Journal Club: The Evaluation of Cxbladder and Adjudication of Atypical Cytology - Sia Daneshmand

October 15, 2019

Sia Daneshmand discusses a study published in European Urology which evaluated the atypical cytologies to evaluate the performance of Cxbladder diagnostic tests to either rule out low-risk individuals or identify patients at a high risk of urothelial carcinoma. Cxbladder diagnostic tests combine genomic information from urinary mRNA with phenotypic information.  

Siamak (Sia) Daneshmand, MD. Dr. Siamak Daneshmand is an Associate Professor of Urology Keck School of Medicine USC University of Southern California with Clinical Scholar designation and serves as director of clinical research as well as the urologic oncology fellowship director. 

Read the Full Video Transcript

Sia Daneshmand: Hi everyone. My name is Sia Daneshmand. I'm the Director of Urologic Oncology at USC, University of Southern California in Los Angeles, and from the Norris Comprehensive Cancer Center.

And today we're going to talk about this article on "Evaluation of Cxbladder, and Adjudication of Atypical Cytology and Equivocal Cystoscopy". You see there are multiple authors here, all of whom are bladder cancer experts. Dr. Konety from Minnesota, Neal Shore from the Carolina Urologic Research Institute, Karim Kader from UC San Diego, Sima Porten from UCSF, myself, Tony Lough, who's part of the Cxbladder or Pacific Edge team. And of course Yair Lotan, who's at the University of Texas and Southwestern Medical Center. I'm used to saying UT Southwestern.

So in any case, these are bladder cancer experts. We have experience using the test Cxbladder. And a lot of us struggle with this atypical cytology. And so we had assembled a team together to, along with Pacific Edge, to look at all the atypical cytologies to evaluate the performance of this test in this setting. So let's just get started.

Cytology interpretation. Again, we all struggle with this. At every center there's a high proportion of cytology samples that are either atypical, inconclusive, and that's just because either there's not enough cells to be looked at under the microscope or it's very, very difficult. I mean, and this number varies significantly from center to center depending on the expertise and the competence of the cytopathologists looking at these urine cytologies. So for us as clinicians, this is a diagnostic dilemma. What do we do with atypical? Atypical could be positive and it could be negative in many cases. So putting a patient through additional procedures when it's unnecessary is really not where we're trying to go. What we want is a test that can tell us with high confidence there was no cancer there.

So when you look at this study, 153 of 852 cytology, that's roughly 18%, were classified as atypical. And I think that's sort of the number that you find in many, many studies. It can be even higher than that in many centers.

So different labs have different ways of describing the urine cytology test, but these are essentially the ... there are four categories here. There was the unsatisfactory specimen. That means you just don't have enough cells and we call it possy cellular and there are various other terms for it. But taking that out, there are really four results that you can get. Negative, which means absolutely no cancer cells seen. Or sometimes you can have the cytopathologist say no high-grade cancer seen in the urine specimen.

So atypical means there were some abnormalities found in the urine sample, but it's not really quite enough to consider it cancer. That's what the topic of this paper is. Suspicious of course makes us all worry about the possibility of there being cancer because these are found to have psychological abnormalities that are consistent with cancer. And of course positive, you're actually looking at cancer cells. So a significant proportion of these fall into the atypical and suspicious. So we don't really use this test alone to diagnose cancer. If it's atypical or suspicious, of course, we have to recommend a cystoscopy and/or a CT scan to further examine the bladder.

So this study design was actually putting together four separate studies. There was a US primary study. That data was not published, but it included over 350 patients. There was a study called the Monitor study. It was a Cx Monitorâ„¢ that had 750 patients. There was another paper with 480 and then importantly, a real-world study where this was used sort of in clinical practice in New Zealand.

So there are lots of samples here. There are 2050 samples from over 1700 patients undergoing Cxbladder testing for recurrent bladder cancer or hematuria. There were some excluded because the result wasn't available. So there was others excluded because no local cytology result was available. So really we wanted a quality study to zoom in on the samples that were valid samples from hematuria patients tested with Cxbladder or recurrent patients. So 400 in each arm is what we ended up with. Again, very, very high-quality data here where local cytology was also done for this comparison. So good numbers here.

So looking at the results in Table 1, the negative predictive value and sensitivity. The best performance characteristic of this test is its negative predictive value. And that's what we want, right? We want a test that tells us there's no cancer. That's what we want to tell the patients. This is not a positive/negative test, but a test that will tell us with high certainty whether or not there's cancer present.

So when looking at the 852 samples, you see the negative predictive value for Cxbladder was 97.4%. Cytology was 92.6. How many samples had a negative result? Well, 300 had a negative result. That's 35% of the patients, whereas 94% of the cytologies were negative. There were very few tumors that were missing in those samples with a negative result, only eight of them. Whereas in cytology it was 59. So looking at now at the bottom here the Cxbladder true negatives, you see 294. The false negatives was only eight. So the negative predictive value, you see the calculation there, is 97.4. You see with cytology there were false negatives were 59 of these cases. So yes, cytology had a lot more true negatives. But what's worrisome is these false negatives, and this is what we worry about. That we're missing cancer. So cytology ruled out two-and-a-half more times patients but missed almost eight times more tumors.

