bourbiza mohamed
Zach Klaassen: Hello. My identify is Zach Klaassen. I am a urologic oncologist on the Georgia Most cancers Heart in Augusta, Georgia. We’re in Chicago at ASCO 2024. I am delighted to be joined by Dr. Mark Markowski, GU medical oncologist at Johns Hopkins College. Mark, thanks a lot for becoming a member of us right this moment.
Mark Markowski: Thanks for having me.
Zach Klaassen: We have heard a ton about AI over the past two or three years. It is infiltrating medication, particularly the ArteraAI Prostate Take a look at. Simply as a manner of background, we’ll speak concerning the information that you simply introduced and charted, however how did it come to be by way of the place we have seen this begin and the place it’s now?
Mark Markowski: Yeah, completely. Artera has been round for various years. And I consider again in 2022, they explored their multimodal synthetic intelligence prostate take a look at in sufferers with localized prostate most cancers. They checked out a number of of the big NRG trials as their coaching cohort to essentially educate their mannequin and set up their biomarker, after which they used one other NRG trial to validate the cohort. They have been in a position to present, based mostly on their biomarker take a look at, that they’ll predict these sufferers who had native illness who wanted androgen deprivation remedy.
And so actually, it was the primary AI take a look at that truly predicted responses to remedy. We used that mannequin right here to use to the CHAARTED research.
Zach Klaassen: Even simply a few weeks in the past at AUA, we’re seeing it now within the post-prostatectomy setting too. So it is persevering with to maneuver all around the illness house and exhibiting some actually cool outcomes.
Mark Markowski: Proper. And it will proceed to evolve. In order the mannequin will get higher and extra information goes into the mannequin, it must be higher in a position to predict outcomes.
Zach Klaassen: Completely. So simply stroll us by means of, at a excessive degree, the CHAARTED research. It appears loopy to say that is an outdated research now, and it’s; it is within the historical occasions of 2015, 2016. However simply perhaps a high-level research design for our listeners to set the stage.
Mark Markowski: Proper. And it even goes again additional than that. It is an fascinating historical past lesson. In 2004, we bought docetaxel within the castration-resistant house, the primary chemotherapy exhibiting survival. After which in 2005, as a part of ECOG, now ECOG-ACRIN, they started designing the CHAARTED research. In 2006, it opened, they enrolled about 800 sufferers. It was a randomized Section 3 research in sufferers with metastatic, basically hormone-naive prostate most cancers. Everyone was getting ADT, and it was plus or minus docetaxel. So half of the sufferers bought docetaxel, 75 milligrams per meter squared, each three weeks for as much as six cycles. And the first endpoint was total survival.
It took a number of years to accrue the research. After which lastly, a few decade in the past, we noticed the outcomes, that chemotherapy extended survival. And as you began to dig deeper into the information, we discovered that high-volume sufferers have been actually those deriving profit. And that turned practice-changing nearly instantly when that paper got here out.
Zach Klaassen: Yeah, completely. In that context, nice background, how does the MMAI mannequin incorporate into that? There are just a few totally different excessive, low quantity … Possibly simply stroll our listeners by means of the way you guys set that up.
Mark Markowski: Yeah. We talked about it earlier than. This mannequin was designed for sufferers with localized illness and it predicted response to ADT. In order that they took the identical mannequin after which utilized it to the CHAARTED information. Basically what the mannequin does is take digital histopathology slides, run them by means of their algorithm, after which add scientific information to that. It generates basically a biomarker rating: excessive, intermediate, and low. We took these scores and utilized them to total survival within the cohort.
I believe the top-line findings from our research have been that the mannequin was in a position to prognosticate survival within the CHAARTED sufferers that we had. We had about 450 sufferers, so simply over half of the sufferers had imaging accessible. So it was a reasonably sizable cohort. We confirmed that the mannequin can independently predict total survival.
