The Role of Radiology and Radiomics in Staging & Risk Stratification of NMIBC - Manjiri Dighe

July 1, 2021

Ashish Kamat hosts Manjiri Dighe for a discussion on the evolving role of radiology in bladder cancer diagnosis and staging. Dr. Dighe elaborates on the limitations and advancements in imaging techniques like CT scans, MRI, and ultrasound, emphasizing their increasing importance for assessing local tumor invasion and lymph node spread. The conversation also delves into the potential of newer methods such as virtual cystoscopy, diffusion-weighted MRI, and contrast-enhanced ultrasound, based on a study at the University of Washington. Dr. Dighe underscores the role of artificial intelligence in enhancing staging accuracy and expresses optimism about radiology's growing capability to predict tumor grades and treatment responses.


Manjiri Dighe, MD, FSAR, FSRU, Professor, Department of Radiology, Medical Director of Ultrasound Abdominal Imaging, Director of Obstetric Imaging for Radiology, Adjunct Professor, Obstetrics and Gynecology, University of Washington, Seattle, WA

Ashish Kamat, MD, MBBS, Professor, Department of Urology, Division of Surgery, University of Texas MD Anderson Cancer Center, President, International Bladder Cancer Group (IBCG), Houston, TX

Read the Full Video Transcript

Ashish Kamat: Hello and welcome to UroToday's Bladder Cancer Center of Excellence. I'm Ashish Kamat, Professor of Urologic Oncology and Cancer Research at MD Anderson Cancer Center in Houston. It is my pleasure to welcome today, Professor Manjiri Dighe, who is the Professor of Radiology and Medical Director of Ultrasound and an Adjunct Professor of OB-GYN at the University of Washington Department of Radiology. Dr. Dighe is going to talk to us today about the role of radiology and radiomics in staging and risk stratification of bladder cancer, which, as you all know, is a field that has exploded recently onto the scene after a relatively quiet period over the last several years. So with that, Dr. Dighe, the stage is yours.

Manjiri Dighe: Thank you, Dr. Kamat, for the introduction. And thank you UroToday for inviting me to give this presentation. As Dr. Kamat mentioned, this is an important and exploding field, not only in urology, but also in every aspect of imaging. So if you look at the traditional role of radiology, we do not use imaging for the initial diagnosis. Yes, we do sometimes see bladder masses incidentally but, rarely.  It is cystoscopy that is needed not only for looking at the bladder but also for a biopsy diagnosis as well. Even the NCCN Evidence Blocks from 2020 talked about using cystoscopy imaging only if indicated and, yes, examination under anesthesia or TURBT, but we know those can understage patients as well to approximately 30% to 50% of the time.

So why do we use imaging? Imaging is mainly used to look for local tumor invasion, to look at the tumor spread to lymph nodes, which I think is really important. And, of course, the spread to the upper urinary tract and other organs. Now imaging, we know, has a limitation in that it can not identify microscopic diseases. And even in terms of local spread, it can be difficult in the bladder because of inflammation from cystoscopy or biopsy. The newer techniques in imaging are mainly to improve the pretreatment staging in terms of prediction as to which patients would respond to treatment. These techniques also can provide non-invasive alternatives to cystoscopy, especially in patients who are on surveillance. The most important, I think, is the prediction because if we can predict which patients can respond to a particular treatment, I think that would be helpful in deciding the treatment for these patients.

So the techniques that are available today include CT, MRI, ultrasound, and PET scans. CT, of course, we know, is the workhorse of imaging, especially for bladder tumors. MRI is increasing as well. And there are new techniques that can be helpful in evaluating bladder lesions and, of course, extension. Ultrasound hasn't been used as much and, of course, PET scans not as much in primary bladder cancers. I will primarily talk about CT, MRI, and ultrasound, and especially the ultrasound research that we have done. So in terms of CT, it is a workhorse, and we usually do the three-phase CT for patients who are greater than 50 years old because there is a higher risk of bladder cancer. We may use the split-bolus technique in younger patients because the risk of cancer is low, and really, you're looking for stones in these patients. CT scan has a problem and limitations in terms of flat lesions, which will be difficult to see on CT. Carcinoma in situ is a problem or even tumors that are less than one centimeter in size can be a problem on CT.

