Utah hospitals tapped to develop new algorithms that flag cancer-causing genes

Intermountain, U. researchers get $3.8 million from feds to couple electronic patient record systems with computer physician support to improve cancer detection.

(Scott Sommerdorf | The Salt Lake Tribune) Dana Lombardi poses for a photo at the University of Utah Medical Center, where she is a physician's assistant Friday September 1, 2017. Lombardi tested positive for the BRCA-1 gene mutation for breast cancer about four years ago.

Dana Lombardi’s life changed with one phone call from her mother’s cousin four years ago.

The relative recounted how she, all of her female cousins and her 32-year-old pregnant daughter had been diagnosed with breast cancer at some point. Given that Lombardi’s mother also had the disease, her mom’s cousin urged Lombardi to get tested soon as well.

The 39-year-old Salt Lake City resident immediately made appointments for herself and her sister. She tested positive for a breast cancer-causing gene, now undergoes routine screenings and has since removed her ovaries to further reduce her cancer risk.

“What if she had never called me?” Lombardi, now 43, asked. “I wouldn’t have even thought, other than my mother, to think outside my immediate family.”

“If there was a tool prompting me and my doctors to ask about my extended family,” she said, “maybe I would have put it together.”

Intermountain Healthcare and the University of Utah Health have received $3.8 million from the National Cancer Institute to create computer algorithms that scan digital medical records for the genetic markers of cancer. The systems will not only help detect patients likely to get cancer earlier but give physicians extra support with diagnoses and treatment options.

Lead researcher Kensaku Kawamoto said the systems under development could eventually give patients and doctors the equivalent of a digital second opinion.

“We want to prevent these tragedies where people have a family history of a disease but no medical evidence,” said  Kawamoto, also an assistant professor of biomedical informatics at the U. 

The programs will flag notes, data points and other clues in a patient’s digital medical file for signs that might indicate the need for cancer screenings or follow-up appointments, said Kawamoto. The software will focus on breast and colorectal cancer risk — types of cancer with a strong hereditary factor — but researchers said it could expand to other forms of cancer and other diseases.

Kawamoto and his U. colleague Guilherme Del Fiol are already working with the Huntsman Cancer Institute to find keywords and other indicators in a patient’s file that might be flagged.

Family history is still one of the best tools for identifying people with an increased risk for cancer, said Wendy Kohlmann, a Huntsman Cancer Institute genetic counselor involved with the project.

“This will help us do our jobs and ensure the right people are getting to physicians and genetic counselors,” Kohlmann said. “We can focus our time on seeing these families quickly.”

Kawamoto and Del Fiol will create the systems over the next two years, then test them in U. Health clinics and hospitals. Intermountain will then integrate the software to its electronic medical record system, which is different from the U.’s.

Typically, this kind of technology only works within an institution and other networks have to create or find their own, said Scott Narus, Intermountain’s medical informatics director. By developing a platform that can work across different hospital systems, Narus said it will allow for quicker implementation at other sites.

The team hopes the system is fully operational within five years, said Narus.

“We have an obligation,” he said, “to provide the best evidence-based medicine to our shared community.”