Faculty: Baris Turkbey, Peter Choyke & Rajan Gupta

Moderators: Rafael Sanchez-Salas & Tom Polascik


Baris Turkbey

Dr. Turkbey obtained his medical degree from Hacettepe University in Ankara, Turkey in 2003. He completed his residency in Diagnostic and Interventional Radiology at Hacettepe University. He joined Molecular Imaging Branch, National Cancer Institute, NIH in 2007. His main research interests focus on prostate cancer imaging (multiparametric MRI, PET CT), prostate biopsy techniques, focal therapy for prostate cancer and image processing (segmentation, decision support systems). Dr. Turkbey directs the Artificial Intelligence Resource (AIR), which makes AI tools available to CCR investigators with the goal of developing better screening and detection methods or predictive markers for patients.

Peter Choyke

Dr. Peter Choyke is a Senior Investigator and Chief of the Molecular Imaging Branch of the National Cancer Institute. He received his medical degree from Jefferson Medical College and trained in Diagnostic Radiology at Yale and completed a fellowship at Penn. He has focused on the translation of imaging from the laboratory to the clinic for several different cancer types, with an emphasis on cancers of genitourinary tract. In addition, he has co-developed image guided therapies based on MRI, radionuclides and optical fluorophores. For instance, he has co-developed an optically activated form of cancer therapy known as photoimmunotherapy which is currently in Phase 3 testing.

Rajan Gupta

Rajan T. Gupta, MD, is the Chief of the Division of Abdominal Imaging at Duke University Medical Center and holds the rank of Associate Professor of Radiology with tenure. He has also been awarded a dual appointment in the Duke Department of Surgery based on his work with colleagues in Duke Urology and holds the secondary titles of Associate Professor of Surgery and faculty in the Duke Cancer Institute. Dr. Gupta has authored/coauthored approximately 100 peer- reviewed publications on topics ranging from imaging with hepatobiliary specific MR contrast agents to multiparametric prostate MRI. He has given 100+ invited lectures across the United States and internationally and has also been a co-investigator on two NIH R01 grants with Duke Biomedical Engineering. Dr. Gupta has also been named the Director of Imaging for the newly created Duke Cancer Institute Center for Prostate & Urologic Cancers. This position oversees and supports emerging imaging modalities and techniques that assist in furthering research and clinical care. Dr. Gupta will lead the team in fostering collaborations across nuclear and radiologic imaging at Duke, while partnering with industry investigators to further develop image-guided therapeutics. He will also continue to develop rising stars in the imaging field through mentorship.

Rafael Sanchez-Salas

Rafael Sanchez Salas in a Franco-Venezuelan Urologic Surgeon born in Mérida-Venezuela. He received his medical degree from the Universidad de los Andes in Mérida-Venezuela in 1998 with a Magna laude diploma and the first place of his class rank. Subsequently, he obtained degrees in General surgery and Urology from the Universidad Central de Venezuela, both at the University Hospital of Caracas. He followed with a Fellowship research on Urologic oncology at Memorial Sloan Kettering Cancer Center and clinical fellowships of Endourology and Advanced Laparoscopic Surgery and Urological Robotic Surgery at the Santa María Clinic in Santiago, Chile and at the Institut Mutualiste Montsouris in Paris,France, respectively. At the end of his training period in Paris, he joined the Department of Urology team at the Institut Mutualiste Montsouris, where he is currently an Attending Surgeon. He holds French State Diplomas in Clinical Oncology and Urological Oncology. He sits on several international Urological Boards and serves as reviewer for major Urological journals. Dr. Sanchez-Salas career is characterized by sustained work with academic orientation.

 

Webinar Transcript 

Dr. Tom Polascik:

Welcome to today's segment of the masterclass series. I'm Tom Polascik from the Focal Therapy Society. The Endourological Society and the Focal Therapy Society would like to welcome you to our Masterclass in Endourology.

Dr. Tom Polascik:

We wish to thank Wasatch Medical Specialties, the exclusive national distributor for Galil Prostate Cryoablation, for their grant in support of this activity.

Dr. Tom Polascik:

This is the general housekeeping in terms of the purpose, the audience, objectives and the faculty, which we'll introduce in a minute.

Dr. Tom Polascik:

In terms of CME, what [inaudible 00:00:51] happening is you'll receive a survey from Michele Paoli's group. Please indicate which seminars you have attended and you will get your CME certificate. And you fill out the evaluation question at the end of each seminar.

Dr. Tom Polascik:

Use the Q&A function to ask questions today. All seminars and webinars will be uploaded on the Endourological Society website.

Dr. Tom Polascik:

We have an exciting session today. We're going to discuss the perils of multiparametric MRI. We have several esteemed colleagues. I like to introduce the moderator, Dr. Rafael Sanchez-Salas. Rafael did his work, he's [inaudible 00:01:36] in general surgery and urology. He's fellowship trained in Memorial Sloan Kettering. He's done advanced laparoscopic surgery and robotic surgery. And currently works at Montsouris Institute in Paris. Rafael.

Dr. Rafael Sanchez-Salas:

Thank you very much, Tommy. It is a real pleasure to moderate this session on multiparametric MRI. As we know, multiparametric MRI revolutionize prostate cancer diagnosis and treatment, but the tissue characterization is not without limitations. We know that not all MRIs are created equal. Today we have a distinguished faculty, Dr. Turkbey, Dr. Choyke and Dr. Gupta. We're going to address pitfalls, PI-RADS 3 assessment and how to improve their reading. To start with this exciting session, we will go with Dr. Baris Turkbey. Dr. Baris Turkbey is a senior clinician, the molecular imaging branch at the National Cancer Institute.

