AI Expands Clinical Trial Access, Advances Drug Repurposing Efforts: Vivek Subbiah, MD

Artificial intelligence (AI) can democratize clinical trial enrollment by analyzing electronic medical records (EMRs) to identify eligible patients across community and academic settings, expanding access and improving outcomes, says Vivek Subbiah, MD, chief of early-phase drug development at Sarah Cannon Research Institute, in part 2 of an interview at the Patient-Centered Oncology Care® conference in Nashville, Tennessee.

He adds that AI supports drug repurposing by uncovering new actionable pathways and identifying patient subsets who may benefit from therapies that previously failed in trials.

Watch part 1 to learn more about how AI is guiding precision oncology and improving patient care.

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

Transcript

How can AI help democratize clinical trial enrollment, and what impact does this have on patient outcomes?

Clinical trials have been very, very challenging because they've been mainly centered in niche, boutique academic centers and in the community through our community networks. Again, many of the community practices do not have access to clinical trials. I think AI can democratize, hopefully, clinical trials by opening up and going through [EMRs] and seeing which patients would be eligible for clinical trials in an automated fashion.

To democratize clinical trials across the board, to do it at scale across the entire US, I think it would take an AI-based system that goes through the EMR so that it can screen the patients so that they can be matched to the right clinical trials at the right time.

With AI changing how trials are designed, what new possibilities are emerging for treating previously ineligible patients?

I think what we used to think about is actionable vs nonactionable. With AI, we are discovering the nonactionable genes as actionable because AI is coming up with new pathways. It's uncovering and deciphering new pathways that we have drugs for.

That leads us to the next part: drug repurposing. We may have drugs, and we know that they target certain pathways. AI can discover new targets for the same drugs; that is drug repurposing. Another thing that is exciting is that we have many drugs that have failed in clinical trials, but there is a subset of patients that do respond to those [drugs]. I think AI-based algorithms can detect those patients who respond so that we can use those AI algorithms to benefit patients in the future.

I can give an example for a recent TROP2 ADC [antibody-drug conjugate]. The clinical trial with the TROP2 ADC, the phase 3 trial, was a failure in all comers. Later, in a retrospective review, they figured out that this TROP2-based NMR [nuclear membrane ratio] scoring system, called the NMR QCS [qualitative continuous scoring] system, can predict patients who benefit from this antibody-drug conjugate.

If it pans out and it's validated in a prospective clinical trial, I think this can set a benchmark for future drug repurposing and rescuing failed drugs. Again, these are not failed drugs; these are failed trials. We can rescue the drugs so that we can benefit more patients with these precision medicines.