Advancements in Oncology through AI Virtual Health Interventions

Drs. John Sweetenham and Angela DeMichele discuss potentially ground-breaking abstracts in breast and lung cancer as well as notable research on artificial intelligence and its impact on cancer care, all of which were featured at the 2024 ASCO Annual Meeting.

TRANSCRIPT

Dr. John Sweetenham: Hello, I'm Dr. John Sweetenham from UT Southwestern's Harold C. Simmons Comprehensive Cancer Center and host of the ASCO Daily News Podcast. My guest today is Dr. Angela DeMichele, the Marianne and Robert McDonald Professor in Breast Cancer Research and co-leader of the Breast Cancer Program at the University of Pennsylvania's Abramson Cancer Center. Dr. DeMichele also served as the chair of the 2024 ASCO Annual Meeting Scientific Program. Today, she'll be sharing her reflections on the Annual Meeting and we'll be highlighting some advances and innovations that are addressing unmet needs and accelerating progress in oncology.

Our full disclosures are available in the transcript of this episode.

Dr. DeMichele, congratulations on a very robust and highly successful program at ASCO24, and thanks for joining us on the podcast today.

Dr. Angela DeMichele: Well, thanks so much for having me, Dr. Sweetenham. It's a pleasure to be here.

Dr. John Sweetenham: The presidential theme of the Annual Meeting this year was the "The Art and Science of Cancer Care: From Comfort to Cure." And this was certainly reflected throughout the meeting in Chicago that welcomed more than 40,000 attendees from across the globe. I know our listeners will be interested to hear some of your own reflections from the meeting now that we're on the other side of it, so to spea

Dr. Angela DeMichele: Yes. Well, I will say that playing this role in the annual meeting really was a highlight of my career, and I feel so fortunate to have had the opportunity to do it. We had over 200 sessions, and in many, if not all of these sessions, we really tried to make sure that there was a case that really sort of grounded the session to really help people understand: you're going to hear about science, but how are you going to apply that? Who is the patient for whom this science really is important?

We had over 7,000 abstracts submitted, and our 25 tracks and their chairs really pulled through to find really the best science that we could present this year. I think what you saw really was a representation of that across the board: incredible advances in lung cancer, breast cancer, melanoma, GI cancers; also really cutting-edge technologies: AI, as we'll talk about in a little while circulating markers like ctDNA, new drug development, new classes of drugs. So it was really an exciting meeting. I mean, some highlights for me, I would say, were certainly the Plenary, and we can talk a little bit about that. Also, we had a fantastic ASCO/AACR Joint Session on "Drugging the "Undruggable Target: Successes, Challenges, and the Road Ahead." And, if any of the listeners have not had a chance to hear this, it's really worth going in and watching this because it really brought together three amazing speakers who talked about the successes in KRAS, and then really, how are we using that success in learning how to target KRAS to now targeting a variety of other previously thought to be undruggable targets. I learned so much. And there's really both the academic and the pharma perspective there. So I'd really encourage watching this session.

The other session that I really thought was terrific was one that I was honored to chair, which was a fireside chat ("How and Where Will Public Investment Accelerate Progress in Oncology? A Discussion with the NIH and NCI Directors") with both Dr. Monica Bertagnolli, who is the director of the NIH, and Dr. Kim Rathmell, who's now the director of the NCI. And boy, I'll tell you, these two incredibly smart, thoughtful, insightful women; it was a great conversation. They were really understanding of the challenges we face conducting research, practicing medicine. And maybe different from leadership at the NIH in the past, they've really taken the approach to say that everything they do is focused on the patient, and they don't limit themselves to just research or just science, that everything that the NIH does, and particularly the NCI does, really has to be focused on making sure we can give patients the best possible care. And I think they're being very thoughtful about building important infrastructure that's going to take us into the future, incorporating AI, incorporating new clinical trial approaches that are going to make it faster and easier to conduct clinical trials and to get the results that we need sooner. So just a few of the highlights, I think, from some really interesting sessions.

Dr. John Sweetenham: It certainly was an extremely enriching and impactful ASCO24. And I think that the overall theme of the meeting was extremely well reflected in the content with this amazing mix of really, truly impactful science, along with a great deal of patient-centered healthcare delivery science to accompany it. So, I completely agree with you about that. There was a lot, of course, to take in over the five days of the meeting, but I'm sure that our listeners would be very interested to hear about one or two abstracts that really stood out for you this year.

Dr. Angela DeMichele: Sure. I'm a breast cancer specialist, so I can't help but feel that the late breaking abstract, the DESTINY-Breast06 trial, was really important for the field of breast cancer. So just briefly, this is a study of the antibody drug conjugate T-DxD, trastuzumab deruxtecan. This is a drug that is actually now approved in metastatic breast cancer, really effective in HER2-positive disease. But the question that this trial was trying to answer is, can this drug, which is built with the herceptin antibody against HER2, then linked to a chemotherapeutic molecule, can this work even in the setting of very, very low HER2 expression on a tumor? I think this is an incredibly important question in the field of antibody drug conjugates, of which there are now many across diseases, is how much of the target do you really need to have on the surface of the tumor?