Dr. Baris Turkbey:

Thank you. Good afternoon everybody. I'm very honored to be here today to give this brief talk about prostate MR imaging pitfalls. As outlined, prostate MR imaging is very popular in the last decade. This is all the results of those nice phase 1A, sorry, level 1A studies and results. It has a big popularity but it comes with some certain pitfalls. These are my disclosures.

Dr. Baris Turkbey:

There are several pitfalls or challenges of prostate MRI and it's not possible to outline all of them here today but I just brought you some highlights of some of those. The biggest issue or important pitfall is the quality of the prostate MRI. When you are looking into an MRI, you need to understand the quality level of it. I would like to use some examples. This is a sample from our cohort. As you can see, we're looking at an axial T2-weighted MRI, and then the ADC map, and the b1500 diffusion-weighted MRI.

Dr. Baris Turkbey:

Interestingly, half of the data on this ADC map and high b-value DWI is gone. And if you look at closely, this is because of this rectal gas. This is rectal gas is resulting a significant susceptibility artifact in the posterior aspect of the prostate, saw in this particular patient. You don't have a read out for the diffusion and the ADC data. This MRI is absolutely not acceptable.

Dr. Baris Turkbey:

We call this patient back and we imaged him again with some bowel preparation. As you can see, the diffusion data and the ADC map is there this time because there is no air in the rectum. Interestingly, the high b-value DWI is showing us a focal lesion which is a PI-RADS 4 lesion. You need to be aware of such an important pitfall. What kills diffusion-weighted MRI which helps us to pursue cancer process is almost always the rectal gas related artifacts.

Dr. Baris Turkbey:

This is another example from another vendor. As you can see, there's minor air in the rectum and the whole diffusion data is gone. This is another sample from a different vendor. This time, and as you can see, there is minimal gas and the prostate is appearing elongated like a comet and this data is not acceptable. You need to be able to pursue this important pitfall.

Dr. Baris Turkbey:

Quality related issues do not appear in diffusion-weighted MR imaging and also it happens in morphologic imaging such as T2-weighted MRI. This image shows a prostate with a significant motion. As you can see, like the internal structures are visible to you but you cannot distinguish between like this anterior lesion versus the rest of the gland. Technologist at that time have ceased this problem and immediately repeat the T2-weighted image and here you can see the BPH process, the peripheral zone and this anterior lesion. We need to be aware of this pitfall, motion related artifact on T2.

Dr. Baris Turkbey:

Another thing I want to mention is geometric consistency. This is an example referred to us last week from another center [inaudible 00:06:18] image. There is a focal lesion in the left apical of the prostate, apex of the prostate. The same patient had another imaging in 2019 and this was reported as a new lesion but if you look at a cross check, the lesion is visible more anterior, then. Why did this happen, because they scanned the prostate with different geometry in 2019 versus 2021. This is a sort of progression related with geometric inconsistency. This is also an important pitfall for the quality of imaging. We may have several examples but I try to bring you the most common ones.

Dr. Baris Turkbey:

Let's go back into details of how we read the MRI. Finding cancers in peripheral zone is relatively easy compare to transition zone because transition zone there's a BPH process and unfortunately, this is different for every patient. Again, this is a sample from our cohort. As you can see, there's a diffusion restricted area here in the right transition zone. If you look at carefully, [inaudible 00:07:25] boundary but this suggest the BPH process. However, you need to confirm it with other planes such as cited in this [inaudible 00:07:34] plane. This is part of a BPH nodule which is very [inaudible 00:07:38]. So, this is, like in atypical BPH nodule and biopsying them is giving a very low yield, almost 4% cancer and within that 4% very rare you can find clinically significant cancers.

Dr. Baris Turkbey:

This is another example for a BPH process. This time we have a very well defined peripheral zone but in the transition zone, we have two large nodules. Often times, they send me a reports and I see that they are calling this is a PI-RADS 2 MRI, this is wrong because this MRI does not include a focal lesion and to reach a suspicion [inaudible 00:08:20] to be called that on the report. Instead, what I recommend is, you need to report those MRIs as PI-RADS 1 with BPH changes because none of these lesions or BPH nodules are worth to sample.

Dr. Baris Turkbey:

Another pitfall is inflammation. It is often mixed with high grade PI-RADS lesions. This is an example referred to me from outside NIH, 54-year old patient, PSA 6.2. There is a focal heterogeneous lesion in the left peripheral zone. If you look, this lesion is showing some positive features on diffusion and ADC map, and it nature shows heterogeneous enhancement. If you go by the book, you need to give PI-RADS 3 to this lesion and DCE is positive it becomes a PI-RADS 4. But, a real four should like this, like from 10 meters distance, you need to be [inaudible 00:09:16] about this hyperintense and hypointense. So therefore, I really suggest everybody to give the PI-RADS score but give a differential such as a chronic inflammatory process in this patient.

Dr. Baris Turkbey:

Another pitfall is AFMS versus central zone and they are so rare cancers but we are still seeing lesions in those locations. This is an example from our cohort. If you look at the T2-weighted image only, there is a PI-RADS 5 lesion here. But interestingly, this lesion is not showing diffusion-positive nor it's enhancing focal. If you [inaudible 00:09:53] with the book, it is PI-RADS 5, however it is wrong because, take the other plane, again you are seeing the nice AFMS anterior with no cancer features on the other part sequences, so you can just say, in order to avoid any confusion, this patient has a prominent anterior fibromuscular stroma to avoid confusions.