If a patient has had recent resection or biopsy, it is important to keep that in mind when reading the CT scans because the changes seen may not necessarily be only from a tumor but could also be related to these interventions. Patients who have had systemic chemotherapy and intravesical drugs can also have an inflammatory response within the bladder and that is important to remember as well. The newest improvement in CT, in addition to the dual-energy CTs, which can decrease the radiation dose and cut down on one scan that these patients will get for a multi-phase scan, I think virtual cystoscopy is important. Virtual cystoscopy is basically where you do 3D surface remodeling, and it can be done with CT or MRI. You can look at the mucosa. For example, in this particular patient, there is a lesion in the bladder along the right lateral bladder wall, but you can actually look at this in turn as if you're looking at a cystoscopy and you can see this lesion over here using virtual cystoscopy.

So you need dedicated workstations for this with computer algorithms in them. Of course, the advantage is that it's non-invasive. Some of the areas can be a problem. For example, in the bladder mucosa or if you are looking at a diverticulum that can be a problem on cystoscopy. But on a virtual cystoscopy, you can easily get to these areas. The disadvantage of virtual cystoscopy is that, of course, you don't get pathology and the low sensitivity in smaller tumors that are less than one centimeter or even flat lesions and CIS.

Now, there are studies that have looked at virtual cystoscopy, and they've had, for example, in 2010, this particular study had a sensitivity and specificity of approximately 92%. Other studies have similarly shown higher sensitivity and specificity. There is good inter-observer agreement as well, and especially when you are looking at tumors that are greater than five millimeters in size. Because of the amount of computer algorithm or extra work that is needed, it isn't mainstream at present but hopefully, in the future, this could be something that could be done in patients who are not good cystoscopy candidates.

MRI has excellent soft-tissue resolution and multiplanar capabilities. We've used it significantly for bladder cancer staging.  It has good accuracy, 85%, for deciding if it's not non-muscle-invasive versus muscle-invasive or if it's an organ-confined disease. And the thing I would like to talk about is the newer techniques, specifically diffusion-weighted MRI. So this is a diagrammatic representation. You are looking at the different layers of the bladder wall and on just routine T2-weighted images, the muscularis propria appears hypointense and you don't really see the urothelium lamina propria. When you give contrast, you are able to see all these layers, and they appear at different phases of contrast enhancement. And then, diffusion-weighted MRI is something that has come up recently where bladder cancers basically have high-signal intensity compared to the adjacent urine and the bladder wall, and you can see them easily on diffusion-weighted imaging.

This is a patient who had a tumor along the posterior right lateral bladder wall, and this is a T2-weighted image.  You can see the tumor, the urine appears bright, the tumor is a little hypointense compared to the urine, and the bladder wall appears hypointense. Comparing that to the diffusion-weighted imaging, and you can see that the bladder tumor stands out compared to the wall and the urine as well, and this is just a fusion between these two images, the T2 and diffusion-weighted images.

So the benefit is that the lesion conspicuity is increased. You can see adjacent lymph nodes quite easily because they would have diffusion restriction on diffusion-weighted images as well. And I think it's complementary to the dynamic contrast-enhanced images that we do in these patients. So again, another patient shows... One of the advantages of this diffusion-weighted imaging is that you can easily see the extension outside the bladder wall. And if you are looking at small lesions, whether they have infiltrated, especially if there are submucosal lesions, they can be seen easily on diffusion-weighted imaging.

There are VIRADS and everything nowadays and so, of course, there is Vesical Imaging-RADS as well. This was created by Dr. Panebianco in 2018, and it is a standardized approach to reporting bladder lesions or bladder cancer. And that's mainly because if you are doing research, it's helpful to have this standardized approach and standardized reporting. And what they decided to do was, in addition to a schematic map as shown in the previous slide, dividing the bladder into different areas, you have a description of what the lesions look like on different imaging. So T2-weighted imaging, contrast-enhanced imaging, and diffusion-weighted imaging. And you would use this to then decide what the VI-RADS for a particular lesion is. We have not started using MRI VIRADS, but that is something that we will be looking at closely in the future.