We had seen previously HER2 overexpressing tumors respond really well to this drug. HER2 tumors that have an intermediate level of expression were tested in the DP04 trial, and we saw that even those 2+ intermediate tumors responded well to this drug. The DP06 trial that was presented at ASCO was looking at this group of patients that have even less HER2 on the surface. So we typically measure HER2 by immunohistochemistry as 0, 1+, 2+, or 3+. And this was looking at patients whose tumors were over 0, but were at 1+ or below, so low and ultra-low. And it turned out that compared to treatment of physician's choice, the drug really had quite a lot of activity, even in these patients who have very little HER2 on their tumors, really showing progression-free survival benefits in the HER2-low and HER2-ultra-low groups that were appreciable on the order of about 5 months, additional progression free survival hazard ratios around 0.6, so really demonstrating that utilizing an antibody drug conjugate, where you've got very little target, can still be a way to get that drug to a tumor.

And I think it'll remain to be seen whether other ADCs can have activity at very low levels of IHC expression of whatever target they're designed against. I think one of the tricky things here for implementing this in breast cancer will be how do pathologists actually identify the tumors that are ultra-low because it's not something that we typically do. And so we'll go through a period, I think, of adjustment here of really trying to understand how to measure this. And there are a bunch of new technologies that I think will do a better job of detecting low levels of the protein on the surface of the tumor because the current IHC test really isn't designed to do that. It was only designed to be focused on finding the tumors that had high levels. So we have some newer technologies with immunofluorescence, for example, that can really get down to very low levels. And I think this is going to be a whole new area of ADCs, target detection – how low can you go to still see activity? So I thought that this was an important abstract for many reasons.

I will just say the second area that I was really particularly impressed with and had a big impact on me were the two lung cancer abstracts that were presented in the Plenary, the LAURA trial (LBA4) and the ADRIATIC trial (LBA5). And I think, I've been in the field of oncology for 30 years now, and when I started in the late '90s, lung cancer was a disease for which we had very few treatments. If we didn't catch it early and surgery wasn't possible for non-small cell lung cancer, really, it was a horrible prognosis. So we knew this year was the 20th anniversary of the discovery of EGFR as a subtype of lung cancer. That was really, I think, a turning point in the field of non-small cell lung cancer – finding a target. And now seeing the LAURA trial show that osimertinib really had such an enormous impact on progression-free survival amongst these patients who had EGFR-positive non-small cell lung cancer, progression-free survival hazard ratio of 0.16; there was a standing ovation.

And one of the really big privileges of being the Scientific Program Chair is getting to moderate the Plenary Session, and it's a really amazing experience to be standing up there or sitting there while the presenter is getting a standing ovation. But this was well deserved because of the impact this is having on patients with EGFR positive lung cancer. And it was similar with the ADRIATIC trial, which looked at the benefits of adding immunotherapy in limited-stage small-cell lung cancer. Again, a disease that treatment has not changed in 30 years, and so the addition of durvalumab to the standard backbone of chemotherapy for small cell lung cancer had its survival advantage. These patients are living longer and it was really an impressive improvement. And I think it really underscores just the revolution that has happened in lung cancer between targeted therapy and immunotherapy has completely changed the prognosis for patients with this disease. So to me, these were really landmark reports that came out at ASCO that really showed us how far we've come in oncology.

Dr. John Sweetenham: Yeah, absolutely. I think that, as you mentioned, those results are truly remarkable, and they reflect extraordinary advances in science. I think we see that both in terms of the therapeutic arena, but also, I think we've started to see it in other areas as well, like symptom control, remote patient monitoring, and so on and so forth, where some of the newer virtual technologies are really having major impacts as well.

Dr. Angela DeMichele: Yes, we really wanted to have a focus on artificial intelligence in this meeting, because it's having such an enormous impact on our field in everything from care delivery to diagnostics. I'd love to hear what you thought was the most interesting, because there really was just new data across the board presented.

Dr. John Sweetenham: I've actually chosen 3 abstracts which I thought were particularly interesting for a couple of reasons, really. They're all based on virtual health interventions, and I think they're interesting in really reflecting the theme of the meeting, in that they are extremely advanced technology involved in the virtual platforms, a couple of which are artificial intelligence, but very impactful to patients at the same time in terms of remote symptom control, in terms of addressing disparities, and in one case, even influencing survival. So I thought these were three really interesting abstracts that I'll walk the listeners through very quickly.

