I'm a cardiothoracic radiologist at Vanderbilt. My question's for Florian. So what's so great about CBL for people who haven't used it is that it's this one-time CT scan. It gives you this risk prediction. You don't need to do anything. You don't need any demographic data. As you mentioned, there is an opportunity to incorporate demographic data as well and perhaps augment it. I was thinking when you showed the slide of baseline screens and as we think about expanding screening to other high-risk populations that maybe we aren't defining yet, is there an opportunity for CBL to become, or a different algorithm to be longitudinal and that it's incorporating data from multiple CT time points? And then would that change? Could maybe your screening interval not necessarily be determined by your baseline, but could this be something that changed over time as we see risk sort of change as well? Thank you. I think both is possible. I think you could use it in the current iteration to assess the risk on the scan that you have in front of you to make decisions about the future. The option to integrate serial observations, meaning multiple CT scans, is something our engineering colleagues are working on and I think that can only stand to improve the performance because you have more information. We all know that in thoracic imaging, change over time is as important as the observation itself, whatever you're seeing. So yes and yes. The next step for us is figuring out what are relevant levels or what score would you need to, how would a score affect what you do next? And we don't have the answer yet, but we're getting very close. In the back of the room? Yeah. Hey, this is for Peter. Peter, what does that conversation look like? Let's assume for a second that one of the lung cancer specific biomarkers makes the grade. A few things that I've heard from pretty thoughtful people are, hey man, if I have this test and it's really sensitive, maybe I don't go get my CT scan. You mentioned that this could increase uptake and adherence. What does that conversation look like at the patient level in the shared decision-making business? Yeah, I think that they're all great questions that will be influenced by the accuracy of these tests. I think at this point in time, I would be thinking of any test developed in the currently screen-eligible population as not being the first test in somebody who is willing to undergo a low-dose CT scan, has access to a low-dose CT scan, is covered and is going to follow. That is still standard of care unless you get your home-run biomarker out there. So we know uptake is low, we know adherence is low, and if some of that is related to fear of getting a scan or poor access to high-quality lung cancer screening program, then perhaps we can pull people in by getting that blood test done. That's another variable that needs to be proven. Can we increase uptake and adherence by using these biomarkers in practice? I'm Allison Chang, I'm a medical oncology fellow from Mass General. This question's for Dr. Mazzone about the circulating biomarkers, which is really exciting technology, I think especially for earlier today hearing about our global colleagues without as much CT scan or access. But some of the other commentary I've heard about these tests is that it's not only a screening test, but can also be a prognostic marker potentially because we're identifying cancers that have already leached some of their tumor DNA into the bloodstream. And I know at our center, we've actually started to see some patients who are obtaining some of these tests direct to consumer and are presenting to us with positive tests. And I'm curious, based on your experience with the biology of these assays, as thoracic oncologists and surgeons, should we be thinking about these patients in a different risk category as well? Do you think that this could be something to take into consideration when considering, for example, adjuvant therapy? Yeah, I think that that's a very important question that still needs to be answered. So as we talk about what accuracy is necessary, we have to learn a little bit more about what is a false negative. Is a false negative all of the ground glass nodules that we really don't want to be resecting anyway? And maybe that lower sensitivity is actually a better performing test than it is just on the surface. As of now, I don't think we have enough information. Biologically speaking, it makes sense. Cells that are turning over more, spilling their contents into the bloodstream are more likely to be detected by a lot of these tests. So they may be more aggressive. You could argue at the other end of the spectrum, are they so aggressive that they're past their point of being able to cure them as well? Again, something we just don't know, but there are potential applications for these tests over time. To the back of the room. Thanks, Ella. I'm Christine Berg. I was the National Cancer Institute PI for the NLST. And for the purpose of this comment, I was the NCI PI for the Prostate Lung Colorectal Ovarian Cancer Screening Trial. And wanted to comment on Dr. Lee's presentation on the use of AI for chest X-rays. In my opinion, there were quite a number of flaws with the PLCO. And for the lung cancer arm in particular, we had at least two major errors that could have contributed to us not finding a mortality reduction. One was only the chest X-rays were only PA, posterior anterior, and there were no lateral films done. And that was perpetuated into the NLST except at Hopkins where they made a mistake and did PA and lateral. Good for them. We also analyzed the PLCO results like 12 years to be consistent with the prostate results. And at six years, there was a suggestion that there was actually a mortality reduction in the lung