Sam Chang: Hi, my name is Sam Chang. I'm a urologic surgeon in Nashville, Tennessee. And we're joined with great pleasure and great excitement to have actually Dr. Vig Packiam, an associate professor at the Rutgers Cancer Institute. Dr. Packiam is really one of the renowned experts in the US and actually internationally, focusing on urothelial carcinoma and specifically bladder cancer.

He's been doing a significant amount of work, actually, utilizing AI computational histology type of test to help us determine best treatment options for our patients. And we've asked him to actually present something that was presented at the AUA in 2025, as well as recently published in Urologic Oncology. So Vig, thank you so much and we look forward to your presentation.

Vignesh Packiam: Thank you so much for having me and for the introduction. And I'm excited to talk about this topic. This is something we've been working on for three to four years and it's finally come to fruition. And I think the timing is perfect, as these therapies and their data supporting them continue to mature.

So as a background, the standard treatment for high-grade non-muscle invasive bladder cancer for the last 40 to 50 years, as well described by multiple guidelines, has been BCG. The main issue with BCG is, especially in the last decade or so, that its availability has been tight. And because of that, people have looked for viable alternatives.

Gem/Doce has emerged as something that is effective and well tolerated in the BCG-unresponsive setting. And due to BCG shortage, people have tried to utilize it earlier in the treatment paradigm and use it as an effective option in the BCG-naive setting.

The curve on the right side is a study that we did at University of Iowa comparing patients who got either Gem/Doce or BCG in the treatment-naive setting. And we found that the results were fairly comparable for both treatments, which is promising. Obviously, this has to be prospectively validated and that process is currently underway. But using this data, a lot of people are excited to have a treatment option when they're hit with a severe BCG shortage.

The trouble is, we don't really have a rational way to know which treatment to use if we have both of them available. And biomarkers are really needed to let patients have a nice precision medicine approach so that each patient can get the best treatment for themself rather than a generic treatment recommendation. So we really do need a predictive biomarker.

So that's what this test offers us. This is the Computational Histology AI biomarker, CHAI biomarker. And the way that this works is we take the H&E slide from the TURBT prior to treatment. And that slide gets digitized, zoomed in at 40x magnification.

And then the algorithm looks for 6 or 700 microscopic features that could be associated with recurrence or progression. These could be immune markers, stromal markers, nucleoli changes, anything microscopic that could be associated were fed into this algorithm.

This was initially developed and validated on a cohort of almost 1,000 patients from over 10 institutions. And what it showed was that the biomarker was able to stratify for risk of recurrence, progression, and even risk of development of BCG-unresponsive disease. That's what all of those curves on the right-hand side show.

So we hypothesized that patients with CHAI biomarker positivity would also be more likely to respond to Gem/Doce versus BCG, and that this could be more than just a prognostic marker. Maybe this could also be predictive.

So in our study, we pooled patients from two centers. We collected over 250 patients with treatment-naive, high-grade non-muscle invasive bladder cancer who received either BCG or Gem/Doce based on drug availability. For the most part, the treatment decision was made due to BCG shortage or just institutional protocol.

A natural question is, was there any selection bias? And were certain patients given certain treatments due to risk features of their tumor? For the most part, we thought the answer was no. And the data kind of shows that as well. There was no difference in clinicopathologic features between patients who got Gem/Doce or BCG. Digitized whole-slide images of pretreatment TURBT were analyzed as discussed. And then we did stats to compare high-grade recurrence-free survival between biomarker groups and treatment groups.

And what we found was that biomarker positivity really did stratify who was likely to respond better to which treatment. So when the CHAI biomarker was present, there was a 90% chance of a patient who got Gem/Doce to have recurrence-free survival at two years, and the two-year recurrence-free survival of patients who got BCG was only 56%. So a significant difference and a notable difference.

In contrast, when the CHAI biomarker was absent, there was no significant difference in outcomes between groups. We did an interaction test that confirmed that biomarker results did influence prediction of response to different treatments, and all of these analyses were maintained even after adjusting for AUA and EAU risk-group clinicopathologic features.

