AI-Powered Prostate Cancer Risk Assessment Tool Validated for Low-risk and Active Surveillance - Andre Esteva

October 17, 2024

Alicia Morgans interviews Andre Esteva about ArteraAI's expanded prostate cancer test, now validated for low-risk and active surveillance patients. Dr. Esteva explains that the AI-driven test helps clinicians decide between active surveillance and definitive treatment by assessing the risk of adverse pathology at prostatectomy. The test, which uses digital pathology to analyze H&E slides without consuming tissue, provides personalized risk assessments across all prostate cancer stages. Dr. Esteva highlights the test's validation in multiple studies and its inclusion in NCCN guidelines. He discusses ArteraAI's partnership with PathNet, enabling faster result delivery by integrating AI analysis with pathology reports. The conversation emphasizes the test's potential to improve decision-making for low-risk patients, with about 80% showing a low risk of distant metastasis. Dr. Esteva concludes by noting the test's broad applicability across risk groups and its Medicare reimbursement status.

The ArteraAI Prostate Test is intended for patients who have been diagnosed with localized prostate cancer and have not yet received definitive treatment.

Biographies:

Andre Esteva, PhD, Co-Founder & CEO, ArteraAI

Alicia Morgans, MD, MPH, Genitourinary Medical Oncologist, Medical Director of Survivorship Program at Dana-Farber Cancer Institute, Boston, MA

ASTRO 2022: Prostate Cancer Risk Stratification in NRG Oncology Phase III Randomized Trials Using Multi-Modal Deep Learning with Digital Histopathology

Read the Full Video Transcript


Alicia Morgans: Hi, I'm so excited to be here today with Dr. Andre Esteva, who is the CEO and co-founder of ArteraAI. Andre, I'm so excited to talk with you today about the prostate AI assay or test that you have certainly integrated into clinical care, but this test has a newly expanded patient population that it can help, and that's the active surveillance or low-risk population. Can you tell me a little bit about this?

Andre Esteva: Absolutely. And thank you, Alicia. It is a pleasure to be here and a pleasure to be speaking with you. At Artera, we're really excited about an expanded validation of our test. As you know, our test is one that is personalized to each patient. So depending on the specific risk group of a patient that's sent to us, we will render different use cases to help the clinician make the most informed decision. And now, we've validated in a cohort of patients that are in a low-risk setting, and our prognostic tool can now help clinicians make the very challenging decision of whether or not they should take their low-risk patients and put them on active surveillance or on active therapy.

Alicia Morgans: Wonderful. So when I'm thinking about this, and certainly when I'm sitting across from a patient in clinic, we are always thinking about, do we need to do definitive therapy? Is this disease going to be higher risk than we realize? And there's a limit, of course, to what we're able to understand from biopsies. Even when they are MRI-guided and under the best conditions, we're always worried that perhaps we're missing something. So it sounds like this test is able to help us in terms of giving us a prognostic risk of whether we will have adverse pathology at the time of prostatectomy if a patient like this went to surgery. Is that true?

Andre Esteva: That's exactly right. So we validated in a cohort of about 300 patients who were managed with active surveillance and were tracked for the endpoint of adverse pathology at radical prostatectomy. And so what we will do if you send us a low-risk patient is we will show you the percentile of risk relative to this cohort of patients of your particular patient. And if it's below a certain threshold, you can consider active surveillance. If it's above a certain threshold of percentile risk, you might want to consider either active surveillance or definitive treatment for your low-risk patient.

Alicia Morgans: I think that's so helpful, and what I understand is as you were developing and then validating this model, it does sound like a majority of patients ended up being in a situation where they could be monitored very comfortably without active and definitive treatment. It sounds like about 80% of patients actually had a risk of developing distant metastasis and having ongoing challenges that was less than 3%, which sounds actually very reassuring when I'm speaking to a patient. Is that correct?