The newest stuff is about AI, and that is applicable to both CT and MRI, where people have developed models that use machine learning to help in staging lesions. So in 2017, Garapathi decided to use AI to stratify tumors into greater than T2 or less than T2. Some of the other stuff people have done are where they've used radiomics nomogram, so preoperative prediction of lymph nodes, they've done segmentation of lesions, and they are looking at texture analysis. For example, Zhang in 2020, looked at the pathological grade of bladder cancers. And then Cha used models to compare pre-and post-treatment. So there is segmentation done in a particular patient looking at pre-treatment, post-treatment, and then looking at the changes between the pre-treatment and post-treatment enhancement and size of tumors, and that is mainly used to predict whether the patient would respond or not respond to a particular treatment.

Ultrasound is something that I do routinely, but it's not used routinely for bladder cancer imaging. What we use ultrasound for is to mainly assess for hydronephrosis or to investigate patients who have hematuria. So for example, renal masses, looking at the upper tract, stones, et cetera. But contrast-enhanced ultrasound is something that has come up recently. It is superior to conventional ultrasound in differentiating non-muscle-invasive versus muscle invasive. Nicolau, in 2011, was the first person to use contrast-enhanced ultrasound on bladder cancer, and they found that lesions less than five millimeters, you don't have good sensitivity; but bigger lesions, greater than five centimeters, you have good sensitivity and accuracy for bladder tumors. So what you would be looking at is something like this, where there is a lesion seen along the posterior wall of the bladder, and you can see the enhancement in this particular lesion.

What is contrast ultrasound? Contrast ultrasound is basically... it's a low-soluble, low MI imaging agent that basically vibrates and creates its own sound, and we use that to image with ultrasound. What we use in the US is something called Optison, it is sulfur hexafluoride. It is approved for liver imaging, so using it in the bladder or other organs is off-label imaging. In my institution, we use it for renal, testicular, and bladder imaging, mainly for research but for clinical uses as well.

There are quite a few studies done using contrast-enhanced ultrasound, of course, they have a smaller number of patients. We do not have large multicenter studies with contrast ultrasound, but it has been shown that contrast ultrasound can have high utility in differentiating between high-grade and low-grade urothelial carcinoma. Basically, high-grade urothelial carcinomas have fast wash-in and slow wash-out, so the contrast comes in quickly but leaves slowly. Low-grade urothelial carcinomas have fast wash-in and fast wash-out. This is debatable because there have been differences in what these different studies have found. So I think larger studies would definitely be needed.

At UW, University of Washington, our study was basically to look at contrast-enhanced ultrasound in bladder cancer patients mainly to see if it can be used as an initial modality to evaluate the urinary tract and also to assess staging. This was funded by Phillips Medical Imaging. We had a smaller number of patients, we had 13 patients. This was the initial study, and we recruited from the Urology Clinic. We had IRB consent. And then, after we injected contrast and collected these images, we did quite a bit of post-processing to figure out what the contrast is doing, looking at the time-to-peak, wash-in, and the peak intensity slope.

The average age of our patients was 77 years approximately. and most of these were male. Most of them were urothelial carcinomas. Some patients did get systemic chemotherapy or BCG instillation before contrast ultrasound, and then we did look at histopathology in these patients after they had the contrast ultrasound. And what we found was, we could see the masses in 10 patients. The smallest mass we have had was two centimeters in size when it was actually visualized. And then we looked at the time-to-peak in the mass and in the bladder mucosa. And we could see normal bladder mucosa, and we could see a difference between the two. The unfortunate thing was wall invasion was detected by ultrasound, and only one of the six masses actually had wall invasion.

