Chapters Transcript Video Setting the Stage: AI Breakthroughs in Gastric Cancer So we have been able to successfully use deep machine learning, artificial intelligence. It's actually the most sophisticated form of machine learning, convolutional neural network, to come up with a way to look at the primary tumor on the CT scan and predict what is the risk or presence of carcinomatosis in this patient based on that CT scan. And we have a 98% accuracy on the initial model that was developed at UCI and validated on UCI patients. Hi, I'm Lindsay Carrillo, director of business development at UCI Health. Hi, I'm Doctor Sunil Verma, the associate Chief Medical Officer for ambulatory at UCI Health. I'm also a laryngologist who sees patients as well. Welcome to Physician Huddle by UCI Health. Today we're joined by Doctor Maggie S Santo, Division Chief of Surgical oncology and the vice chair for research at UCI Health. Welcome, Doctor Sento. Thank you. Happy to be here. Doctor Sentel, tell us a little bit about your clinical role at UCI. I know you wear a lot of hats. I am the, as mentioned, chief of division of surgical oncology, and I oversee all of the work on surgical cancer care that is rendered at UCI as personally as a clinician, I take care of patients with advanced gastrointestinal cancers called peritoneal malignancy. That's actually my area of expertise. So I take care of patients who present with these advanced cancers as part of my clinical practice and all of my research is also focused in this area. How do patients come to you with this diagnosis? I mean, an advanced diagnosis like that. Is this patients that are often treated elsewhere or is this an initial diagnosis that they come unknown and sort of treatment naive? Yeah, it's interesting that you have that, you know, there are several avenues through which patients come to my clinic. One is referring physicians from outside, both surgeons and medical oncologists, when they see this type of manifestation of stage 4 metastatic disease, they do know that these patients need a different level of expertise. And they're all very familiar with the work I do and and what UCI Health has to offer for this kind of a complex metastatic disease management and that's one avenue through which patients get into our clinics. The second is through the clinical trials that we have. So patients hear about the clinical trials that we have available for patients with peratorial carommatosis, particularly the gastric cancer one, and I get referral from all over the country and patients seek. To come here, even they themselves initiate that conversation with their physicians knowing that this work is being done. So that's another form as to how patients end up in our clinic. So those are like two major ways that we get patients from outside of course internal referral continues to happen, but externally that's how um patients come to us. You are working on some revolutionary tools to change the management of gastric cancer. Start with how you're identifying these patients at the earliest stage. So that's a, that's a challenge that needs to be solved, right? So the reason why we approach this, the way that we are currently approaching in terms of figuring out as to how to help patients with gastric cancer, is by truly understanding what actually results in the poor outcomes. One of the main challenges with gastric cancer is that it has a very high incidence of cancer spreading to the abdominal cavity, peritoneal carcinomatosis, 30 to 50%. Unfortunately, it doesn't get diagnosed at the initial stages of presentation because it is image occult. So the typical CT scans that we do to identify this problem doesn't necessarily pick it up, and that's actually one of the biggest challenges. And subsequently, the recommendations to do a staging laparoscopy to see such a manifestation is present in the patient is not often done. The compliance with staging laparoscopy for patients with gastric cancer is about 30%. So it puts the patients at a risk, a huge risk of under staging and undertreatment. And we know that the treatment for patients with stage 4 versus stage 3 is radically different, so properly staging them is important, but that's a challenge. It's a practical challenge when you think about the logistical issues and, and many, many issues associated with this, you really have to get to the root of the problem. So what we know is every patient with gastric cancer would have gotten a CT scan. The question is, can we use the CT scan using artificial intelligence and deep machine learning to predict the risk or the presence of this advanced metastatic disease at the time of their initial presentation? Once we are able to accurately predict that, now we can change their treatment because in my clinic I'll end up seeing a patient with gastric cancer who was diagnosed with gastric cancer 3 months ago. They have a CT scan. They have gotten 3 months of chemotherapy. Now it is up to me to decide did they actually have stage 4 disease or did they only present with stage 3 or stage 2 disease at the time of initial presentation. Now how do I reconcile that? I can't go back in time. But what I have is access to that CT scan. So I can use that CT scan to figure out did this patient actually have this disease, stage 4 disease at the time of the initial presentation, which is a tool that we have developed right now. So we have been able to successfully use deep machine learning, artificial intelligence. It's actually the most sophisticated form of machine learning, convolutional neural network. To come up with a way to look at the primary tumor on the CT scan and predict what is the risk or presence of carcinomatosis in this patient based on that CT scan, and we have a 98% accuracy on the initial model that was developed at UCI and validated on UCI patients. And what we have been able to do through our international collaborations is now we have been able to validate this in a completely