrasound that very often we see hypoechoic lesions at the sight of the target. I created an arbitrary scale for measuring the hypoechoic lesion and compared it to cancer detection. And what we see is that the ultrasound hypoechoic lesion indeed does predict an increasing likelihood of clinically significant cancer. And if you overlap that with PI-RADs in the high PI-RADs categories there are a number of cancers missed by conventional ultrasound, but in the lower PI-RAD categories, the majority of patients who would have clinically significant cancer have a hypoechoic lesion.

Well, what's micro-ultrasound? This is simply the idea that if you increase the frequency of the ultrasound up to initially 21 megahertz and now in the commercial application 29 megahertz, you improve the resolution of the ultrasound. That's been translated by Sangeet Ghai at Toronto into a scale similar to PI-RADs called PRI-MUS™. And when we look at micro-ultrasound, a few observations have been made. It finds more cancer than conventional ultrasound, both overall and clinically significant. When you add it to a systematic biopsy, it detects more clinically significant cancer than the systematic alone. If you're targeting abnormalities like PI-RADs on the right there it correlates with high-grade cancer detection, but one potential concern is that most patients on the left graph are found to have PRI-MUS™ 2 category readings. And if that's the case you may end up biopsying a lot of men to find those cancers. When you compare it with MRI, the synopsis of this slide is basically that of 35 clinically significant cancers that were found, 33 of them were found by the PRI-MUS™ scale. So it seems that it's pretty good and it needs good comparisons with MRI for the future.

So in conclusion I would say that pre-biopsy MRI is useful in selecting men for biopsy, in addition to guiding the biopsy, it misses some clinically significant cancers but misses more indolent cancer. And I think the critical question for the urologists will be what are you willing to give up? In order to overcome the overdetection problem, can we give up the detection of a few of these very small 3+4 cancers? Its role in the pre-biopsy setting relies on that because I don't think it will be cost-effective if we just simply do it to guide biopsies. Ultrasound has historically failed due to poor sensitivity, operator dependence. I think micro-ultrasound offers improved sensitivity. The jury is still out on ease of use and reproducibility but it's potentially less costly than the MRI. Thank you.

Gordon Brown: Excellent talks, thanks everybody for the content. Now it's interesting to see some of the data that's coming out. You're looking at a rapidly changing field both in diagnoses as well as treatment of prostate cancer. You know, Ben, you have any insights or thoughts about the talks?

Ben Lowentritt: Well I kind of wanted to think of something that brings all this discussion together and I'm thinking about one of the dilemmas that we have, because there's just still no real consensus. But in that patient that was diagnosed, let's say with low risk disease for purposes, from a traditional transrectal ultrasound approach, and we're considering they're on surveillance, and you're considering how to confirm that you have the adequate and accurate staging and grading, what is your thoughts on do they need a biopsy? Do they need an MRI? What's the approach to the biopsy? What's the timing? Maybe we can all discuss it because I think still there's a lot of variability there and we're hearing some different options about what might be added on. So I don't know, Dr. Taneja, you want to start?

Samir Taneja: Well I think the problem in assessing risk with MRI is it's not really as well correlated with some of the longterm outcomes. So we know that somebody who has an abnormal high suspicion MRI, they're more likely to progress on surveillance. They're more likely to recur after surgery in general. But that's hard to quantitate. And the one thing that's missing is it's not really been correlated as of yet with metastasis and death. So we use it in our risk assessment. I'm a big believer in MRI for everybody, but maybe that reflects a little bit of a bias. I do realize that as time goes on, that's not going to be effective from a cost perspective. So I like the idea of sequential risk stratification, starting with the simple PSA density, family history, obviously genetic history, I still like doing digital rec... I don't like doing digital rectal exams, but I still like the idea of doing digital rectal exams.

But start with the simple and if you perceive the gentleman to be at a substantive clinical risk then I'll usually go to an MRI first to determine the need for further biopsy, et cetera. We like using low suspicion MRIs to help us decide if someone needs a biopsy. But there still are going to be a lot of guys with low suspicion MRIs who based on the first qualification still need a biopsy, and so we use it selectively to avoid biopsy and people who have marginal indications and in the low pirate setting, then that's where I'll use the biomarkers to further refine the risk if we're not sure if that patient should have a biopsy. That's a costly approach, but if you use that first part, the PSA density, the family history, maybe avoid the MRIs for the people who don't really need it.

Ben Lowentritt: I have just a couple thoughts. I still believe that the most important question nobody can answer on this panel is if you diagnose somebody with a grade group one lesion, if you're certain that's what they have, no one knows if that lesion could evolve into a grade group two or maybe a three. We know it's probably not going to turn to a four or five, but I know from experience that I believe that many of those do evolve into a pattern two, maybe a three over time. So I don't feel that this idea that not knowing about that disease is better off for everyone. I'm not an advocate, so my bias is that we need better tissue. We need to bring the patient through the diagnostic experience. Ideally the least pain, the least complications, with the highest diagnostic yield. MRI I think is a fabulous tool, but its negative predictive value is still not perfect and if we can adequately sample the entire prostate, I think that's obviously where my research has all been focused on.

Matthew Allaway: To address your question, I think there are many great ways of going about it. Everyone does it a little differently. For the garden variety low-risk patient, I say, "Come back at some point in the next six months. On that day you'll have an MRI, you'll have a re-staging biopsy, which is a standard 12 core plus two cores from any MRI PI-RADs 3 to 5 lesion." And the data shows about 20% of them will have a higher Gleason score, and we say surveillance may not be the best idea and 80% get good news, and then we cruise on surveillance.

Neal Shore: Can I ask, so based upon, Samir, what you were saying, your comments about upgrading, even if we find a six, and the data that you presented. So yeah, markers, right? So we don't have a talk on markers today, but you get somebody that comes in and says, "You know, my PSA increased by 1.5 over the last six months." Rechecked, again in six months, maybe it's up a little bit more, sort of in a bit of a gray zone, maybe has sort of a quasi interesting family history. He's 63, 64 years old. What markers are you getting... And now he says, "I don't want a biopsy." DRE's normal. "I don't want a biopsy, what are the markers that you would use now? And are there any trials that you would say for us to keep our eye on in terms of using MR with markers for the future to just avoid biopsy in general?".

Samir Taneja: Well the last part of the question, I think that there are some trials ongoing. The one, in particular, is the PRIORITY trial, which is endorsed by the SUOCTC. That's a trial in men who've had previous negative biopsies, but they're all getting confirmed MDx, SelectMDx and MRI prior to biopsy. I don't know that the MRI is required, but I think they're all getting it. And so when that study is done, it's 800 men we'll have a comparison of MRI, confirm MDx, and SelectMDx and the prediction of cancer in that group. In the biopsy naive population that you're talking about, there's some smaller industry-sponsored trials, but I'm not aware of any EDRN trials that are comparing MRI. I think there's been a reluctance to do these large comparative trials. But I agree with you, that's kind of what we need is cohorts that have gotten all of them so that we can compare the performance of them over time.