iwNHL 2025 | Classification systems for hematologic malignancies: the need to evolve and the role of AI

I think we’re going to move into an era where the classification systems for cancers are going to need to evolve, and specifically in diseases like lymphoma, where historically we’ve used the microscope, or more recently we’ve used gene expression profiling, mutation profiling to define subsets of disease that genomically or histologically look distinct from each other. But the problem is that it’s really not moved the needle in the ways that we’d hope within clinical trials within drug development and there really are not great examples of drugs approved based on a specific genetic abnormality. In other diseases, for example, lung cancer, if you have a particular mutation and you get a drug that targets that mutation, that can work wonderfully, much better than chemotherapy, and in hematologic malignancies, despite being in the vanguard of the early days for how we treat people, how we diagnose people, we’ve really fallen behind in terms of being able to classify patients based on what drug will work and have a prognostic tool that’s usable in the clinic that can actually tell me my patient will benefit from this treatment and not from that treatment. I do think AI tools will help us to go back and look at ongoing and recently completed clinical trials to analyze responders versus non-responders in a way that we can better understand patterns that are reductionist models of one mutation or a few genes in a gene expression profile might miss. And using AI to understand complicated patterns that are maybe hundreds of parameters to say this patient is in this group and not in that group could be very useful in terms of deciding which treatment for which patient. And I think we’ll see that in the coming years.

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