separate cohort of patients from the National Cancer Institute of Ukraine. And we are now studying this in an Asian cohort of patients from South Korea because we have collaborations with the Yonsei University, so we have almost 2000 patients, Asian patients that we will be testing this model on. So how, how many years did it take or how long did it take to develop this model and then which patient cohort did you use? Were there patient, were these patients at UCI Health or sounds like this is an international collaborative. Did you take patients from sort of all demographics? Great question. So it has taken us about 2 years to create that initial model, validate this model, and now we are moving into the international cohort of patients. So the initial model was developed entirely based off of UCI patients. And when you look at UCI, one of the. The great advantages that we have with UCI is that we have a diverse group of patients. So when you look at the initial cohort of patients, 1/3 are Asians, 1/3 are Hispanics, and 1/3 are non-Hispanic whites. So now you have a model that is developed on a diverse cohort of patients, and now you can validate that on an entirely Caucasian cohort that comes from Ukraine or an entirely Asian cohort that comes from South Korea and see if the model still holds strong. And that's actually the next step that we're doing and essentially making this um available as a clinical diagnostic tool. I'd love to go back a step actually, because it sounds like a lot of patients are presenting with gastric cancer. Or maybe you can help clarify for this for me at, at later stages of disease. So what are the symptoms of gastric cancer that a patient would normally experience? Are these found on screening, uh, endoscopies? Yeah, you know, unfortunately, gastric cancer symptoms are so non-specific. Who doesn't have a stomach upset? Who doesn't take a Pepcid or a Mylanta just to take care of some of the symptoms that they have experienced. So it is actually almost insidious. Because it it masquerades as common gastric upset that patients and you and I normally experience, and there is no screening protocol to identify gastric cancer at its earliest stages. That's also the reason why we end up finding gastric cancers at later stages, one, because it's so non-specific, and two, there are no real screening for high risk population. You know, it's interesting so in head and neck cancer we have all different types. So vocal cord cancer, laryngeal cancer, which is what I treat patients often present very, very early because if your voice goes out for more than a couple weeks or a couple of months, you're very likely to go and seek care. So the majority of laryngeal cancer we see is presented at an early stage. Compare that to base of tongue cancer or tonsil cancer who doesn't have a sore throat, who doesn't have some maybe some some vague discomfort. Unfortunately, those patients can present with metastatic neck disease. So the neck lymph node is what they present with, which makes them automatically stage 3 or stage 4. And so similar challenges with how do you get those patients in earlier and who is appropriate to screen. Because of course the later stages are harder to treat. So interesting that you're having similar challenges in in your area. Yeah, you know, it's, it's analogous, right? You know, every one of us have certain cancers in which we experience these kind of challenges where patients always present during the later stages. There's also an opportunity for us to really think about. Population health and figure out although screening is not routinely done, is there a high risk population for which screening could be recommended or at least an initial workup and the threshold for escalating it to an endoscopy should be lower in certain high risk patients, you know, even if it is not a screening when a. And a young patient, particularly of Hispanic race ethnicity, presents with a gastric symptom, that patient should probably end up getting an endoscopy because of the higher risk of gastric cancer and higher risk of aggressive and advanced stage gastric cancer in the Hispanic race at Nico, and so is the nation's. Are there any dietary um behaviors that that increase the risk of disease or the risk factors a sort of standard alcohol and tobacco use, or are those risk factors for gastric cancer? Yeah, so the risk factors for gastric cancer, particularly the young onset gastric cancer, is essentially multi-dimensional and multi-layered, right? So there are so many factors and the interplay between the factors really influences that aggressive manifestation. There may be some genetic predisposition as well and how with that genetic predisposition, patients respond to the usual triggers, even a HPyLOA infection. HPBOA infection, which is considered to be one of the common risk factors for gastric cancer, the response to HPyLRA infection might be different in different groups of people. So there is so much to study when it comes to what incites that initial pathogenesis of gastric cancer, the initial beginning of cancer in any human might be so many factors that interplay and influence the presentation. A lot of people are talking about artificial intelligence. A lot of people are afraid of it. There's plenty of implications for healthcare. Talk to us a little bit about like 10 years ago, 5 years ago, 2 years ago, how is this gonna change the game? Yes. See, artificial intelligence, when we think about artificial intelligence, it's an, it's an all encompassing term. We gotta ask which artificial intelligence model are we talking about? So the large language models, the chat GPTs and the deeps are different. So that is a large language model, which kind of crawls on the internet, finds what the information is and brings up a summary for you. Whereas these type of diagnostic tools is where I really feel artificial intelligence is going to drive the health care forward, where physicians