Dr. Baris Turkbey:

This is a real cancer type which is affecting the anterior fibromuscular stroma. It has to positive in all pulse sequences, as you can see here. This is a PI-RADS 5 lesion, with possible EPE.

Dr. Baris Turkbey:

Central zone is a different area because it changes a lot along with the transition zone during the aging process. In this particular example, there is a focal hypointensity here, and there's another focal hypointense over there on the other side. Either call both of them as cancer suspicious or you go to the other pulse sequences as you can see on ADC map, the left side of lesion is more hypointense, high b-value tells me the same thing, and interestingly, DCE-MRI shows focal enhancement. So, this is a real PI-RADS 4 lesion affecting the left central zone and the guided biopsy was positive [inaudible 00:11:13].

Dr. Baris Turkbey:

Sometimes, lesions can be hidden. We all talked about transition zone mostly and the other areas but not too much peripheral zone, this is an example to that. Let's look at the T2-weighted imaging here it's [inaudible 00:11:28] peripheral zone. There is no focal lesion but if you look at the high b-value DWI, there is a focal restriction here. This is confirmed by the ADC map and it is showing enhancement here on DCE-MRI. So, let's score it together. This is a PI-RADS 1 on T2, sorry. It is PI-RADS 3 in DW and it's showing enhancement so it's overall four. But I really recommend you to comment to on the heterogeneous peripheral zone signal pattern in such patients because finding cancers, over or underestimation is quite likely in those patients.

Dr. Baris Turkbey:

Again, pitfalls are numerous but I try to bring you the most common ones. I really appreciate for [inaudible 00:12:13] opportunity and the end I guess there will be a QA session and happy to answer your questions. Thank you so much.

Dr. Rafael Sanchez-Salas:

Thank you very much, Dr. Turkbey. Very important variable to take into account when we do the analysis. We saw that not all MRIs are created equal but we move forward. Let's say that there's a high quality MRI and we look for a PI-RADS 3 lesion. Those lesions that are with equivocal likelihood of a significant prostate cancer. For that, Dr. Peter Choyke is going to address the issue of this lesion. Dr. Choyke is the chief clinician at the molecular imaging branch at the National Cancer Institute. Dr. Choyke, please.

Dr. Peter Choyke:

Thank you Dr. Sanchez-Salas. Great. Today, I'll be focusing on should PI-RADS 3 lesions be biopsied? Here my disclosures. Let's just review PI-RADS. PI-RADS 1 is highly unlikely to represent cancer and therefore, no biopsies recommended. PI-RADS 2 is unlikely to represent cancer, clinically significant cancer, so again, no biopsy. Let's get over to PI-RADS 4, likely and PI-RADS 5 highly likely. So those, both, would mandate biopsy. But PI-RADS 3 is this equivocal range where we're putting the likelihood of cancer in an equivocal level. That creates, it was known from the beginning, this would create problems but some lesions just don't fall neatly into one or two or four or five. And so, this category was created.

Dr. Peter Choyke:

Let's look at the definition. In the peripheral zone, the dominant sequence is the diffusion-weighted MRI. Focal, discrete and different from the background, hypointense on ADC. Like in this example. And/or focal hyperintense on high b-value, so here slightly hyperintense. It can't be markedly hypointense on one or markedly hyperintense on both of them but it can be one or the other. This is a classic example of a PI-RADS 3. It's sort of almost a PI-RADS 4 but is not quite.

Dr. Peter Choyke:

On the other hand, in the transition zone, the dominant sequence is the T2-weighted sequence. The definition of a PI-RADS 3 is a heterogeneous signal intensity with obscured margins. You can see in this situation, there is a lesion. You can barely see it on the T2 as distinct but it doesn't really enhance. And so, it would be a... It's neither a two, a four, or a five so it becomes a three.

Dr. Peter Choyke:

It's always important to remember the special exceptions for PI-RADS 3. Dr. Turkbey mentioned enhancement in the peripheral zone is something that can bump, ordinarily a PI-RADS 3 lesion to a PI-RADS 4 lesion which should be biopsied. And in occasion of the transition zone, if the diffusion-weighted image is very strongly one way that is positive, you can bump to a four as well. That makes the overall category of PI-RADS 3 somewhat complicated to interpret.

Dr. Peter Choyke:

Here, just some examples. A 60-year old male with a PSA of 3.91, PSA density not very elevated. You can see there's a very subtle lesion on the T2. It's kind of visible on ADC but not really dark, not super bright on the b2000 and the enhancement is equivocal. It's a overall three, nonetheless biopsied and it was Gleason 3+4 lesion which [inaudible 00:16:51].

Dr. Peter Choyke:

Here we're going to over example of a TZ lesion. 73-year old male, slightly elevated PSA, PSAD is 16%. The T2, we're just going to ignore these lesions on the peripheral zone which are PI-RADS 4 lesions. But just focusing on this TZ lesion, you can see that it has a DWI of four and it does enhance but the dominant sequence here is a T2-weighted and it's really looks pretty much like a BPH nodule. That is an example of a lesion that we would call three but because of the four lesions in that case, we would end up biopsying.

Dr. Peter Choyke:

Now, the clinical studies, what the outcomes of PI-RADS 3s are, there are numerous studies in the literature. Here's just one that I've selected for more detailed look. They collected cases between 2014 and 2019. They had done over 3,000 MRIs and 400 of these were PI-RADS 3 which is good sign. You don't want to see a collection of PI-RADS 3s that were the dominant classification. It should be a minority of the cases are PI-RADS 3. If we took these, 90% of them did have biopsy, 8% had radical prostatectomy or 2% TURP. They were verified. 26% were clinically significant PC, 50 point percent low grade PC, and 23% normal.

