• Posted May 28, 2025

AI Helps ID Candidates For Targeted Breast Cancer Treatment

Thanks to artificial intelligence (AI), more women might soon benefit from targeted breast cancer treatment, a new study says.

AI can help identify women who might be helped by cancer therapies that target HER2, a protein that spurs on the growth of tumor cells, according to results scheduled to be presented at an upcoming  meeting of the American Society of Clinical Oncology (ASCO).

Up to now, only about 20% of breast cancers carried high enough levels of HER2 to make women eligible for these treatments, the American Cancer Society says.

But the new AI could help extend HER2-targeted therapy to another 65% of women by helping doctors detect low or ultralow levels of the protein, lead researcher Dr. Marina De Brot, head pathologist at the A.C. Comargo Cancer Center in São Paolo, Brazil, said in a news release.

“Some of these tumors could be treated with HER2-targeted drugs, but only if we detect their HER2 expression levels,” De Brot said.

At least 55% of breast cancers contain low levels of HER2 and another 10% contain ultra-low levels, researchers said in background notes.

These tumors might respond to monoclonal antibody drugs that target HER2, as well as newer antibody-drug conjugates that link antibody treatment with chemotherapy. 

In those conjugate drugs, monoclonal antibodies use the HER2 protein on cancers to steer chemo directly to tumor cells.

“Our study provides the first multinational evidence that artificial intelligence can help close a critical diagnostic gap and open the door to new therapies like antibody-drug conjugates for a majority of patients who, until recently, had not been offered these options,” De Brot added.

For the study, researchers used an existing AI called ComPath to help 105 pathologists from 10 countries in Asia and South America with their HER2 scoring of breast cancer biopsies.

Over five sessions, the pathologists performed a total of 1,940 readings, with 1 out of 3 performed using AI assistance.

Even among experienced pathologists, about 1 in 3 breast tumors with ultralow levels of HER2 can be mistakenly labeled as HER2-negative, researchers said. Those cases miss out on potentially helpful HER2-targeted therapy.

Results show that AI improved pathologists’ ability to accurately identify HER2 status by nearly 22%, from about 67% accuracy to just under 89%.

AI assistance also reduced by more than 25% the cases of ultralow HER2 levels that were mistakenly categorized as HER2-negative, researchers added. Only 4% of these cases were misclassified when using AI, compared to nearly 30% without the AI assistance.

“Accurate HER2 scoring is important to ensure that patients receive the best treatment for their breast cancer,” said Dr. Julian Hong, medical director of radiation oncology informatics at the University of California-San Francisco, who reviewed the findings.

“This international study shows that an AI-assisted approach improved HER2 scoring, including in situations that would affect treatment decisions,” he continued in a news release. 

“These findings shed light on the promising role for AI in oncology, not as a replacement for the physician, but as a powerful tool to help us work smarter and faster to deliver high-quality, more personalized care,” Hong concluded.

The researchers next plan to incorporate the AI in routine cancer care, to see if it improves patients’ treatment and outcomes.

The findings are scheduled for presentation at ASCO's annual meeting Friday through Tuesday in Chicago.

Findings presented at medical meetings should be considered preliminary until published in a peer-reviewed journal.

More information

The American Cancer Society has more on HER2 and breast cancers.

SOURCE: American Society of Clinical Oncology, news release, May 22, 2025