Dr. Stone presented at the 26th International Prostate Cancer Update on Thursday, January 21, 2016 on the “Role of mpMRI: Implication and Implementation.”
Keywords: prostate cancer, MRI, biopsies, target therapies
How to cite: Stone, Nelson N. “Role of mpMRI: Implication and Implementation.” Grand Rounds in Urology. January 21, 2016. Sep 2019. https://grandroundsinurology.com/role-mpmri-implication-implementation.
Role of mpMRI: Implication and Implementation
In your decision to use an MRI, it’s probably one of the more important things to consider is that’s it’s extremely difficult to get consistency in the observers interpreting the MRI both correctly and consistently. There’s no question if you go an institution where they’re doing thousands of these, in their experience they’re going to be more likely to tell you what’s actually something that should be biopsied.
Sending somebody just for an MRI who hasn’t been doing a lot of these is probably not the best idea, just because you want to get the study. That’s the take-home message. Another commonly asked question is can I use a 1.5T versus a 3T, knowing that many of the centers have upgraded their magnets and their software, and the answer is yes.
The only difference between the higher energy magnet is that it’s better at discriminating patients who have prostatitis, and trying to distinguish the prostatitis from prostate cancer, especially if it’s more interiorly. So, the 3T magnet’s better at that, but it doesn’t mean you can’t do it, especially if you’re using a surface coil and you have updated software.
So, what is the multi-parameter MRI? We’re all familiar with the T1 weighted and the T2 images, so these are enhancements. Using the T1 weighted image, which doesn’t give you a very good anatomic picture, we use a contrast, and that’s called DCE, or Dynamic Contrast Enhancement.
The T2 weighted image, which we’re used to looking at because it gives a good anatomy of the prostate, there’s two enhancements, which are the apparent diffusion coefficient, with comes with the diffusion weighted imaging. So, DWI and ADC on the T2, and DCE on the T1. Spectroscopy is also an enhancement that I find that most centers are not using; spectroscopy.
Part of the problem is really no standard way to identify what are suspicious lesions. What it turns out to be is is if you see them, the radiologist sees them on the different sequences, they’re more likely to call the lesion highly suspicious and therefore that’s a target, but if they only see them on one or two of the lesions, one or two of the sequences, they’re less likely to label that lesion highly suspicious.
As we go forward and using regression analysis and discrimination studies, we’re probably going to be able to come up with more statistical mechanisms to evaluate the MRI result, but right now it’s really in the hands of the radiologists, and that’s based on his or her experience.
This is a fairly simple image. You can see the hypointense cancer in the posterior of the prostate there. Nobody would have trouble calling this a prostate cancer. Here is the diffusion weighted imagine. You see the bright white area. Here is the ADC map of that, which looks hypointense, and there is the T2 image for those other sequences.
So, what a radiologist would do, if you’re performing a targeted biopsy, for example, using the Phillips machine, now we’re using the Artemis Hoggin machine. They would have multiple screens up on the radiology suite, and they’d be looking at the T2 weighted image. They would see the DWI, they’d see it on the ADC, and they would circle it so they’d call it segmentation, and that would be the region of interest.
Then it has to appear on the T2, because at the end of the day, they have to get that dichroine electronic information outputted to the ultrasound machine so you can have a target. So, the requirement is to have software. The radiologists need software so they can visualize multiple screens, looking at the different sequences simultaneously, and then co-register everything back to one image, which would be the anatomical image, which is the T2 you see down there in the corner.
This is an example of the dynamic contrast enhancement, and you can see the lesion there. So, this requires an injection, so we have to give the patient a contrast image, and then the image of the lesion on the T1 weighted image. This would be another sequence the radiologist has to look at, and hopefully he or she sees it on this, and on the other images you’re going to have a region of high suspicion.
Lastly here is the spectroscopy, that’s of course choline citrate ratios, and you can see down there. You can see in the lower corner there an area of suspicion and the whole amount showing you that was Gleason 6 lesion. So, if the radiologist also has the spectroscopy, then that will add to the specificity of detecting prostate cancer.
Futterer put together a study using all these different sequences and they found that the area under the curve was 68, 91, and 80%. If you look at a reading from the radiologist and they’re looking at the different sequences, on the body we see the red line and everything was positive, they’re going to tell you that’s high probability, and sure enough, if you target that lesion you will probably find prostate cancer, and it will probably be high-grade.
Here’s a couple examples of the equipment used for the co-registrated target biopsies; on my left there is the Ika machines, and on the right is the Phillips machine.
Now, the problem comes–because things are not so simple–first we got to figure out whether they’re a lesion, and then we’ve got to take the information from the MRI and we’ve got to marry that information to the ultrasound image during the targeted procedure.
