How new drugs and AI are helping to treat HER2-low breast cancer patients
Every year, around two million women are diagnosed with breast cancer. Over the past few decades, the survival rate has rapidly increased, due to an emphasis on earlier detection and the development of better treatments.
One such group of treatments is HER2 inhibitors. These drugs are targeted at the one in five patients whose tumors test positive for a growth-promoting protein known as HER2.
But not all women whose tumors contain HER2 have historically been included in this treatment group, because their levels of HER2 have been considered too low for successful treatment. It is thought that around half of all breast cancer patients may be classified as ‘HER2-low’, and their treatment is currently limited to chemotherapy, if no other targeted therapy is available.
Previous efforts to bring the benefits of anti-HER2 therapies to HER2-low patients have yielded little to no success.
But in landmark news for patients, AstraZeneca recently announced that Enhertu, its anti-HER2 drug, can significantly help women with HER2-low breast cancer. The company said that results from a Phase III trial showed that Enhertu prolonged survival and slowed the progression of metastatic HER2-low tumors, and delivered better results than the physician’s standard choice of chemotherapy. The news suggests that millions more women could now receive potentially life-saving treatment every year.
It also opens up a number of exciting research opportunities that could spur on further advances in treatment, such as:
- the need to better understand why Enhertu can effectively target HER2-low tumors;
- if these treatments can effectively target low-expressed genes, it opens new doors to new potential targets for treatment;
- innovative methods to evaluate the HER2 status of patients that will lead to the most effective treatment.
To address this third question, Owkin is working with Guy's and St Thomas' NHS Foundation Trust and King’s College London to develop machine learning models to better detect HER2-low tumors in patients.
At present, accurately quantifying HER2 levels remains a laborious effort for doctors. Currently, human pathologists analyse tissue sample slides to detect the presence of HER2 proteins.
The new AI models being developed will use deep learning trained on hundreds of retrospective tissue samples to accurately detect the expression level of HER2, including HER2-low. We will then use this to develop a clinical tool to allow patients’ HER2 status to be rapidly classified, allowing them to receive the most appropriate treatment as soon as possible. Using AI to better spot some of the most aggressive forms of breast cancer has the potential to radically improve outcomes for patients.
AI is helping to improve breast cancer treatment in other ways, too. Through our RACE AI project, Owkin is working with Gustave Roussy to develop an AI diagnostic that can identify early breast cancer patients’ risk of relapse, helping to inform clinical decision making.
While this is excellent news for breast cancer patients, it could also help millions more patients suffering from other cancers, as HER2 is overexpressed in numerous cancers, including breast, gastric, lung and colorectal. While further research is needed, the success of Enhertu against HER2-low breast cancer tumors is promising news for these patients, too.
Authors
Testimonial