Dr. Peter Choyke:

This is pretty much middle of the road kind of result. I've outlined some other studies. There are numerous studies and literature. If you lay variant the size of the study, and here's the variation in the percent positivity for clinically significant PI-RADS 3 lesions. They ranges in the literature from 10 to 38%. The example that I just gave you was ones in the middle at 26%. Clearly, different places have different criteria for PI-RADS 3. It's going to vary a little bit. It's going to be on the low side but inside this group, they are obviously patients who are higher risk than other patients. Majority of the patients will be low risk for clinically significant cancer but there's a significant minority that will have clinically significant cancer.

Dr. Peter Choyke:

How do you suss those out? If we look at more detail, the total number of PI-RADS 3 from this one example, 17% were insignificant cancer, 19% significant cancers. You can look at whether they're biopsy naïve or not, that doesn't seem to make a huge difference. Whether they had targeted biopsy or systematic biopsy or the combination, well the combination did increase the amount of clinically significant disease. And whether they were using version one or version two, again not a game changer in terms of the outcome. There really are no clean solutions to this problem.

Dr. Peter Choyke:

Here's our approach. And that is if there's PI-RADS 3 lesion on a dominant sequence and the PSA density is elevated or there's something going on with the velocity of the PSA that is suspicious then we would biopsy a standalone PI-RADS 3 lesion. Actually, if there're in the gland other lesions that are more significant, like PI-RADS 4 or 5, then we're going to be doing a biopsy anyway. At that point, doing another pass of the needle is not a big deal. We would recommend biopsy the PI-RADS lesion. If the lesion shows, is rated greater than three on the non-dominant sequence, for instance if it looks suspicious on the T2 on the peripheral zone or on the diffusion in the transition zone. That, just sort of elevates the risk a little bit and we would do a biopsy in that case. That's perhaps the most controversial in this list. But if there are no added risk factors, and this is a standalone PI-RADS 3 lesion, we typically defer the biopsy, continue to follow the patient but [inaudible 00:21:59] biopsy.

Dr. Peter Choyke:

To conclude, PI-RADS 3 is the hardest classification in PI-RADS and it's hardest to achieve agreement among radiologists. The cancer detection rate for this category varies from 10% to 38% in recent literature with fairly large [inaudible 00:22:22]. The only hints are to look at the risk factors of the patient, a priori, high PSAD, doubling times or velocity associated with greater risk of prostate cancer probably warrants the biopsy. In some cases, we can avoid biopsy if there are no extra risk factors and the patient can go back into screening. The bottom line is to consider the risk-benefit in every patient.

Dr. Peter Choyke:

With that, we'll just acknowledge my colleagues and thank you all for your attention.

Dr. Rafael Sanchez-Salas:

Thank you very much Dr. Choyke. That brings up that one of the question is PI-RADS going to evolve to add some clinical variables with it because definitely the surrounding variables, clinical variables are, most important we will probably discuss on that later. To follow, we need to improve the reading. How it is important to make out the adequate evaluation, the adequate assessment of the information we're getting out of MRI. And for that, we will have Dr. Raj Gupta talking. Dr. Gupta is the chief of abdominal imaging of Duke University Medical Center. He's a director as well of the imaging at Duke's Cancer Institute Center for prostate and urological cancers. Raj, please.

Dr. Rajan Gupta:

Really appreciate the opportunity to be here, very excited to be here with colleagues and friends and talking about something I think we just said a natural progression of what we've talked about today which is the next step to improve the technology or the community expert reader.

Dr. Rajan Gupta:

Would like to first thank the Endourologic Society, the Focal Therapy Society for the invitation. Friend and colleague, Dr. Tom Polascik and again my esteemed moderator and colleagues.

Dr. Rajan Gupta:

Before we talk about what's next, let's talk about where we are. First off, we are currently at the setting what we're with PI-RADS version 2.1 that was released a little while ago and is in the process now of being iterated, PI-RADS is a continuously iterating document. Where we are is, the Society of Abdominal Radiology and the American Urologic Association have put out a joint statement talking about the use of prostate MRI in patients who have negative biopsies. There are ASCO guidelines which start to reflect the use of prostate MRI. We know about the studies of PROMIS and PRECISION. And then, we're currently in the midst of an #MRIFirst movement especially in Europe.

Dr. Rajan Gupta:

When we think about what's next and where we are, the first thing we really need to be thinking about is how do we expand the use of prostate MRI? How do we expand the capabilities? What we need is more consistent study performance and interpretation across the multiple settings. Dr. Turkbey talked a little bit about pitfalls. Dr. Choyke talked a little bit about image quality in PI-RADS 3. What we must agree on is that overall, we cannot restrict prostate MRI to academic and major centers only. If we're going to really have this technology expand out, it's going to have to go the community as well.

Dr. Rajan Gupta:

How do we do that? Well, first off, what's required is that we have to have image quality as paramount. And the way that we do that is that we look to the PI-RADS committee and PI-RADS document to provide us technical parameters that very, very important to follow. The second thing what we need to do is really put in place robust quality improvement programs. I'll talk a little bit about some of the things that are out there but I think that quality improvement is really going to be what's next to take us to the next step.

Dr. Rajan Gupta:

There was a study that was performed in academic radiology in 2018 that talked about the adherence to PI-RADS technical parameters and it being variable across a 107 centers. It turned out that some of the things that had the lowest adherence from a technical perspective was T2 frequency resolution, the field of view of the DWI images and the temporal resolution in dynamic contrast-enhanced imaging. It turned out that the adherence actually to the technical parameters was higher at teaching institutions. But one of the major punchlines of this paper was that the standards warrant greater community education. And I think that's a critical importance that we talk about this topic.