The first of these was a study, Abstract 1500 ("National implementation of an AI-based virtual dietician for patients with cancer") which looked at an artificial intelligence-based virtual dietitian for patients with cancer. This is based on the fact that we know nutritional status to be a key driver of patient experience and of cancer outcomes. And as the authors of the presentation noted, 80% of patients look for nutritional support, but many of them don't get it. And that's primarily a workforce issue. And I think that's an important thematic point as well, that these new technologies can help us to address some of the workforce issues we have in oncology. So this was an AI-based platform developed by experts in nutrition and cancer patients, based on peer reviewed literature, and a major effort in terms of getting all of these data up together. And they developed an artificial intelligence platform, which was predominantly text message based. And this platform was called INA. And as this is developing as a platform, there's a machine learning component to it as well. So in theory, it's going to get better and better and better over time.

And what they did in their study was they looked at little over 3,000 patients across the entire country who were suffering from various types of cancer, GU, breast, gynecological malignancy, GI and lung. And most of them had advanced-stage disease, and many of them had nutritional challenges. For example, almost 60% of them were either overweight or obese by BMI. And the patients were entered into a text exchange with the AI platform, which would give them advice on what they should eat, what they shouldn't eat. It would push various guidance and tips to them, it would develop personalized recipes for them, and it would even develop menu plans for the patients. And what's really interesting about this is that the level of engagement from the patients was very high, with almost 70% of patients actually texting questions to this platform. About 80% of the patients completed all of the surveys, and the average time that patients interacted with the platform was almost nine months, so this was remarkable levels of engagement, high levels of patient satisfaction. And although at this point, I think it's very early and somewhat subjective, there was certainly a very positive kind of vibe from patients. Nearly 50% have used the recommended recipes. More than 80% of them thought that their symptoms improved while they were using this platform. So I think as a kind of an assistant for remote management of patients, it's really remarkable. And the fact that the level of engagement was so high also means that for those patients, it's been very impactful.

The second one, this was Abstract 100 ("AI virtual patient navigation to promote re-engagement of U.S. inner city patients nonadherent with colonoscopy appointments: A quality improvement initiative") looked again at an AI-based platform, which in this case was used in an underserved population to address healthcare disparities. This is a study from New York which was looking at colorectal cancer screening disparities amongst an underserved population, where historically they've used skilled patient navigators to address compliance with screening programs, in this case specifically for colorectal cancer. And they noticed in the background to this study that in their previous experience in 2022, almost 60% of patients either canceled or no-showed for colonoscopy appointments. And because of this and because of the high burden of patients that this group has, they decided to take an AI-based virtual patient navigator called MyEleanor and introduce this into their colorectal cancer screening quality improvement.

And so they introduced this platform in April of 2023 through to the end of the year, and their plan was to target reengagements of around 2,500 patients who had been non adherent with colonoscopy appointments in a previous year. And so the platform MyEleanor would call the patients to discuss rescheduling, it would assess their barriers to uptake, it would offer live transfer to somebody to schedule for them, and then it would go on closer to the point of the colonoscopy to call the patients and give them advice about their prep. And it was very nuanced. The platform would speak in both English and Spanish versions. It could detect nuances in the patient's voice, which might then trigger it to refer the patient to a live agent rather than the AI platform. So, very sophisticated technology. And what was most interesting about this, I think, was that over the eight months of the study, around 60% of patients actually engaged with this platform, with almost 60% of that group, or 33% overall, accepting a live transfer and then going on to scheduling, so that the completion rate for the no show patients went from 10% prior to the introduction of this platform to 19% after it was introduced. So [this is] another example, I think, of something which addresses a workforce problem and also addresses a major disparity within cancer care at the moment by harnessing these new technologies. And I think, again, a great interaction of very, very high-level science with things that make a real difference to our patients.

So, Dr. DeMichele, those are a couple of examples, I think, of early data which really are beginning to show us the potential and signal the impact that artificial intelligence is going to have for our patients in oncology. I wonder, do you have any thoughts right now of where you see the biggest impact of artificial intelligence; let's say not in 20 years from now, but maybe in the next year or two?

Dr. Angela DeMichele: Well, I think that those were two excellent examples. A really important feature of AI is really easing the workload on physicians. And what I hope will happen is that we'll be able to use AI in the very near future as a partner to really offload some of the quite time-consuming tasks, like charting, documentation, that really take us away from face-to-face interaction with patients. I think this has been a very difficult period where we move to electronic medical records, which are great for many reasons, but have really added to the burden to physicians in all of the extra documentation. So that's one way, I think, that we will hope to really be able to harness this. I think the other thing these abstracts indicate is that patients are very willing to interact with these AI chatbots in a way that I think, as you pointed out, the engagement was so high. I think that's because they trust us to make sure that what we're doing is still going to be overseen by physicians, that the information is going to get to us, and that they're going to be guided. And so I think that in areas where we can do outreach to patients, reminders, this is already happening with mammograms and other sorts of screening, where it's automated to make sure you're giving reminders to patients about things that they need to do for some of their basic health maintenance. But here, really providing important information – counseling that can be done by one of these chatbots in a way that is compassionate, informative and does not feel robotic to patients.

