Novo Nordisk got a major boost in March when the FDA approved its GLP-1 diabetes and weight loss drug Wegovy to also prevent heart attacks and strokes in patients with certain risk factors. The decision opened the market up to more than six million people and broadened Medicare coverage for the blockbuster drug, now on track to become one of the best-sellers of 2024.
And that may only be the tip of the iceberg, according to a sweeping AI study from Dandelion Health, which identified an additional 44 million lower-risk cardiovascular patients who could benefit from a GLP-1 treatment, hinting at the potential to greatly expand insurance coverage.
Putting AI to work
Dandelion leaders ran the study as a proof-of-concept test of their real-world data platform, which uses commercial and academic AI programs to analyze patient information obtained through revenue-sharing agreements with health systems, including Sanford Health, Sharp HealthCare and Texas Health Resources. The platform could be a faster and less expensive way to deliver results than traditional clinical trials. Studies could take weeks, not years, and incorporate larger and more diverse patient groups, according to the company’s co-founder and CEO, Elliott Green.
“No one ever does preventative trials because they're just too big,” he said. “But this AI proof-of-concept started to show that there's a path forward to being able to achieve that.”
Mining real-world data
For the observational study, Dandelion examined whether GLP-1 use might reduce the risk of major adverse cardiac events such as heart attack or stroke in overweight and obese patients. The study used inclusion criteria similar to the Novo Nordisk-sponsored trial that helped convince the FDA that Wegovy had heart benefits. But, unlike SELECT, Dandelion’s patient group didn’t have severe preexisting cardiovascular disease.
Researchers applied an AI algorithm from Pheiron trained to spot cardiovascular risk indicators on patient electrocardiograms. The AI assigned patients a risk score and compared predicted risk for heart attacks and strokes in GLP-1 users versus non-users. The results expanded on the findings of Novo’s trial, which found Wegovy reduced the risk of major cardiac events in the study group as much as 20%.
Dandelion’s study, seven times the size of Novo’s, determined that people in the lower-risk group who took GLP-1 drugs had 15% to 20% lower risk scores for major adverse cardiovascular events after taking the drug for three years. Treating these additional patients could head off 17,300 heart attacks and 16,700 strokes a year, according to study authors.
“With further validation, this AI-driven approach could potentially unlock a new biomarker or surrogate endpoint that measures [major adverse cardiovascular event] risk, enabling shorter and smaller clinical trials by predicting cardiovascular outcomes earlier without waiting for [adverse events] to occur,” they stated.
Speeding research
Real-world data-based AI could validate GLP-1s for broader applications, from neurological and liver diseases to addiction and arthritis.
Green said Dandelion already has another GLP-1 study in the works, using AI to assess muscle changes on patient abdominal CT scans to determine whether GLP-1 drugs also cause muscle to wither during weight loss. The platform could also help pharma companies explore and validate label expansion opportunities and perform other types of research. Dandelion is working with several companies, Green said.
“Clinical trials are very focused. You come in with a very specific question, and you get a very specific patient population,” Green said, noting that a broader approach is often too challenging due to time and cost constraints.
With AI, instead of looking at a limited area, similar to what you might see under a single street light, Green said, you can illuminate the whole neighborhood.