Dr. Rajan Gupta:

What prevents greater adoption of MRI and also these technical parameters? Well, the fact of the matter is, we are in the middle of a ground swell of support for prostate MRI but very candidly, the demand for high quality prostate MRI is vastly outpaced by the ability for most to provide it. It's important to realize that many radiologists who are performing prostate MRI may not have learned this technique in their training programs that's hopefully becoming less common as we get further and further into a more established baseline of use of prostate MRI. But the real challenge for those of us in the space is really to train an entire generation of radiologist to not only perform but to interpret prostate MRI at very, very high level without sacrificing quality and patient care. I think that it's important to realize that a bad prostate MRI maybe worse than no prostate MRI at all.

Dr. Rajan Gupta:

Sorry about that. First, when we think about how we train these folks, we first need to understand that there's a problem. And so I think that where that really comes from is that there needs to be a communication between the radiologist and their referral base. First thing is that many radiologist have been asked to start prostate programs because their urologist would like to be doing fusion biopsies. What is required to do a fusion biopsy is a high quality diagnostic exam with a high quality interpretation. So, the referring clinicians are critical here to press the quality of the radiologist to go even higher. And that really requires that urologists need to know what to look for in a good high quality MRI and to ask for it from the radiologists. From a radiologist standpoint, one of the things that we need to do is provide hands-on courses, provide online education and provide mentorship for those that are really committed to learning prostate MRI and how to do it. And again, what it really gets back to is understanding what a good prostate MRI looks like.

Dr. Rajan Gupta:

One of the big misconceptions about prostate MRI is that it's the same as a pelvic MRI. The fact of the matter is the part of the body you might be imaging might be similar, but the field of view is different, the diffusion technique is different, all of the different things are different. Let's talk a little bit about that.

Dr. Rajan Gupta:

This is a 57-year old male with an elevated PSA of 12.7, had a prior TRUS-guided biopsy which showed positive clinically significant cancer at this right lateral base, showed one out of 12 cores positive with 3% positive tissue. A little bit of unusual situation and the fact that we only have one out of 12 cores positive but we've got a four plus four equals eight cancer. One of the places that MRI's being used is in the pre-operative planning for these patients to determine is this patient a candidate for robotic assisted laparoscopic prostatectomy, basically for staging purposes.

Dr. Rajan Gupta:

This was the examination that was provided to me to look at from the outside, and the surgeon asked, is this an appropriate staging examination? And so, hopefully, the first thing that you can see here on this T2-weighted images is that it looks very different than the T2-weighted images that you've seen already today. Part of the reason being, that you have this tiny little organ sitting in the middle of this entire field of view. This is why prostate MRI is not just pelvic MRI plus. You must take this into account. As we come through here, you have to be able to delineate out what normal gland is compared to tumor. And then, you need to be able to do high quality ADC and b1400 images. When this exam was performed, despite the fact that the biopsy was positive for four plus four equals eight, this was read as a PI-RADS 1 study. Okay.

Dr. Rajan Gupta:

Of course the surgeon, the urologist basically said, "Look, I need some more information. I need to know whether or not this patient is a surgical candidate et cetera." That exam was redone. And so here, this is a three test with three Tesla with endorectal coil. Obviously, endorectal coil and non-endorectal coils is one of the probably most controversial topics in prostate MRI today. We're not going to dive too deeply into that, but we will dive in image quality.

Dr. Rajan Gupta:

Here, what you can see along the posterior gland on the right, is a lesion that is relatively vague, admittedly. But this is really where the dominant sequence comes into play. When you take a look at the ADC map here without the endorectal coil at 1.5T, you take a look at the 3 Tesla, you can nicely delineate that there is a tumor there. When you really start listening to what the PI-RADS committee has told us regarding high b-value imaging, that's where you really start taking it to the next level.

Dr. Rajan Gupta:

A comprehensive modern multiparametric MRI looks like this. T2-weighted images and at least two orthogonal planes. ADC maps composed of at least two b-values. High b-value imagine of at least 1400. And then, profusion maps which basically show us that this is what I call a triple match. You can nicely delineate in the right posterior peripheral zone a lesion that is admittedly vague but not vague when you start thinking about the fact that the dominant sequence here is the ADC and the diffusion imaging that you see here. And then, I really think this is sort of the cherry on top, it really shows that there, in fact, is a tumor there.

Dr. Rajan Gupta:

We go from what we, what's being called the PI-RADS 1 study to what we are calling a PI-RADS 4 lesion at the right posterior peripheral zone at the level of the base, compatible with the findings that were seen on biopsy. We see the markedly restricted diffusion. We see the high signal on b1400 and see abnormal profusion. From a staging perspective, what we basically say, is that we say there's high clinical suspicion of the Gleason seven tumor, compatible with the findings seen on biopsy. And we see some broad based capsular contact which suggest the possibility of microscopic extraprostatic extension.

Dr. Rajan Gupta:

In terms of what's next, this patient is staged as a T2 disease, not a T3 disease based on this thing. And we're also able to clear the nerves on both sides. Therefore, the concept would be this patient could get a robotic assisted laparoscopic prostatectomy potentially with the nerve sparing approach. The teaching points being that 3 Tesla [inaudible 00:32:46] endorectal coil really helps in diffusion-weighted imaging and DCE even more than T2. Many people get lulled to sleep because their T2-weighted imaging is pretty good. They don't realize that where the real decisions are made is diffusion-weighted imaging. And that's really what separate prostate MRI today from the prostate MRI 20 years ago. And realize here, that without optimal technique, you could be calling a negative MRI or PI-RADS 1 study when you already know you got biopsy proven disease there and that is major impacts.