And this is not unique to this study either. We have seen the same thing with the flexible blue light study that we did where we were looking at detection of CIS with flexible blue lights, cystoscopy in the office. And we found the same thing, that cytology was not sensitive in picking up the high-grade cancers in many, many of the cases. In fact, all the lesions that were seen with blue light alone had negative cytology. So again, this is consistent with those results.

Now, let's look at hematuria patients on Table 2. These are the percentage of patients with negative cytology. There's a Cxbladder-Triage, and then they had a Triage and Detect results, number of percentage of tumors missed in the samples with a negative result. So again, this is 430 samples from patients. Just to clarify that the Cxbladder-Triage and Detect, people ask, are these different tests? They're not. The Cxbladder test is the same. It's the algorithm that's different for Triage and Detect based on clinical and patient characteristics. So basically you're combining two different algorithms here to get a better result here.

So you see for the Cxbladder-Triage, again, the negative predictive value was very high, 97.6%. And if you combine the Detect and the Triage, again 97.8%. And cytology was 94. Again, very consistent with previous studies. 39% had negative Triage, but if you combine them, you got 73% who had a negative result. So the number of tumors missed was four with the Triage, and all these four had low-grade TA lesions, which is not of much concern to us or the patient. And seven were missed if you combine the two. Four had low-grade TA tumors, and two had low-grade with a detect, and only a single high-grade tumor was missed in this 436 sample database. Whereas if you looked at the cytology, 14 were atypical. We had seven high-grade tumors and one mixed. And 11 were negative by cytology. So again, cytology is great when it's positive. The positive predictive value is very, very high. So we use it for that reason, but you can miss significant tumors because the cells are just not shed in the urine. And we're not looking at this.

So looking at the atypical cytology, so now we're zooming in on all the patients who had atypicals on Table 3. This is urothelial carcinoma diagnosis according to Cxbladder results, and you see that these were 153 samples total. Cxbladder was negative in 47 of these patients and positive in 80. Urothelial carcinoma positive patients was zero. So it did not miss a single tumor in atypicals. And again, that is exactly what we want to see. These are the patients we can confidently call and say, "I don't think you have cancer. You certainly don't have high-grade cancer. We can see you back in three months or six months." And this is how I use the test, and I think most of my colleagues use the test. If we have equivocal findings either on surveillance cystoscopy or on cytology. You see that of the positive ones, 26 of them actually had a urothelial carcinoma.

So again, as I said in the beginning the strength of the test is in its negative predictive value. I tell the patients often that this is not a positive/negative test. This will tell me with high certainty whether we can rule out cancer. So the atypical cytology result: 13 were low grade, 12 were high grade, and one with mixed. And these were again ruled out.

So in this study, Cxbladder was able to really adjudicate all the urothelial carcinoma positive samples among the 153 atypicals and rule out the diagnosis of cancer in 47 samples, so 31%. If you take this into a real-life situation, we're talking about a third of the patients not undergoing further testing or cystoscopies or cysto and biopsy. For us clinicians who have to take the elderly patients with multiple comorbidities to surgery to get biopsies and things like that, we know what a value this has because really, the patient goes through quite a bit to get to the OR, to get the preop clearances and things like that. So we're sparing these patients further invasive workup.

If you look at atypical cytology and equivocal cystoscopy, there were 14 patients with equivocal cystoscopy, and two of them were diagnosed with urothelial carcinoma. One was a high grade T1 and one CIS. And both were tested positive for Cxbladder. So again, this is a pretty strong argument for using Cxbladder instead of cytology avoiding this atypical cytology results. I think that'll take some time. We're really ingrained in using cytology as part of our workup, but this is certainly an adjunctive test that can help us in these situations. In fact, whenever I have sort of an equivocal cystoscopy finding, I have the urine set aside, and I will send the urine knowing that at least my cystoscopy is equivocal and if the cytology comes back also atypical, that's the time where this test is really helping me.

So again, to conclude the Cxbladder adjudicated all the UC positive samples among the 153 atypical cytology. Really strong results here. Correctly ruled out 47 of the atypicals again in about a third of the patients. And of the 14 that were atypical for cytology with equivocal cystoscopies, it tested positive for one of the high grade and one CIS tumor.

So it definitely outperforms urine cytology for identifying patients with a urothelial carcinoma. And it's very, very useful again in this atypical cytology and equivocal cystoscopy situation. And so again, is it helpful in every setting? Certainly not. We don't use it on every single patient, but anyone I think who has atypical cytology or equivocal findings on cystoscopy.

People ask, "Well what if you didn't send it and you get an atypical cytology?" Well, those are patients who are, just tell them, give them a choice. We can either biopsy you in the OR or just come in and drop off a urine test and we'll do a Cxbladder. So that's always an option to do it afterwards as well.

So I think with that we'll stop. And again, these are the conclusions that I think we went through before. Thank you very much.