That means because a patient was prone, most likely, when the MRI was done, and now he’s on the side, or his legs are up if you’re doing it transperineally, that you’ve got to pull the margins together. One of them needs to be deformed to the other, and when you start doing that, lesions and sides start moving around. It’s always been a problem that if you have smaller lesions or hard, distinguished lesions, whether you’re going to actually a proper overlay and successfully be able to target it. So, that’s the weak point of this system and systems like it.
Here’s an example. You see some pictures of the DCE and T2 image and you see the contour of the prostate, which is now overlaid over the ultrasound image and the targets are overlaid, so the targets are made on the MRI, that information is then co-registered to the ultrasound if you’re doing this in the office, and then you use a targeting system to get to the biopsy. You see the results, and there’s another example of the same.
Now, of course, when you put pictures in the literature you always show the big, giant lesions, because they’re easy to see. You wish life was that way, because most of the time it doesn’t look like this. Here’s what this is telling us, this study, where the MRI showed no target, and PHI, meaning all the criteria are positive. So, when you look at, for example, the PHI group, the number of cases was 15 positive out of 16, so it’s got a very high accuracy rate.
The problem is that only represents 9% of the cases. The overwhelming majority’s in the middle, which is in the gray zone, no surprise.
This is a seminal study that came from Peter Pinto’s group that looked at the comparison of MRI-targeted biopsies with the addition of trans-rectal semi-blind biopsies that we’ve been doing for years. What they found is when you added the extra biopsies, random biopsies, you found more of the high-risk cancers; in fact, you found one-third more high-risk cancers when you added the random systematic biopsies in.
Now, this has really changed the approach, where before if people were at this meeting last year, Mark Emberton was up here talking about their MRI experience, and Mark Emberton said, “If I see that lesion, I can hit it. I don’t need to do more than one or two biopsies.”
That’s dramatically changed now. The average number of biopsies being done for one of these cases is probably 15-20. You’re getting 12 systematic biopsies, and you’re getting at least six targeted biopsies, and each lesion probably gets two biopsies, the targeted lesions, so if you see an average of three lesions, which is typical, you get 6 of those, and 8, 12 of the others, so you’re up to 18 biopsies.
Not that it’s bad, but that’s the way we’re doing it now. This is a review study that just came out in European Urology, and this is probably one of the best studies to read, so you can look it up in the book there, or on the slides. This is the question I was asking, which is what’s the most important statistics? A little hard to see there. That was a negative predictive value.
The way the world is moving now, we want to get an MRI not to do a biopsy, not to do a biopsy, but get an MRI not to actually have to perform the biopsy. So, if the MRI is negative and you don’t need to do a biopsy, you save finding possibly cancer that doesn’t need to be treated, because an MRI is really good at finding the larger high-grade lesions.
It’s a negative predictive value, which is the true negatives divided by the true negatives plus the false negatives. We don’t want to have any false negative scans. We want the scan to be correct 100% of the time, which means there’s no lesion there. The negative predictive value is wrong between about 80-95%; the average is low 80s.
But the problem is, what’s the bar? The bar is going to be for most of these studies, the definition of clinically-significant cancer. If you’re talking about any cancer, the negative predictive value is not going to be very good, but if you’re talking about clinically-significant cancers, and you can have a whole discussion about whether or not we should find them or not find them, then these are the criteria that these different 12 centers use.
For example, is it above the Epstein Criteria? Is it the UCLA Criteria? Is it 4 plus 3? Does 3 plus 7 get in there? These are substantial cancers, so there’s probably a gray zone for the cancers they’re not finding. But just keep in mind when you look at these statistics, it’s because these are the criteria that these centers use for clinically-significant disease. You may or may not agree with it.
So, in conclusion, the MRI has a high negative predictive value. It’s limited to the definition of clinically-significant disease. It finds harvest lesions of the interior of the gland missed by routine TRUS biopsies; no question, it’s very good at doing that.
Cost is an issue. I’ve been hearing in Europe now that the tests are down to $200-600, but in the U.S. it’s very expensive. It’s not covered by all insurance. We haven’t even mentioned the fact that the MRI-guided biopsy is not covered. There’s no reimbursement for the extra time for a urologist to spend doing these targeted biopsies.
I think the real question is if you’re using this test to exclude who’s going to have a biopsy, then will a negative biopsy mean no biopsy, or will it change your follow-up?
So, if you’re getting an MRI because the PSA was 4.2 in a 56 or 58-year-old man, and an MRI is negative, and then the patient comes back six months and the PSA has gone up a little bit, or he starts getting worried, or you start getting worried, are you going to just say, I’m not going to trust that MRI and bring that patient in? Because we know there’s a 20% miss rate. So, that really needs to be determined. I don’t think we know enough about that yet.
One more slide. This is a book that Dave Crawford and I just published on the prostate cancer dilemma. Many of the lecturers are authors in that book. It goes through MRI, it goes through surveillance, goes through markers, it’s got mapping; it’s a really nice book to read. It’s not that long. I even put movies in that book, which you can access online to see the procedures. So, hopefully you’ll take a look at it. Thank you.