Dr. Rajan Gupta:

This gets to the critical need for feedback. One of the things that we tell our referral base is we're very happy when we're right, and we're very happy when the answer is correct but we need to be told when we're wrong. We need to be told when we find PI-RADS 5 lesions that do not show a positive on biopsy. And the reason why those critical is because it allows us to get into a feedback loop. And it allows us to understand, are these misses that we made from the interpretation standpoint? I will tell you that I learn much more from the cases that I've missed than the cases that we've gotten correct. This has allowed me to learn and adapt. They allow us to find these blind spots which for a very long time was anterior transition zone cancers. They're key for internal peer review. And maybe most importantly, they build accountability to our referrers so that we are fallible but we will do everything we can to fix those issues.

Dr. Rajan Gupta:

This is the feedback loop that would advocate. The concept of performing the MRI, reading and segmenting the MRI, joint review the images and segmentation, performing diffusion and biopsy, and then most importantly, a team histopathology review. Now, in a busy clinical practices, maybe impossible for every case and that's certainly the case in may places. But the idea that identifying what discordant pathology is and concordant pathology and certainly reviewing those discordant cases to make sure that we get the correct answer and the correct next step is critical.

Dr. Rajan Gupta:

What do we need in the future? As we look to the future, we need higher consistency of imaging and reporting. Dr. Choyke showed the results and variable results of PI-RADS 3. We need to know what the positive predictive value needs to be in each of those categories and really have a much more fundamental agreement on how often should a PI-RADS 3 be positive. How often should a PI-RADS 4 be positive? How often should a PI-RADS 5?

Dr. Rajan Gupta:

We need better biopsy guidance. We need to ensure that the image quality is high enough to get high quality diffusion imaging. We need to improve therapy planning through better staging on prostate MRI. We need to incorporate MRI into algorithms for treatment selection for patients. PI-RADS talks about the diagnosis of it but stays outside of treatment for the most part. We really need to incorporate that in. Also, we need to incorporate MRI based metrics for active surveillance. It's important to realize what is a clinically insignificant tumor. What are the metrics that allow us to understand that, that clinically insignificant tumor is actually becoming clinically significant. And then perhaps, and maybe most importantly, we need to really confirm and establish the role of MRI pre-biopsy. We're seeing this movement of #MRIFirst in Europe. The fact of the matter is, that movement needs to continue to move forward into the US. We need to have more MRI in the pre-biopsy setting for those patients who really need it to define their next steps in management.

Dr. Rajan Gupta:

What do we maybe need in the future? Maybe some accreditation for prostate MRI. Many things inside the radiology space have got ACR accreditation. The process of ACR accreditation for prostate MRI likely will raise the stakes when it comes to getting high quality imaging. We need to continue to iterate PI-RADS to go to version three and beyond and really with continued emphasis on what is needed to interpret and acquire diagnostic imaging. Patients need to be advocates for quality as well. We talked about referring clinicians having that relationship with their radiologist, patient advocacy is also going to help us. And then, we need better alignment with vendors to continue to ensure that more people can obtain optimal image quality on their scanners and very [inaudible 00:36:40] with the people who are on this call in pushing AI forward. We need AI to help bring the readers up to higher level and our colleagues at the NIH-NCI, Dr. Turkbey and Dr. Choyke are pushing that as well as others.

Dr. Rajan Gupta:

To summarize, the technology is here. We need to know how to use it. Most modern scanners can do prostate MRI if you know how to optimize your diffusion-weighted sequences. We need to develop the community expert reader because without them, we will not get widespread adoption of multiparametric MRI worldwide. And, this will only occur if there referrers demand high quality interpretation and performance of exams, if there's greater collaboration from radiologists and if all of us are afraid from not learning from our mistakes.

Dr. Rajan Gupta:

With that, I want to thank you all very much for you attention and we'll be happy to take any questions. Thanks so much.

Dr. Rafael Sanchez-Salas:

Well. Thank you very much indeed for these wonderful presentations. I would go with one question that's coming out of the audience. We already have some difficulties and controversies on the evaluation of multiparametric MRI upfront. How about in the post-treatment setting, [inaudible 00:37:55], post-radiation setting, do you have some hints to approach those cases that already had treatment? Perhaps, we can start with Dr. Turkbey.

Dr. Baris Turkbey:

Yeah. That's a very good question. I forgot to tell, the pitfalls I covered today were related with treatment naïve patients prostates. But that's a different league of disease, that's a different form of disease, biochemical recurrence or residual cancers. For those, first of all, we can't apply PI-RADS. There's a new system released mainly by European colleagues and we try to support it as much as we can. It is PRRR. So, it is being evaluated for its accuracy settings. But I think, doing a good MRI is much more critical in that setting because this is like the, one of the last stops before the patient goes to a different phase of the disease or to be able to control the disease locally. The same quality settings apply and it has to be more tailored like radiation, [inaudible 00:39:07] or focal therapy or surgery. Also, MRI has been trying to be more useful if it is used in a joint fashion with some targeted PET tracers such as PSMA, Axumin or choline. That is my short message about post treatment setting. Thank you.

Dr. Rafael Sanchez-Salas:

Good. Dr. Choyke.

Dr. Peter Choyke:

Well, the only thing I would add to that is the focal therapies of prostate cancer can leave the gland very distorted and irregular. One of the consequences of focal therapy is making the diagnosis of occurrence that much harder. That is one of the pitfalls, I think, that were seeing with say cryo or ablative therapies.

