Can AI and liquid biopsies take precision oncology to the next level?

Prof. Philippe Bedard, Clinical Director of the Cancer Genomics Program at the Princess Margaret Cancer Centre, Toronto, ON, Canada, was presented with the 2025 ESMO TAT Honorary Award during the Opening Session of the ESMO Targeted Anticancer Therapies Congress 2025 (Paris, 3–5 March) for his significant contributions to early-phase drug development. In addition to his clinical trial work, Bedard has been part of the American Association for Cancer Research project Genomics Evidence Neoplasia Information Exchange (GENIE) (Cancer Discov. 2017;7:818–831) since its conception in 2015, serving as its chair for 4 years (2021–2024). This international data-sharing consortium focuses on generating a publicly accessible evidence base for precision cancer medicine by integrating clinical-grade genomic data with outcomes data and now includes information from over 200,000 samples from patients treated at multiple institutions worldwide.

In your keynote lecture in Paris, you discussed how learnings from mutational testing and data-sharing initiatives such as GENIE have helped advance the field of precision oncology. How can cancer genomics and drug development be better bridged in oncology?

New methodologies could increase the wealth of information generated from trials and registries and help refine precision treatments. Currently, only a small minority of patients with cancer actually participate in clinical trials. Added to that, trials are complex, time consuming and involve a lot of inefficiencies. Natural language processing and large language models can be used not only to improve trial enrolment – by better integrating information from medical records – but also to characterise outcomes more efficiently and gain greater insights from patients outside of clinical trials. These technologies can also be used to expand the scale, and therefore impact, of large registries, including GENIE. Although the exploitation of rapid advances in artificial intelligence represents a considerable opportunity to improve clinical trial research, it comes with new challenges.

In addition, there is still a lot to understand in terms of why patients with targetable mutations do not respond and why some patients develop resistance. There is substantial research scope in the fast-moving area of liquid biopsy to better characterise cancers using minimally invasive approaches from blood samples. This will further our understanding of response and resistance dynamics and help to detect patients who relapse before it is clinically apparent.

In your research career, you have worked on many investigational agents both in first-in-human trials and subsequent phase trials. What do you consider to be your greatest achievements so far?

A notable involvement was the long-and-winding road that led to the approval of PI3K inhibitors. Investigations with the early pan-class I PI3K inhibitor, buparlisib, showed some signals of activity, but it was challenging to deliver because of toxicity (Clin Cancer Res. 2015;21:730–738). In the search for better-tolerated agents, I was involved in the phase I development of the alpha-selective PI3K inhibitor, alpelisib (JCO Precis Oncol. 2019:3:1–13), which is now used in the treatment of hormone receptor-positive, HER2-negative, locally advanced or metastatic breast cancer with a PIK3CA mutation.

Although development has since stopped, a genome-driven basket study with the selective β-sparing PI3K inhibitor, taselisib, helped to improve understanding of the activity, limitations and resistance mechanisms of using PI3K inhibitors as monotherapy to target PIK3CA-mutant tumours (Clin Cancer Res. 2021;27:447–459). I also contributed, along with many others, to the successful development of combination therapies containing inavolisib, a selective p110α inhibitor that promotes the degradation of the mutated p110α subunit of the PI3 complex (J Clin Oncol. 2024;42:3947−3956). Targeting HER2 in breast cancer with tucatinib has been another career highlight (JAMA Oncol. 2018;4:1214−1220; N Engl J Med. 2020;382:597−609).

What new anticancer agents hold most promise for the further advancement of precision oncology?

Advances in medicinal chemistry are offering new opportunities that did not exist when I began studying PI3K inhibitors over 15 years ago. Evolution in this area continues with the development of mutant-selective PI3K inhibitors to further circumvent the off-target toxicity that often limits tolerability and efficacy. New pan-KRAS and mutant selective inhibitors hold promise for patients with KRAS genomic alterations beyond KRAS G12C mutations, opening up new patient populations with targets that were previously considered undruggable.