Roger Li: Thanks so much, Sam, and thanks to UroToday for the kind invitation. I've had the pleasure of giving a talk on some of my thoughts on minimal residual disease detection in the space of non-muscle-invasive bladder cancer, specifically speaking to the role of urinary tumor DNA, which is, as you know, one of my areas of research interest. So I started to talk about looking at some of the biomarkers, specifically urine biomarkers that have been looked at in the past and have failed to make it into the clinic. And on reflecting on some of the reasons why these urine markers failed, it's usually because it's not really clinically actionable. And when we look at clinical actionability, we have to really break it down into low-risk NMIBC and high-risk NMIBC, and in each of those settings, how the urine biomarkers can actually be used to be clinically actionable. And in my opinion, for the low-grade NMIBC patients, we can really use the urine biomarkers to reduce the frequency of surveillance cystoscopies, whereas in the high-risk NMIBC setting, we're really trying to use it to detect tumor recurrence early on so that we can implement treatment intensification in order to decrease the chances of disease progression. And so in considering specifically the setting of high-risk NMIBC, I think this urine biomarker has to have a high positive predictive value.
In other words, we have to have high confidence that once the test turns positive, that it's truly reflective of disease being detected within the urinary tract somewhere. And a good example of this is the use of urine cytology, where even in the face of a normal cystoscopy by white light, if there is a urine cytology that's positive, we are obligated to take the patient to the operating room for a biopsy of the bladder, if not to take a look even in their upper tract by ureteroscopy, which is an invasive procedure, because we are very confident under these circumstances that there is a high likelihood for disease to be present within the urinary tract. And so when looked in that light, I think it's not hard to understand why it is that a lot of the urine biomarkers that have been looked at in the past have not made it into the clinic. For the first-generation urine biomarkers, these are biomarkers that typically use the detection of single proteins or cell morphology to detect tumors. And as you can imagine, it doesn't take much to throw these tests off in the setting of benign urological conditions. The second-generation urine biomarkers expanded upon these panels by using a few genes, either by the way of detecting mRNA or DNA mutation panels or methylation panels, but nevertheless, their positive predictive value still remains relatively low. And so how do we move the needle?
Well, the first thing is that we have to acknowledge that what we regard as ground truth by cystoscopy is actually not the ground truth. So from the Cysview study, we know that at least 20% of the NMIBC patients are missed on white-light cystoscopy alone. So the first thing that we need to do is to really use the best available modality of detecting disease, which today I think is Cysview cystoscopy. But in addition to that, we also have to continue to innovate in this field so we can detect these occult diseases by white light. And of course, recently there's been a lot of enthusiasm for urine tumor DNA based off of the success of plasma circulating tumor DNA that, as you know, have been shown to be very useful in the setting of adjuvant MIBC in the IMvigor011 trial. But nevertheless, when we are trying to incorporate these tests into practice, we still have to keep in mind the limitations, and also to keep in mind the lessons that we've learned in the past. So this is a table from the Salari paper in 2023 looking at the UroAmplitude test. And as you can see down here, even by tweaking the model to maximize the positive predictive value of the test, we're still only looking at about a 60% positive predictive value. And in considering why that may be the case, one reason that I think can contribute to this is the well-described phenomenon of field cancerization that's in bladder cancer, whereby the normal urothelial cells can still share some of the mutations that we know to be tumorigenic. So even in the absence of tumor cells, these histologically benign cells can potentially shed tumorigenic DNA into the urine.
So on the one hand, you may have this phenomenon causing a false positive by showing that there is positive utDNA within the urine when there is truly no tumor. But on the other hand, this may actually act as a precursor lesion or a precondition for forming tumors in the future, because these field cancerizations may in turn turn into cancer recurrences down the line. So which is it? And there's obviously a lot of complex dynamics there. So we actually performed a study here at Moffitt taking advantage of a standard-of-care situation where physicians are obligated to perform a repeat TURBT in the setting of high-risk NMIBC patients. And we asked the question, number one, some of these re-TURBT specimens are going to be benign lesions, whether these benign lesions had contained some of the mutations that are found in the index tumor, and secondly, the urine that was obtained right before the re-TURBT, whether you can find the urine tumor DNA that's associated with the benign lesions within these urine. So this is the schema of the study that we performed. We initially took the index TURBT and ran it through whole exome sequencing. Before the repeat TURBT, we also collected urine samples, and from these urine samples were able to detect using the personalized mutations from the tumors in order to detect urine tumor DNA.
And then finally, we performed the repeat TURBT, and using the tissue samples from the repeat TURBT, was able to run that through the whole exome sequencing again, looking at whether or not these repeat TURBT, sometimes on benign tissue, also contained the same mutations that were found in the index. So I'll just talk you through some of the results that we had found, because our study has not been published yet. But essentially, we found that in a cohort of about 90 patients or so, that we were able to find some mutations, about 5% of the mutations were indeed contained within the repeat TURBT that were shared with their index tumors. But more importantly, we also found that the performance characteristics looking at specifically the mutations and also the copy number burdens from the urine samples gotten prior to the repeat TURBT, was able to parse out those patients with positive disease versus negative, with an area under the curve of over 93%. And there were some patients with false-positive values, as you would expect. So in other words, their molecular urine tests were positive in the face of a negative re-TURBT study, and we specifically looked at the mutations that were found in the index tumor, the repeat tumor, as well as in the urine. And what we found was that for all of these patients who were false positives by the urine, were able to find mutations that were existent in the index TURBT that was not found in the repeat, but was then found in the urine, suggesting that perhaps the urine is capturing some of the mutations or some of the lesions that were not biopsied at the time of the repeat TURBT. And finally, we're able to track some of these patients out. Because as you would imagine, if these were precursor lesions, you could imagine a situation where even though you have a negative biopsy, that they have a very quick time to recurrence. And indeed that was the case that we found in these positive MRD patients, were able to find recurrences usually within the first year.