And then I was really impressed with, in the abstract on the screening colonoscopy, the ability of the AI instrument to really hear nuances in the patient's responses that could direct them directly to a care provider, to a clinician, if they thought that there might be some problem the patient was experiencing. So again, this could be something that could be useful in triaging phone calls that are coming in from patients or our portals that just feel like they are full of messages, no matter how hard you try to clear them all out, to get to them all. Could we begin to use AI to triage some of the more mundane questions that don't require a clinician to answer so that we can really focus on the things that are important, the things that are life threatening or severe, and make sure that we're getting to patients sooner? So there's just a few ways I really hope it'll help us.

Dr. John Sweetenham: Yeah, absolutely. I think we're just scratching the surface. And interestingly enough, in my newsfeed this morning through email, I have an email that reads, "Should AI pick immunotherapy combinations?" So we'll see where that goes, and maybe one day it will. Who knows?

Dr. Angela DeMichele There was a great study presented at ASCO about that very thing, and I think that is still early, but I could envision a situation where I could ask an AI instrument to tell me all of the data around something that I want to know about for a patient that could deliver all of the data to me in real time in the clinic to be able to help me make decisions, help me quote data to patients. I think in that way it could be very, very helpful. But it'll still need the physicians to be putting the data into context and thinking about how to apply it to the individual person.

Dr. John Sweetenham: Absolutely, yes. And so just to round off, the final abstract that caught my eye, which I think kind of expands on a theme that we saw at an ASCO meeting two or three years ago around the impact of [oncology] care at home, and this was Abstract 1503 ("Acute care and overall survival results of a randomized trial of a virtual health intervention during routine cancer treatment"). So, a virtual platform but not AI in this case. And this was a study that looked at the use of an Integrative Medicine at Home virtual mind-body fitness program. And this was a platform that was used to look at hospital admission and acute care of patients who used it, and also looked at survival, interestingly enough. So what was done in this study was a small, randomized study which looked at the use of virtual live mind, body and fitness classes, and compared this in a randomized fashion to what they called enhanced usual care, which essentially consisted of giving the patients, making available to the patients, some pre-recorded online meditation resources that they could use. And this was applied to a number of patients with various malignancies, including melanoma, lung, gynecologic, head and neck cancers, all of whom were on systemic therapy and all of whom were reporting significant fatigue.

This was a small study; 128 patients were randomized in this study. And what was very interesting, to cut to the chase here, is that the patients who had the virtual mind-body program, compared with the control group, actually were less likely to be hospitalized, the difference there being 6.3% versus 19.1%, respectively. They spent fewer days in the hospital. And remarkably, the overall survival was 24.3 months median for patients in the usual care arm and wasn't reached in those patients who were on the virtual mind-body fitness class platform. So very preliminary data, certainly are going to need more confirmation, but another example of how it appears that many of these non-pharmacological interventions have the potential to improve meaningful endpoints, including hospital stays and, remarkably, even survival. So again, I think that that is very consistent with the theme of this year's meeting, and I found that particularly interesting, too.

I think our time is up, so I want to thank you, Dr. DeMichele, for sharing your insights with us today on the ASCO Daily News Podcast. We really appreciate it. And once again, I want to congratulate you on what was really a truly remarkable ASCO this year.

Dr. Angela DeMichele: Well, thanks so much for having me. It's been a tremendous pleasure to be with you today.

Dr. John Sweetenham: And thank you to our listeners for joining us today. You'll find links to the abstracts discussed today in a transcript of this episode. Finally, if you value the insights that you hear on the ASCO Daily News Podcast, please take a moment to rate, review, and subscribe wherever you get your podcasts.

Disclaimer:

The purpose of this podcast is to educate and to inform. This is not a substitute for professional medical care and is not intended for use in the diagnosis or treatment of individual conditions. Guests on this podcast express their own opinions, experience, and conclusions. Guest statements on the podcast do not express the opinions of ASCO. The mention of any product, service, organization, activity, or therapy should not be construed as an ASCO endorsement.

Follow ASCO on social media:

@ASCO on Twitter

ASCO on Facebook

ASCO on LinkedIn

Disclosures:

Dr. John Sweetenham:

Consulting or Advisory Role: EMA Wellness

Dr. Angela DeMichele:

Consulting or Advisory Role (an immediate family member): Pfizer

Research Funding (Inst.): Pfizer, Genentech, Novartis, Inviata/NeoGenomics