Andre Esteva: That's exactly right. So in the cohort of patients that we validated on, about 80% of patients had less than a 3% risk of a distant metastasis. Those are the patients that also correlated really nicely with a good prognosis with respect to adverse pathology at radical prostatectomy. And so if you take a look at the ArteraAI test report for active surveillance insights, what we will do is we will show you that cutoff and show you where your patient sits on one side or the other and help you make that decision. Moreover, we also accept for active surveillance insights patients that are favorable intermediate risk. And in that context, what we will do is we'll render both the active surveillance insights as well as our biomarker for whether or not they would benefit from short- or long-term hormone therapy so that you as a clinician can make the best decision on whether or not you should treat that patient and how.

Alicia Morgans: I think that's really, really helpful. And from a clinical perspective, this is something that I can imagine using day to day, and it's wonderful that it is actually already available. And certainly there is an ArteraAI prostate test actually call-out in the NCCN guidelines, which is helpful. I wonder for those who are interested in how this was developed, can you tell me a little bit about how the model was initially formed and then how it was validated?

Andre Esteva: We have now validated our test in around 30 pieces of evidence published in the form of conference abstracts, papers that were in Nature, in the New England Journal of Medicine, and so forth. We take validation very seriously. And one of the really exciting things about AI today is how easily it ingests massive amounts of data. Our first biomarkers from back in 2021, 2022 were trained and validated on 6,000 patients approximately. We've now more than tripled the number of patients that we work with in a prostate cancer setting and have completed validations across the spectrum of the disease. So whether you're dealing with a low-risk patient or a metastatic patient, or any risk group in between, the ArteraAI prostate test can help you make the best decision for your patient, whether it's a prognostic question with respect to the patient's outcome or a predictive question or one that is clinically actionable with respect to whether or not you should intensify therapy.

Alicia Morgans: Great. One more thing that I just wanted to explore, because I don't know that we've talked about it in other conversations, is this concept of digital pathology, which is, I think, increasingly something that we're using in our practices as a way to perhaps get information from something like an ArteraAI prostate test without using a sample, without using up tissue. So can you tell me a little bit about PathNet and what that partnership does for our ability in clinic to potentially get this test?

Andre Esteva: Absolutely. So one of the key advantages of how our test works is that this is purely an AI. It's a computer vision-based AI that will look at an image of a patient's digitized H&E. It fits seamlessly in a workflow. All you need to do to work with the Artera test is digitize an H&E image or send a slide to us for digitization. We don't consume any tissue, and we render results immediately after digitization. So what that means clinically is that if you have digital pathology already set up, or it's simply a scanner locally at your clinic, you can digitize your H&E slide and get a test report back immediately. And in doing so, you'll obtain both the pathologist's report with the diagnosis and the biomarker from the ArteraAI test report at the same time. That is a very meaningful value proposition of one, which we're actively working to build out for all clinicians, irrespective of their pathology setup.

And to that extent, one of the things that we've done towards this is to set up an integration with a very large digital pathology provider called PathNet. They serve about 10% of the U.S. prostate cancer volume, and that is growing. And in this integration, we now offer up the ArteraAI test with their pathology report in one joint package sent to the clinician so that when they're speaking with their patient, they can explain that they do have cancer, but they can also explain how they're going to treat them. And patients aren't waiting around for weeks wondering about their test report results.

Alicia Morgans: That's really powerful. So I look forward to continuing to hear how that partnership and how that opportunity continues to move forward, because I think one of the challenges, at least for patients in my clinic and across prostate cancer clinics, is that it can feel like things take a long time when they're just trying to get answers and understand themselves better. So thank you for that. So as we wrap up, I would love to hear what would your message be to listeners about the ArteraAI prostate test and this new population that it may be useful for?

Andre Esteva: The ArteraAI prostate test, which is in NCCN guidelines, and it is reimbursed by Medicare and broadly affordable across the United States, is now active for active surveillance and is validated in all risk groups from low risk through metastatic.

Alicia Morgans: Fantastic. Well, thank you so much for sharing this update with us, and congratulations to you and the team for ongoing work. We continue to experience and be really excited about where we can use this test both now and in the future. I appreciate your time.

Andre Esteva: Thank you so much, Alicia. It's a pleasure to be here.