AI, Genomic Testing Key to Advancing Precision Oncology: Vivek Subbiah, MD

Vivek Subbiah, MD, chief of early-phase drug development at Sarah Cannon Research Institute, considers cancer a genetic disease and genomics a diagnosis, emphasizing that somatic and germline genomic testing should be part of every cancer diagnosis. He also highlights the role of artificial intelligence (AI) in integrating genomic, proteomic, and transcriptomic data into actionable insights for clinicians.

He discussed these perspectives during the Patient-Centered Oncology Care conference last week in Nashville, Tennessee, where he participated in the panel "Access for Everyone: Using AI in Precision Oncology" alongside Alyssa Schatz, DrPH, MSW; John L. Villano, MD, PhD; Davey Daniel, MD; and Stephen Speicher, MD, MS.

This transcript has been lightly edited; captions were auto-generated.

Transcript

Precision oncology has been around for a while, but access remains uneven. What are the biggest barriers, and how can they be overcome?

I think we are living in one of the best of times in oncology, and also, we should say, to quote Charles Dickens, in the worst of times, as well, because we've sequenced the human genome. We have so many advances in genomically-targeted therapies, immunotherapies, agents that arm the immune system, and, also, antibody drug conjugates. We are looking into an era of PROTAC degraders and a lot of new drugs.

We are able to pick these needles out of haystacks, all these druggable driver fusions, like NTRK fusions, RET fusions. But the challenge we have in the real world is that we are picking the needle, but we are not threading the needle. The main challenges for threading the needle are huge therapeutic gaps and access barriers to standard of care therapy, standard of care precision medicines, standard of care research, and standard of care clinical trials.

Beyond standard of care, access to clinical trials; that remains a huge, huge, huge therapeutic gap here. Most of the genomics are being tested late in the late lines of therapy, not when patients are doing well. Again, at a therapeutically futile stage. Again, these are some of the barriers we have in the implementation of precision medicine in the real world.

How is AI helping clinicians integrate molecular profiling into routine care?

Given the volume of data and information we have from genomics, proteomics, and transcriptomics, it's humanly impossible to put all these pieces of information together.

I think that's why we depend on AI, that's defined by either machine learning or large language model systems, to integrate all this data so that it can give us pieces of actionable information so that we can help patients in front of us in clinic.

How does wider access to testing create value for providers and payers?

Again, my personal opinion is that cancer is a genetic disease, and genomics is a diagnosis. Somatic genomic testing and germline genomic testing should be a part of cancer diagnosis because, in the last 10 years, let's take, for instance, lung cancer. Ten years ago, we had only one drug that was targeted to an oncogenic driver. Right now, we have 10, 11, and 12. In the next decade, we're going to have at least 20 drugs to go after lung cancer.

Again, there is no way that we can test for each and every one of these genes, so we need to do comprehensive genomic analysis on not just lung cancer but every patient with cancer walking in front of us, because we may not have drugs to go after them today, but we may develop drugs tomorrow.

Again, we are seeing unprecedented advances in genomically targeted therapies, immunotherapies, and now we are entering the era of antibody drug conjugates. It's an exciting time to be in oncology, but in order to make sure that the patients benefit from these advances, we need to do the right testing.