Dr. Rafael Sanchez-Salas:

Raj.

Dr. Rajan Gupta:

I would totally agree with that. I think that the post treatment prostate is one of the most challenging examinations that we read. Just because again of the distortion of the gland, and secondary to the various types of therapy that are out there. I think that, to a point that Dr. Choyke made earlier on about the fact of kind of on-book versus off-book reading, PI-RADS obviously does not explicitly state this, but diffusion, DCE-MRI is the most important sequence when were talking about recurrent disease. I think that's an important thing to know. I also think understanding the clinical histories, of critical importance, right? If you have a biochemically recurrent patient, you know the patient has disease. The question is, can you find it? And so, looking for subtle findings on diffusion, looking for the type of therapy, looking for distortion of the bracytherapy seeds, looking for recurrence inside of lymph nodes, all of those different things come into play. It really requires you to be a little more of a detective on a post-treatment prostate MRI and really requires even higher level of experience.

Dr. Rafael Sanchez-Salas:

Right. I agree with that. I certainly believe that teamwork is the bottom line here because as you mentioned, doesn't matter how advanced you are in your reading, you need the clinical factors to make a good call. In that sense, how are we moving from the evaluation of the traditional multiparametric MRI to the biparametric MRI? We mentioned several variables. So, I would like each one of you to mention, are you actually moving to biparametric MRI? Or are you staying on the traditional way?

Dr. Peter Choyke:

Go ahead, Baris.

Dr. Baris Turkbey:

I'm one of the members of PI-RADS training committee and we have very long and fruitful discussions about this. Definitely, there is some good evidence from northern and central and western Europe, actually from all of the Europe and some are from Australia. We need to read those papers very carefully because some of them are retrospective studies. We just need that level 1A evidence to confirm that biparametric MRI is at least performing equal into the multiparametric MRI. When we see that, I think we will be more comfortably be able to tell that biparametric MRI will be the way to go. In our personal practice, also, when we speak to patients, some of them are avoiding contrast due to some concerns that they see in the media so we tailor it biparametrically. But again, scientifically talking, we need to document it in a good experimental study and then we can move more comfortably to that direction. Thank you.

Dr. Peter Choyke:

Yeah. I would say on this that there's going to be a good compelling reason for biparametric meaning just the T2 and the diffusion sequence no gadolinium. That's usually in the setting of a high volume screening situation where you're going to try to reduce the time involved and the cost involved of the MRI. We're not there yet, at least not in our setting. The added confidence that you get from the gadolinium seems worth it to us at this point. But we could definitely, we would definitely sign on to biparametric if we're talking about using MRI much more frequently in the management of prostate cancer and we need to screen large numbers of people more quickly and at less cost, it's significantly less cost. That where it really starts to play.

Dr. Rajan Gupta:

I totally agree with that. I think one of the things that cannot be understated is, overstate is that experience really matters when you're looking at MRI. I think that the confidence that comes from having the gadolinium in place for a multiparametric, especially when you’re starting a practice, is really critically important. I think that, what I would say is biparametric MRI is a little harder to read because it's one less sequence, one less piece of information that you can use to make that diagnosis. And though it may not change your scoring, I think you saw in the cases that were shown today, when something lights up in that way and has all the corresponding features on T2 and diffusion, you feel a lot more comfortable with it. So what I would say is biparametric MRI is best used in more experienced hands and I'd really applaud the PI-RADS committee for actually explicitly talking about biparametric MRI in version 2.1, when it should be used, when it shouldn't be used. And I think it's an evolving conversation.

Dr. Rafael Sanchez-Salas:

Yup. You just mention the idea of using multiparametric MRI more especially in the primary setting. We have previously said that the acronyms of the main multiparametric MRI studies give a message. It's promote, you promise precision if you do MRI first. That was our idea. But how often, and this is a provocative question. How often are you doing multiparametric MRI upfront in the American setting? Not necessarily in the academics, but in the community setting because in Europe, for us, it's like mandatory to have multiparametric MRI. How often are you doing this?

Dr. Peter Choyke:

Raj, why don't you get started?

Dr. Rajan Gupta:

Sure. You know there was study that basically talked about the fact that about less than 10% of MRI was approved in a pre-biopsy setting. That was a study that was done a while ago. I think that, it's quite variable. I think it's very dependent on insurance carrier in regards to how often we're able to do it. I think the biggest question, and it comes down to, is we're not talking about large scale screening. Prostate MRI is not mammography. You're not screening everyone the same way. What we're talking about is pre-biopsy MRI in those patients who are at a little bit of an elevated risk, their PSA is up, they've got family history, African-Americans, things of that nature, where the opportunity comes in. It's a more selective screening process. But I'll tell you, we have struggled with uptake on the pre-biopsy settings secondary to the lack of insurance coverage.

Dr. Peter Choyke:

Yeah. I will say that our view is a little distorted because we don't require insurance at our institution. And I will say that the patients in the Washington area are knowledgeable and they, in general, the practices are doing MRIs or trying to do MRIs upfront when a biopsy is suggested. We got a fair number of those referrals for that reason. Especially for the biopsy hesitant person, and that sometimes creates a problem because if they really need a biopsy but the MRI is negative, you have to go to this, well, the MRI is not perfect. And so, you may still need a systematic labs even if the MRI is negative. Baris, do you have [crosstalk 00:47:09]

Dr. Baris Turkbey:

I agree with all what you said and what Raj mentioned. That's one important observation. It is very common for us to get second opinion imaging studies shipped to us. Especially during this COVID situation, it's more frequently happening. And we are very happy to help our patients out there. My observation is, one in 10 come with biparametric MRI. It's mainly these outside studies from United States is coming with, always, with all three pulse sequences. I don't think it is very commonly taken. Again, the observation's [inaudible 00:47:51] but I have a comfort that these are coming all community setting and it is all, mostly, multiparametric.