like me and others where you already know what information, so it's a hypothesis driven artificial intelligence diagnostic tool, you're not at the mercy of what this tool is going to say or do. You are teaching the tool and then eventually developing the tool to the accuracy that is needed for clinical applications. So it's not random. You know what this tool should do, and you are designing the tool to help manage patients better more than what is currently available. So this is where I do see artificial intelligence to really add to the health care kind of delivery that we have and continue to keep improving both the diagnostic and precision tools that we have. Let's talk a little bit about treatment. So, diagnostic tools is one part of it. What kind of interesting things are you doing on the treatment side? Yes, you know, this gives me an opportunity to talk about my passion, which is one of the trials that I'm doing, right? So we have to work on all aspects of taking care of a patient, right? Yes, we understand this problem. Of high incidence of carcinomatosis exists and the survival for these patients is 6 to 12 months, 6 to 12 months. When you have a 30 year old with a 2 year old child that comes into your clinic and they have a stage 4 cancer and your. Going to have to deliver the news that unfortunately the current standard of care will give you a survival of about a few months. It's not acceptable, not at all acceptable. So we need to be doing something different. And that is one of the reasons why I started this clinical trial at UCI, which is called a stop gap. Now stopgap, the, the main objective of this trial is to ask the question, in addition to what we give through the IV, which is a systemic treatment, can we add regional treatments, meaning putting chemotherapy directly into the abdominal cavity. And see if that makes a difference in these patients' outcomes and add surgery in a specifically selected group of patients, you know, the appropriately selected patients. Now you're adding one more element to that regional treatment and see if that is going to change the outcomes in these patients and we have been very successful in enrolling patients in this particular clinical trial, and I do think it is going to change the treatment paradigm over the next few years and the success of the trial that we have done at UCI. Has led to a national randomized phase 23 clinical trial that is actually approved by NCI, and it is going to launch in the next few weeks. And the significance of this is that this is the first ever clinical trial for carcinomatosis that is actually approved by the National Cancer Institute because of the significance of the problem but also the opportunity to completely change the treatment paradigm in patients with stage 4 cancer. So take us into the operating room. What changes are, what, what do you do at the time of surgery and how long does it take? What effects do you see on the body when the patient's in the operating room and how do you monitor the patient? Great question. So with regards to how we think about this, this is, this is. Very systematic objective kind of process in how we get the patients to the operating room. So there are several steps. See, the first step is doing a diagnostic laparoscopy ascertaining the extent of the peritoneal disease burden, and that's when we place a a port like essentially. Intraperitoneal port to deliver chemotherapy. 3 months of this bidirectional IV IP chemotherapy reevaluation again by a diagnostic laparoscopy just to confirm that you can successfully remove all the cancer in that patient before subjecting them to the big operation. Then they actually have a formal cyto reduction for gastric, and there is a selection criteria they need to have what we call it as peritoneal cancer index, which ascertains the extent of the peritoneal disease burden to be 10 or lower for this particular study. It will be 7 or lower for the bigger study and also the ability to remove the primary tumor. So in the operating room, what I do is after a systematic evaluation at the time of the big operation, which is a gastrectomy surgery, we do perform whatever is needed to remove that cancer. Often it is a total gastrectomy, total gastrectomy with D2 lymph node dissectionomatectomy, and then additional cytoreductions, diaphragmatic peritonectomy, small bowel resections, and of course bilateral oophorectomies, whatever is needed to completely render the patient cancer free. So these are big operations. Usually it takes us about 8 to 10 hours because at the time of the surgical resection, we also combine heated intraperitoneal chemotherapy after removing all of the cancer. So the patients do get about 60 to 90 minutes of the heated chemotherapy, and then we finish off the operation with making the anastomosis and everything else. So it's a big operation, but what we have seen thus far is that 1/3 of the patients who have been in. Golden stopgap one have undergone this operation with great postoperative outcomes. We have not had a single anastomatic leak in these cohort of patients despite such an extensive operation after such extensive amount of chemotherapy, which also speaks to the patient selection and the and the interventions that we give have been tested and the skills that we have, the team that we have all plays a role in providing that great outcomes. That's amazing. What are some of the early findings you're seeing? Is it, I mean, clearly it's growing, so you're seeing some success for sure, you know, when we start using regional therapies, you know, these type of treatments, the first thing that we have to answer is safety and feasibility. Is it safe? Because we are combining intravenous therapy with now something that is given directly to the abdominal cavity. And what we have seen is unequivocally it is safe, it's feasible. Actually, patients feel better. With the intraperital therapy, most patients gain weight, right? So in