arm. So I am personally enthusiastic about considering chest X-ray use in low and middle income countries where access to low-dose CT might be few and far between. So thank you for your work. Hi, Nicole Rankin from the University of Melbourne in Australia. My question is for Professor Tamghamaji. Thank you, Martin, for a great presentation. I'm interested in the patient satisfaction component of thinking about quality indicators. And I guess when we look at the outcomes from other cancer screening programs, satisfaction is just 96% of people's ceiling rate satisfied with the test. And how do we move instead with going forward with quality indicators that are gonna look at process outcomes like timeliness and willingness and intention to rescreen and offers of smoking cessation? And is there any intention for the Early Detection Committee to start developing patient experience modules that move beyond our existing suite of patient experience? Question is that are used in a very different context for trials and for outcomes. Like where do you see the work going? The thinking that I see that the committee had was that you don't have to have a satisfied participant for screening to work to reduce mortality. And the threshold was fairly high. 80% had to be in agreement, a strong agreement to pass the indicator. And so that we might have been just a little bit below. I was involved with the Ontario Health Cancer Care, Ontario screening pilot. And there they valued patient satisfaction, stakeholder satisfaction very highly. And they developed and validated many multiple questionnaires that actually tested everybody involved through satisfaction. And the thinking there is that adherence, as you mentioned, will be better if the individual is satisfied and they'll come back. Fortunately, in the Cancer Care Ontario, the satisfaction was very high. And so the people had a very high adherence at all levels above 85%. And so that was very successful. In terms of the expert panel, the Islaic Working Group, we have at this time not thought of pursuing that any further. And I think that we probably have other interesting themes to study rather than satisfaction. I did notice that the Australia guideline was very rich in satisfaction, I believe, questions. And so that for Australia, they have a different kind of philosophy in thinking about it than North Americans do. But I'm sad to say that I don't think that we're gonna be pursuing it to the extent that you probably will be pursuing it. Thank you. So we have two people left. We're a little tight on time. So if you could make your questions brief, we'd like to get to both of them. Thank you. Okay, this is just in regards, this is Edith Marome, a chest radiologist from Israel. This is just a comment about the AI and the chest radiographs for screening. I just would be afraid that we'd leave the room thinking that would be a solution when we, in order to do that, we would have to conduct a prospective study actually showing a decrease in mortality. And the other thing is that unlike CT, that our usual use, the use of AI on chest radiographs to see pulmonary nodules is still extremely poor. We implement that in our hospital because we have no interpretations for chest radiographs because of the lack of radiologists. This is not out-of-the-box solution. In order for it to work, one has to clean the data, take out all the ones with chest tubes, all the known oncology patients, and even then, after cleaning the data with the work of an IT team, you're left with false positive rates which are still extremely high, like 80% nodules which are not actually pulmonary nodules, still requiring a chest radiologist to sit and sort them out. So I just wanted to know if you think there would be a prospective study looking at chest radiographs with AI. I think we need to do that, yeah. I think with improvements in AI with all the deep learning and continuous learning, it might be a very good tool. But like you said, the jury is still out. We need to do randomized controlled trials. There are some trials that have been done in Korea. There are several commercially available AI programs. So that needs to be evaluated. I agree with you. Thanks, last question. There's time. Sarah Gandahari, pulmonologist from Los Angeles. For programs that are growing and developing in lung cancer screening, given that there is such a difference in terms of how imaging are integrated into the systems or different platforms, and there's still so few patients that are getting screened, I'm often conflicted as to what should be our effort focused on, on increasing the number and volume, or moving towards exploring biomarkers and inclusion of AI. And I wonder at the moment, what would be your collective recommendations for folks like me who are in the position to develop their program and expand it? What direction should we go? Should we focus on biomarkers? Should we focus on inclusion of AI? Or should we just crunch numbers and try to get as many people screened as possible? I'm gonna direct that question to Dr. Mazzella. I would do the final. I would put your efforts into bread and butter, getting people in and screened with standard of care load OCT screening. The structure you put in place to do that well will help as these other tools become more proven, help you to implement them into practice and shift your screening program over time. Thank you. Can I add something? If you have the luxury of extra time, like after you're done doing standard of care to the best of your ability, collect annotated data. None of these efforts can move forward without annotated data, tons of it. So the more you can do that prospectively, the better. Thank you.

lung cancer screening

CT scans

AI in healthcare

biomarkers

patient satisfaction