So in conclusion, I think we're aiming for high positive predictive values for the urine biomarkers, that next-generation sequencing is a great way for us to comprehensively profile the tumor and to understand whether or not the tumor-related DNA can be indeed found within the urine. In my opinion, I think the personalized utDNA way is the best way to go because it increases the specificity. But nevertheless, we still have to figure out some of the performance characteristics of this test in order to maximize the positive predictive value in order to bring it to the clinic.
Sam Chang: Roger, that was a great presentation. It generated obviously a lot of buzz at the SUO. Some simple questions to consider are probably currently unanswerable. Question number one, you have a positive urinary tumor DNA from the urine, obviously. Your biopsies are negative. What do you think in terms of the possibility there is a missed lesion or another similar lesion within the bladder? That's question one. But then question two is, how do you know if you've missed it? I guess with subsequent long-term follow-up, et cetera? Tell me your thoughts about that.
Roger Li: Yeah, so that's a great question, Sam, and a question that's very much forefront on my mind as well. I think we can learn something from how the muscle-invasive bladder cancer world has continued to be pushing on, which is, I think as urologists, we're fixated on the idea that you need to prove that there is disease by histology to understand how to treat the patient. But nevertheless, I don't know whether that is true at all times. So in the muscle-invasive setting, we're fixated on the fact that you need to have pathologic staging of the disease, by radical cystectomy, to understand how to treat the patient in the adjuvant setting. But I would argue, Matt Galsky had put forth at least a great experiment, where through the Hoosier trial, they treated patients with cisplatin, gemcitabine, and nivolumab, and proved that perhaps we don't have to take out the bladder, but rather than just looking at the event-free survival rates after cystectomy, and that's a good enough measure for how effective your treatment actually is.
And I think the same thing could be said in the NMIBC setting, where we may not be able to have histologic proof that there is truly disease behind, but you can imagine a scenario where if you have a positive molecular test, that may actually indicate that your treatment may need to be intensified, and you can also track the response to the treatment by serially collecting your urine to understand whether or not your urine tumor burden is actually decreasing as a result of your treatment. Now, with that being said, of course we still have to understand what truly constitutes positive disease in this setting. And I think in order to do that, we have to do the hard work up front. So not only in the non-muscle-invasive setting, where we're looking at based off of the urologist's assessment of the bladder where the disease is and whether or not there is a cystoscopically clear bladder with biopsy proven negative biopsies, but also in the muscle-invasive setting, we can have these urine assays that we collect and correlate that with the pathologic overview of the cystectomy specimens to truly understand how reflective the urine is of the ground truth, which is how much disease burden there is in the bladder. So I think there's still a lot of work to be done, but we're on our way. But hopefully one day we'll be able to just use the molecular urine test as the true marker for disease in order to select therapies for patients, also to increase the intensification of treatment.
Sam Chang: No, I think your point is really important. And especially now with a growing number of intravesical therapies that at least theoretically, to be honest, would be more effective than TUR in terms of treatment of the entire bladder, as opposed to being visually confined to what's abnormal. Well, obviously there are abnormal changes that we can't see with blue light, white light, whatever, purple light, could be whatever, we can't see. So to have that ability to treat the entire bladder and change, I think it's actually ... I always like history because I'm a lot older than you, Roger, but it reminds me of the fact, when people have had positive cytologies and we "couldn't find tumor," we changed therapies or we initiated BCG or we did something, and many of those patients converted to negative cytologies or didn't develop disease that was clearly evident. So I think that this is really the next step forward, I think, in terms of evaluation, risk stratification, and then early treatment. So it's incredibly exciting. The last question before I let you go is, can we have a test that has ... Because all the statisticians tell me no, because we don't have the perfect test. But can we have that test that has this high predictive positive predictive value, but then just as importantly, that negative predictive value? "Hey, you're okay, we don't need to treat, you're clear, et cetera." Do you think ultimately we'll be able to get that combination?
Roger Li: Yeah. So Sam, I think the negative predictive value certainly is going to be more important in this cancer screening setting. And it's especially important because the incidence of bladder cancer overall is relatively very low. So as you know, a lot of the tests can have very seemingly high negative predictive values, but that's a function of the low incidence of the disease. So we really have to test that in a setting where there's a relatively high prevalence of bladder cancer patients. And traditionally that's been tested in the setting of hematuria patients, but even there, as you know for microscopic hematuria, which there is a great presentation by Yair Lotan on the new guidelines, it's still less than 5% of the patients with microscopic hematuria have bladder cancer. So I think your point is very well taken. There needs to be a test that we have a very high negative predictive value, but we have to also be mindful of the test population that that test has run in. But to answer your question, though, looking at the urine DNA tests that we have, it's exquisitely sensitive. And so I think the negative predictive value for this test will actually be very high. But the problem is that you're getting that at the expense of some false positives. So where to draw that line is still, I think it's going to be the most critical thing that we have to tackle in the future.
Sam Chang: Yeah. No, Roger, thanks so much for the wonderful presentation on UroToday and that presentation at the SUO. We look forward to future trials, and we look forward to the manuscript coming out to look at those numbers and then seeing where everyone and your co-leader, investigators at Moffitt and other places, where we go next. So as always, appreciate it, and look forward to our next time together.
Roger Li: Thanks so much, Sam. It's always fun to talk to you.