Pfizer, Merck, J&J to release cancer drug data

Think a friend or colleague should be getting this newsletter? Share this link with them to sign up.

Good afternoon! This is Annika, and I’ll be on the ground at the American Society of Clinical Oncology annual meeting in Chicago in the coming days. 

So will thousands of cancer researchers and clinicians, who expect to see encouraging new science and data that could advance cancer treatments, health equity and patient care. 

More than 5,000 research abstracts will be presented or published at ASCO, which starts on Friday and lasts until Tuesday. There will be data on existing drugs from pharmaceutical companies, experimental treatments in early to mid-stage testing from biotech companies and even AI tools. 

I’ll be focusing on some of the larger names in the industry. Here’s some of the data I’m looking at:

• Merck and Moderna will present three-year data from a phase two study on their experimental vaccine, used in combination with the therapy Keytruda, in patients with severe versions of the most deadly form of skin cancer.
• Pfizer‘s antibody-drug conjugates, or ADCs: The company will present data on some medications acquired through its $43 billion buyout of Seagen last year. That includes an experimental ADC called sigvotatug vedotin in a type of lung cancer, and another already approved treatment, Adcetris, for a common type of blood cancer.
• Johnson & Johnson will present mid-stage and late-stage data on a more convenient form of its targeted antibody drug, Rybrevant, in patients with a type of lung cancer.
• J&J has also released early-stage data on a radiopharmaceutical drug, which showed signs of efficacy in prostate cancer patients. But four trial participants died.
• Merck and its partner Kelun-Biotech have released positive phase three trial data on an antibody-drug conjugate in a type of breast cancer. 

I plan on writing up some data, along with a wrap-up or two after the conference, so stay tuned for my coverage. If you see me at ASCO this weekend, don’t hesitate to say hello! 

Feel free to send any tips, suggestions, story ideas and data to Annika at [email protected].

Software vendor Epic Systems is trying to make it easier for health systems to cut through the artificial intelligence noise. 

The company last week released a free new tool set called the AI Trust and Assurance Suite, which health systems can use to evaluate the performance of an AI model that integrates with an electronic health record, or EHR.

An EHR is an electronic version of a patient’s medical history. Epic is perhaps the best known vendor in the space, as it houses medical records for more than 300 million people. 

Epic released the suite as an open-source tool on the popular developer platform GitHub. That means it’s available for anyone in the world to use. The suite can help health systems assess an AI model’s equity, fairness and performance, as well as how it is affecting outcomes for patients, said Corey Miller, vice president of research and development at Epic.

For instance, a health system could use the suite to test several different models and determine which one works best for that community’s local patient population. The tool set can also carry out a “fairness audit,” which can look for bias across race, sex and age within a group of patients, Miller said. 

The suite is specifically designed to evaluate health-care AI, and Miller said it will work regardless of whether a health system is using Epic’s EHR. The tools are not specifically built for Epic’s AI models, he added.

“We saw this as an opportunity to create something that wasn’t out there today,” Miller told CNBC in an interview. 

The release of Epic’s suite comes as the health-care sector has been reckoning with how to establish guard rails and best practices around AI. Several organizations like the Coalition for Health AI, Microsoft’s Trustworthy & Responsible AI Network and the Health AI Partnership have been created with these objectives in mind, but there are no hard and fast rules about how to use the technology.

In the interim, Miller said he thinks Epic’s suite will make it easier for health systems to look at how an AI model performs with their local populations, particularly within rural community hospitals that might not have data scientists on staff.

He said the suite has been in the works since late last year. It evaluates models based on existing standards developed by “healthcare systems, health IT software developers, third-party experts, and the government,” according to the post on GitHub. Epic plans to update the tool as best practices evolve, Miller said. 

Epic’s suite currently works with binary classification models, or typical predictive models, and Miller said generative AI models will be up next.  

“We expect to make this tool something that can really look at that whole spectrum of AI,” Miller said.  

Feel free to send any tips, suggestions, story ideas and data to Ashley at [email protected].

Reimbursement is one thing AI programs can’t generate yet

When I had my mammogram earlier this month, I was puzzled by having to pay cash for an AI-enhanced reading of my scan. I opted to pay, but wondered why my insurer wouldn’t cover it.

It turns out the majority of new FDA-approved AI radiology and diagnostic programs coming to market have yet to get billing codes for insurance reimbursement. Early forays into computer-assisted mammograms in the late 1990s proved to be no more effective than conventional screenings.

This time around, medical societies along with government and private insurance plans are taking a more cautious approach toward signing off on paying for new technology. 

Feel free to send any tips, suggestions, story ideas and data to Bertha at [email protected].