To your level, everybody desires to learn about high-volume illness. So we did a multivariable evaluation that included the AI rating, quantity standing, and stage at analysis. If you happen to had native illness after which turned metastatic versus de novo metastatic, these have been all integrated. Even with all these variables, the mannequin was nonetheless in a position to independently predict survival, which is nice. So the mannequin works; although it was designed for native illness, simply taking the older algorithm and making use of it to the metastatic setting, there appeared to be a prognostication there.
Zach Klaassen: The massive query is, we have now all these choices for hormone-sensitive prostate most cancers. What is the subsequent step for this? Will we do it in different trials to see if we must always give this affected person chemo, ought to we give him ARPI? What is the subsequent step for the mannequin on this illness house?
Mark Markowski: I believe that is the million-dollar query: how do you are taking this expertise? For you and me treating sufferers with metastatic prostate most cancers, can we take a look at histopathology slides from analysis or metastatic biopsies, run them by means of the algorithm, and see if it will probably inform us which manner we must be treating these sufferers?
I believe that is sort of the following frontier right here. So the mannequin is for localized prostate most cancers, however as you revise the mannequin and use perhaps the CHAARTED research or STAMPEDE even to re-educate the mannequin, are you able to begin to have the ability to higher predict responses to sure therapies? Once we did it in our research, once more, the mannequin was validated for native illness; it didn’t predict for response to chemotherapy. And once more, that wasn’t the way it was designed, however I believe with extra information, it is probably that we’ll get there in some unspecified time in the future.
Zach Klaassen: That is nice. Clearly not prepared for primetime, however let’s simply take metastatic hormone-sensitive prostate most cancers. If we had biomarker on the horizon, how are we counseling our sufferers? Now we have excessive and low-volume illness. Let’s fast-forward two or three years, how will we counsel our sufferers with this info?
Mark Markowski: Yeah, I imply, I believe that is an awesome query. If we will get a clinical-grade take a look at the place we will put it into the algorithm, get a danger rating, like a decipher rating, and stratify these sufferers as excessive, intermediate, or low danger, then that is going to be a part of the counseling. If you’re high-risk, high-volume, de novo metastatic illness, then you are going to get a sure sort of remedy.
Shifting forward, we have now triplet remedy that is right here for metastatic hormone-sensitive illness. Who wants triplet remedy? Can we de-escalate remedy?
Zach Klaassen: That is precisely what I used to be going to say.
Mark Markowski: So when you’ve got high-volume, de novo mets, and proper now we’re triplet remedy, maybe with AI expertise, perhaps you do not want that. Possibly you may get away with ADT or Abi for these sufferers and save chemo for later. So I believe it is not solely seeing what they want but additionally what they do not.
Zach Klaassen: Appropriate. No, that is properly stated. As a result of I believe as we predict extra about de-escalation, you can see this enjoying into it and saying, “If you’re biomarker low-volume, perhaps we deal with you for 12 months after which offer you a break.” We do not know. I imply, that is the fantastic thing about these conversations.
Mark Markowski: Proper. However I believe we’ll get that info. There are huge information units on the market. We used CHAARTED but additionally STAMPEDE, and there was information at ESMO final 12 months with STAMPEDE and the Artera take a look at with comparable findings. Can we use these huge information units, re-educate the AI algorithm, and apply it to a validation cohort? I believe it will be fascinating within the years to come back.
Zach Klaassen: For certain. Nice dialog. Possibly a few take-home messages for our listeners?
Mark Markowski: Effectively, I believe it is vital to know that AI expertise is right here and it is being utilized to totally different cancers, not solely prostate most cancers. I believe there was an oral presentation in breast most cancers utilizing ArteraAI expertise. So I believe we’re on the very floor of what AI can do, and the extra info we offer, the higher outputs we are going to get.
It might find yourself being a part of our scientific remedy paradigm. You get a rating, you get a stage, you get a Gleason rating, and we put that each one collectively to determine what’s finest for you. We have thought of individualized remedy as perhaps genomics and molecular, however I believe there is a position for synthetic intelligence right here, and we’ll see what occurs.
Zach Klaassen: Yeah, it is nice. Mark, thanks a lot to your time and experience.
Mark Markowski: My pleasure.