Just to show you some examples.  This is a 59-year-old who had pT3 cancer, so it is along the posterior bladder wall here.  This is the contrast ultrasound. You can see the contrast come in and enhance the tissues, and it is pretty quick. So you may have missed the initial part but just to show you what the lesion looked like. So this is a lesion along the posterior bladder wall. This is at 13 seconds where you can see the enhancement in the tumor, and when you graph it out, this is bladder cancer. This is normal mucosa. You can see that the bladder cancer enhanced quite quickly compared to the bladder mucosa, which does not show as much enhancement.

This is a second patient who had squamous cell carcinoma and had pT2b disease, no prior therapy. There is a polypoid lesion along the posterior bladder wall, close to the trigone. And this is the contrast ultrasound, so you can see the contrast bubbles coming in, enhancing this particular tumor, and we actually followed this into the late phase, up to about five minutes as well. So then, we can do metrics and look at this tumor and look at the enhancement characteristics compared to the surrounding mucosa.

So I've shown you two patients, one with urothelial carcinoma, one with squamous cell carcinoma. We do believe that there is a difference in how these tumors enhanced based on the histopathology and the grade. The only issue is that we were hoping we could look at the extra-vesicular extension better with contrast ultrasound, which hasn't worked out. But I think that the utility could be in patients who cannot get cystoscopy or are getting cystoscopy as a routine and may not necessarily need it. For example, patients who are getting therapy and need assessment as to how much bladder mass is remaining and what is it doing? So that is something that we will be working on in the future.

So in conclusion, for contrast ultrasound studies, we think that it is helpful in flat lesions that may be difficult, but looking at this enhancement characteristic, it is important. And perhaps that will be helpful in the future and, of course, larger trials would be necessary. So we have gone through the imaging, CT, ultrasound, MRI, and cystoscopy. PET is not used as an initial imaging modality. There are some alternative isotopes that are under investigation, which may be helpful in the future, but CT and MR definitely are helpful for the evaluation of bladder cancer. Thank you very much.

Ashish Kamat: Great. Thank you so much for that very comprehensive talk. You covered a lot of information in a very short time, which is great. It is exactly what we wanted. Let me ask you a couple of questions so we can get some high-level thoughts for the audience, and I'm sure they could go back and forth and look at the slides to get some detail. But if you had to pick the imaging modality of choice for the most common scenario, for example, a patient that presents with either gross or microscopic hematuria, what would it be even today in 2021?

Manjiri Dighe: Even today, I think, as an initial modality, we definitely would rely on CT, and it does really well. There are a couple of things. The reason I say that is because, of course, you're looking for bladder cancer, but it could be other reasons why this patient has hematuria as well. I mean, even the older patients may have stones and CT is the best modality for looking for stones. So it's a very comprehensive exam. I did not say MRI, and the reason I did not say MRI is one, getting an MRI done, the availability is much lower. I mean, you look at low-resource areas, CT is much easier available compared to MR.

Secondly, I think, MR may miss some of these known malignancy lesions. Though MRI nowadays does it really well when it comes to metastasis, so we use it often. But I think we reserve MRI for those complicated patients where you know that they have proven bladder cancer, they have extension outside, you are trying to figure out how much has it extended? Where exactly is it located? I think those are the patients that MR would be helpful. And so at present, I would still say that CT is definitely my answer if you have a patient who comes in for the first time presenting with hematuria.

Ashish Kamat: And would that also be your current recommendation for patients who are on surveillance for recurrences?

Manjiri Dighe: Yeah. For surveillance, unfortunately, the resolution... If you're looking at smaller lesions... And surveillance, mostly what we are doing is we are not necessarily looking at the bladder. Surveillance, we are looking at, is there extension outside? Are there lymph nodes that have developed? And these patients do get cystoscopy. And as we've seen, smaller lesions, if you have lesions that are less than a centimeter, let's say, or if they are flat lesions, that could be quite difficult for CT to detect. Those then go to cystoscopy and what we are looking at is, is there an extension outside that we can detect with imaging? MR may be helpful in those patients, and if MR becomes easily available, getting appointments for MRI, even at our institution that maybe has four or five magnets can become a problem for these patients. And so I'm hoping in the future MR would be the answer. But right now, it's still CT.

