BTOG 2026 | Managing incidental findings in lung cancer screening with AI and biomarkers

Yeah, sure. So the three key points that BTOG asked us to put into our talk, for me, the ones I stated were the incidental findings have the potential to cause harm in screening programmes. But equally, on the flip side of that, some of them, it may be beneficial to report and there’s potential benefit there. And the key really is knowing which is which and also what to do with them when we find them. And I think another key point that came through with some of the work I presented was about having integrated pathways downstream of what to do with those incidental findings and how we link up with primary care to make sure the findings are actioned and we’re not just giving participants diagnoses and then not doing anything about them so AI-based risk assessments are definitely the way forward because at the moment lung cancer screening involves a telephone triage, you know, person to person asking, working through various questions, now that can easily be done digitally or remotely, and there is an app in development to do that, which would, I mean, make life a lot easier for a lot of participants, it would remove some of the stigma around answering questions about your smoking history to a person in front of you who you might think might judge you, there is obviously issues with digital exclusion and particularly with sort of older age groups who may not be as comfortable using apps or what have you. But yeah, that’s certainly in development and nearly ready, I think, to be rolled out. In terms of biomarkers, so a risk assessment for discriminating, what was it, kind of indolent tumours versus I don’t think we know, I don’t think we have the kind of question or individual participant-based characteristics that will help differentiate that, so I don’t think a risk score could do that, biomarkers could, but again, we’re not there yet, I don’t think, so you know, biomarker development and testing is quite complex because there’s an awful lot of places we can put them within lung cancer screening, in terms of potential for diagnostics, potential for risk assessment. And even within risk assessment, are we identifying high-risk people and bringing them into screening? Or are we identifying low-risk people and sort of pushing them out of screening to reduce the potential for harm in that group? So designing biomarker studies is incredibly complicated. Oh, and then equally, I guess the important bit is a biomarker at the point of having had a CT scan where you found something, can a biomarker tell us, is this an important nodule? Is it likely to be cancer? If it’s cancer, is it going to be an indolent one that we can watch for a while or does it need, you know, immediate, uh, interventional or, you know, investigation, so, and we’re definitely, there’s a lot of work going on out there around genomic markers, proteomic markers, epigenetic markers, but none of them are quite there yet, um, and there’s a lot of big grants, I think, being worked up to look at those. So we need to do that first, the discovery translational science bit. And then we also need to think about how we implement that in clinical screening programmes in terms of the resource, the time, the staffing, and the personnel, and then also how we communicate that with participants, because risk is a really complicated thing to communicate, so yeah, there’s a programme of work to be done there, I mean, it’s a really exciting place to be as an academic, and I do think, you know, 10-15 years down the line, lung cancer screening will definitely benefit from these tools, but yeah, not quite yet.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.