So in summary, the CHAI BCG biomarker was predictive of response to BCG versus Gem/Doce in population of patients with high-grade non-muscle invasive bladder cancer. This may allow for personalized treatment selection, and precision medicine.

It might help us to give the best treatment to individual patients and in the era of BCG shortage, to allocate the BCG supply. And importantly, this does need prospective validation. And we are excited to do that with a variety of trials. Thank you.

Sam Chang: Great presentation, Vig. We usually throw some uncertainty and doubt when there's a retrospective study. But in this case, going back, we know the outcomes of these patients. And blinded to that, you do this test and you see actually how it was able to predict.

And we set up these somewhat—I don't want to say artificial—but we set up some guidelines that the AUA and the SUO have put out when we have limited BCG regarding allocation. But you can really see this being quite helpful in determining who perhaps would better be served if we have a limited supply of BCG with the chemotherapy option as opposed to the BCG option. You agree with that?

Vignesh Packiam: Yeah, I completely agree. I think it would be challenging to use this test without prospective validation, completely to guide decision-making. But I think if you have BCG shortage, we're already doing that. We're already using a lot of soft factors. And this allows for us to have a little bit more of a rational approach.

Sam Chang: And do you think with these H&E stains and the nuclear and microscopic features, any hypothesis on why some of these features make tumors more chemosensitive as opposed to immunotherapeutic-sensitive?

Would this perhaps translate into any of our immunotherapeutic interventions that we currently have—perhaps may not be as effective if the CHAI biomarker test is positive? Tell me your thoughts on that.

Vignesh Packiam: Yeah, that's a really good question. When this was developed, they developed it agnostic to what they hypothesized would be the features that are associated. So it purely just looked at the data and spat something out. But clearly, there is an immune component that's kind of being registered.

We had another similar project that was presented at AUA looking at genomic signatures. And we did find that specifically, the immune component was the most important feature that helped predict response to BCG versus Gem/Doce. So I think this test is also looking at that.

Sam Chang: And then tell us what you and your co-investigators are looking at next in terms of prospective validation in different patient populations. Tell us what you're thinking next.

Vignesh Packiam: Yeah. I mean, a lot of us are very excited about the BRIDGE study. I mean, that's a nice prospective study that's going to really confirm if Gem/Doce is non-inferior to BCG. And it's very popular successful trial. It's been accruing very well. I think the accrual target is going to be hit within the next few months or definitely within the next year. So it would be wonderful to validate the results onto that study.

I think that the company is looking at other trials as well that have had BCG as a control arm versus alternative agents. Recently, we saw the CREST trial data, which was really cool. I don't know if we're working with them as well. But there's a lot of different trials that it would be great to validate this on.

Sam Chang: And then, is this a test now commercially available, Vig?

Vignesh Packiam: Yes, this test is available. We've been ordering it for over a year now. They recently revamped how the test is ordered. So there's a predictive test and a prognostic test, which is cool. So the predictive test can differentiate likelihood of response to different agents, like this paper shows.

And then the prognostic test is really nice. It shows the kind of baseline recurrence rates based on traditional clinical pathologic features. So based on the tumor characteristics. And then it shows the new risk rates based on the biomarker.

So at different time points, we can actually see the difference that's shown there. And it also shows the difference in progression rates, too. So I think that actually is really useful for clinical decision-making when we're thinking about surveillance intensity or duration or use of maintenance. There's a lot of decision factors that it could potentially influence.

Sam Chang: Yeah. And importantly, it's based on H&E slides. I mean, no special antibodies required. No special preparation is required. It's just paraffin-embedded tissue that we all have in terms of slides and tissue blocks. And our pathologists do H&E stains on all their specimens. And so something that is not too difficult.

And now digital processing, sending digital slides is something that almost all pathology departments and pathology institutions and labs have. So you can definitely see where it may become—especially in those problematic patients, as you were outlining—when do we stop? Or should we switch? What should we initiate? We're out of BCG. Unfortunately, we are too commonly in that situation.

So Vig, thank you so much for spending some time with us. We look forward to future research, future papers, your rising superstardom, all those things. And thanks for spending some time with us.

Vignesh Packiam: I appreciate it. Always fun to chat. Thanks.