Ashish Kamat: Yeah. Absolutely. So in surveillance situations, obviously, we are more interested in extravesical lesions, including the upper tracks, obviously, in the lymph nodes and elsewhere for patients who are at high risk for metastasis because these patients do come for cystoscopy. So I agree with you there. Now, the other scenario that we often face, and I'd be interested in hearing what your thoughts are, are the patients that have borderline GFR... At our center, we go fairly low on GFR, and we will often get CTU with contrast, even patients in other places may be considered to have too low GFR to actually have contrast. And in patients who are truly, truly, very low, we'll do MRUs-

Manjiri Dighe: Great.

Ashish Kamat: ... with and without contrast and things like that. What was your recommendation for the average practice per se?

Manjiri Dighe: Yeah, I think, we again use something similar to what you have mentioned. In patients who have low GFR and if the GFR isn't... again, for MRI, we go lower in terms of the GFR even though, as you mentioned, without contrast, using diffusion-weighted imaging, I think, helps significantly because there you can use the T2-weighted images and use diffusion-weighted images in patients who have very low GFR.

Manjiri Dighe: I think in the future, those patients who have very low GFR or have other issues where MRI cannot be performed, a scenario could be where patients have hip implants and they have low GFR, CT causes a lot of streak artifacts. And MR, again, gives artifacts, what do you do in those patients? Perhaps ultrasound contrast may have some utility, and that is what I'm hoping in the future we can get to. Right now, of course, it's mostly used in investigational, it's not used in clinical practice. So at present, for your question, we use a similar protocol as you have and decide whether the patient can get CT or we do MR. And in very low GFR, we do MR without contrast but rely on the diffusion-weighted images in those cases.

Ashish Kamat: Yeah, so it sounds like we are in agreement. And this is something that we would actually even recommend for community practices, is that correct?

Manjiri Dighe: I would say so. Yes. I think MR is becoming more and more available for community practice as well. Diffusion-weighted imaging used to be done mostly as research, but nowadays, we know that we use it day in and day out for liver imaging, for other stuff as well. So it is becoming very common in abdominal work to use diffusion-weighted imaging which is true for community practice as well. I've seen cases that have been sent from outside at our institution to review and people have started using diffusion-weighted imaging, and it's possible to get it done on any magnet, 1.5 or 3 teslas as well. So even for community practice, I would recommend the same thing.

Ashish Kamat: Okay, great. This has been very, very informative and very educational and, obviously, we could spend a whole hour on this topic, but in the interest of time, let me leave you with the stage, just your closing thoughts on where we are in 2021 when it comes to imaging for urothelial cancers.

Manjiri Dighe: Absolutely. So I think in 2021, we've made a lot of progress when it comes to imaging for bladder cancer, I believe, not only as imagers but as working together with the urologist whether urologists are... All of us understand how to use these different imaging modalities better. We've come to a point where we can essentially look towards predicting as to what sort of response or even predicting what sort of tumor grade a patient has based on imaging itself, and I think it's almost getting to the point where we can hope in the future, we have a targeted treatment towards patients based on the imaging itself. And not only imaging, I think every patient you're basically looking at multi-modality evaluation. By that, I mean taking imaging but also pathology, the clinical features, everything together, and making a decision.

Manjiri Dighe: And I think this will keep improving in the future. We will build on it together, as imagers and urologists working towards how can we keep building on getting all this information together for the best care that we can give the patients. So I'm hoping in the future the new techniques become the main streamlined, mainstream imaging modalities and they will be used more often.

Ashish Kamat: Great. Again, I can't thank you enough for taking the time and spending it with us during this crazy schedule of yours. It has really been very informative, and I want to thank you, and stay safe, and stay well.

Manjiri Dighe: Well, thank you, Dr. Kamat, and thank you for the invitation. I enjoyed doing this as well, and stay safe to you as well.