Dr. Rafael Sanchez-Salas:

We have discussed about pitfalls, about PI-RADS 3 and improved reading. Just to round this information up, my question would be, would artificial intelligence will cover this up? If we have good artificial intelligence coming in, and making an interpretation of multiparametric MRI, would that, in your opinion, solve the problems that we have today? Are we moving forward in getting this. the feedbacks, what we have on every day basis put in together and created artificial intelligence interpretation?

Dr. Peter Choyke:

Well, you have the number one expert in the world with Baris here. I'm curious to hear.

Dr. Baris Turkbey:

Okay. First of all, artificial intelligence is very much data dependent. To have a convincing system, especially in the medical field, for high risk, high benefit reasons, you need to have very strong data which is not possible currently in the prostate. Good groups are trying to do good work but currently there is this data problem. The second problem is, we couldn't figure out the interaction between AI and the radiologist, what type of interaction is that. Because you show and AI driven drive [inaudible 00:49:28] to a radiologist but radiologist is using the old school knowledge to interact with the AI systems. We're not ready there yet. I don't think so the AI will replace a radiologist. AI will be, just the second opinion. Like Dr. Choyke defined, sitting right next to your right or left shoulder. And if you want to get the second opinion, you just click the button and get the opinion. We are maximum at that levels. This is my belief. But, again, it's a very good topic to explore in a multi-center, multinational pattern but we're not there yet.

Dr. Rafael Sanchez-Salas:

What would be the accuracy of artificial intelligence at this point?

Dr. Baris Turkbey:

Okay. The best human is performing about 0.7 or 70% sensitivity, we have to be honest, in a real setting not in a surgery setting and in a real practice environment, prospectively. Okay. The artificial intelligence systems we see are always using cross validation techniques and they are reporting more than 90% accuracy which doesn't reflect the reality. The real experiments are very rare and the one that we did in NCI recently, gave us a comparable performance with the human which is between 0.65 to 0.7% sensitivity. Specificity wise, AI is doing worse that human to be honest.

Dr. Rafael Sanchez-Salas:

And if [crosstalk 00:51:04]

Dr. Peter Choyke:

Yeah. I was just going to add, very similar to drug development, this AI algorithms are going to face a kind of valley of death because to get... What you see in the literature is very misleading, lots of AI coming up, but there's single institutions, relatively small data sets. To get FDA approval, you're going to have, to have much bigger data sets and much more diverse and you're going to have to show, validate all these findings. And then, it's going to have to be inserted into a packed system somehow in a routine manner which is costly, and the whole approval process is going to be costly. To actually get a AI functioning on a radiologist desktop, we're still quite a distance away from that.

Dr. Rafael Sanchez-Salas:

Dr. Gupta [crosstalk 00:52:06]

Dr. Rajan Gupta:

Oh, go ahead. I was just to say, yeah. The only thing I have to say is the following. So this best AI algorithms, like Baris was describing, perform at a certain level, but they require a certain level of MRI to be able to perform at that level. This actually, even more so, underscores the need for like high quality imaging to come through because many of the places that the AI is being studied at is places that do incredibly high quality imaging. And so you're putting in a super high quality product into it and training those algorithms. I would challenge any AI algorithm to say, take a bad image and tell me what the prostate cancer is. It's going to be very limited but that's not AI's fault, that's our fault. That's something that we need to do a better job to make sure that the quality of the imaging is high enough to use AI appropriately.

Dr. Peter Choyke:

Now, there is one really interesting algorithm that we play around with, that takes kind of not so good MRIs and uses the assumed knowledge to create a good looking MRI.

Dr. Rajan Gupta:

Yeah.

Dr. Peter Choyke:

We just want to make sure that we're not eliminating the lesions when we do that but those are some hopes that we could sort of get to the quality problem with AI as well.

Dr. Tom Polascik:

And one of the things that we noticed is that, in the academics centers we're commonly repeating MRIs as you guys nicely illustrated, we're used to high quality but sometimes the community ones that come in, if we can get somehow enhance the reading of those, as opposed to having to completely repeat them, I think it'd be very cost-effective.

Dr. Peter Choyke:

Yup, agree.

Dr. Rafael Sanchez-Salas:

So, just wind up. Just want to add it, for me it has been a really learning experience to share this session with you, amazing type information, very thoughtful comments. And I really appreciate the opportunity of having this feedback with people that are really involved in the field. Thank you very much.

Dr. Peter Choyke:

Thank you.

Dr. Baris Turkbey:

Thank you.

Dr. Tom Polascik:

Wanted to say a few words on the Focal Therapy Society. We began in 2019 and have been growing ever since. Dr. Sanchez-Salas is our secretary. This is a society that is involved in imaging and image-targeted therapy. We offer education, hands-on training, masterclass series, quarterly case presentations and tumor board and international symposia. We have the ability to network with direct feedback from key opinion leaders. We'll be launching a Focal Therapy Society Outcomes Registry and clinical trials. And to join, you would simply join the Endo Society and check the box that you also want to be a member of Focal Therapy Society. Our next masterclass will be on robotic radical nephrectomy, troubleshooting and difficult cases presented by Kristen Greene, David Duchene and moderated by Lee Richstone and Benjamin Chung. We all hope to see you this coming fall, September 23 to 25 in September in Hamburg, Germany for the next World Congress of Endourology and Technology Meeting at that time. Again, thank you for your participation.