terms of safety and feasibility, we have established safety and feasibility. Furthermore, with the operative procedures that we have done in the selected patients that we have done, we also have established postoperative safety. The acute surgical outcomes are actually very acceptable. It's actually better than some of the other non um stage 4 patients, right? So when we have no anastematic leaks, so we have established that safety and feasibility. Now, the next step would be to compare this to a systemic therapy alone. So that is why we need randomized control trials, you know, when you're doing these small, smaller single institution trials is to ascertain that safety, feasibility and then all these immediate outcomes. But at the end of the day, you have to compare it to standard of care and ask the question, is this approach better? And that is what the stop gap to, the randomized phase 23 clinical trial would allow us to do. I know it's still somewhat new. You're still working on it. Are you seeing impacts on patients' quality of life, survivability, or is it too early? Yeah, you know, quality of life during this approaches of treatment, we, we do have a quality of life questionnaire that we do. And so so far it's been great, and we will do a full analysis of the quality of life of patients during the intraperitoneal treatment and after surgical intervention as well, because we need to see both and figure out as to how their quality of life. But as a physician who is so closely attached to these patients, I can tell that their quality of life is definitely preserved or even better. In terms of survival results, it's too early. I have not analyzed the survival results overall for the study. We need about one year from the last patient enrolled for us to actually look at the survival data. Um, and so we will be doing that. I'm, I'm actually enrolled the patient today. We are. Coming close to closing that stopgap one trial here and so maybe next year in one of the ASCO meetings I could probably be presenting the data. I think we'll see you there. Yeah, you know, it's, I can just see from the look on your face how excited this makes you and, and I sort of reflect back to medical school for me, which was I guess about 20 years ago. I don't think I learned it much of this. So it sounds like the field has really changed quite a bit. What, what drove you, I mean, to, to be a leader in this field to seek these changes out and, and how did you, yeah, how did you even end up in this space? Yeah, as much as it's a cliche, I think it's destiny, right? Um, it's interesting that that we think about it, you know, when we reflect back on our life and say, how did I end up here? What made me now a peritoneal malignancy expert, you know, serendipitously during my residency, I had exposure to doing. Cases and I was working with a surgical oncologist who did these procedures. I was completely exposed to this. And then during my fellowship, this is one of the major part of our training is to be able to do and manage patients with peritoneal carcinomatosis and lo and behold, my first job that was offered to me, the request was, would you be willing to develop a program? And when I started doing this, you know, the connection is so immense with your patients, you know, eventually you find yourself just engrossed by the gratitude and the relationships that you build in managing patients with stage 4 disease and ask yourself, what better way to actually drive your career, you know, what better way to show up and contribute? Is, is because Sunil, you know, when we think about this, this is an absolute survival threat. Anytime a patient walks into my clinic, I am going to have to work with them to give some clarity. About what they're dealing with and still give them hope. Give them hope and that's Important, you know, as much as it is a devastating diagnosis to be discussing, I want them to leave with a sense of hope. And the sense of hope cannot come by just talking about standard of care. The sense of hope comes from the fact that you are working so hard to figure out better solutions through novel research, translational work, and through clinical trials. To me that's important. I mean, I'm not my intention is not to become a leader in the field. My intention is to serve my patients to the best way possible and never let go of the desire to continue to keep improving their outcomes. So I know you thrive on making an impact in your field and other fields. If you could speak to your younger self or other physicians just starting out, what advice would you give? Be curious. Be open minded. And learn from your patients. Never stop learning from your patients. Every patient is a textbook. And they're asking you critical questions. When you sit there and for a moment, ask what question is this patient asking me? And then take a moment to reflect on it and say, how can I make this better? How can I answer this question? A patient can come into your clinic and say, how do you know if I have recurrence or not? How can you be so sure? Then you're like, no, I can't be sure. Yes, CT scan is not good. You cannot tell me for 100% whether you have a recurrence or not. I should be studying this better. So is there a liquid biopsy that I should be studying? Then study. Never stop being curious and never stop learning from your patients. I just think that our patients are the ones that give us the energy to improve every day and just showing up with that mindset would help any physician to push their field forward. Thank you both so much for your insights today. This has been Physician Huddle by UCI Health. Thank you for joining us. Yeah, thank you. This was an episode of the Physician Huddle podcast by UCI Health, produced by Brett Shaheen, Angelica Yugubi, and Victor Ting. For more episodes, information on clinical trials at UCI Health, or to refer a patient, review the show notes or visit clinical connection. UIhealth.org. Created by
