In a significant leap forward for breast cancer treatment in the United States, artificial intelligence (AI) is poised to transform how doctors identify and treat patients with low and ultra-low HER2 levels. Recent advancements in AI technology are enabling more accurate detection of HER2-low and HER2-ultralow breast cancers, potentially opening the door to life-changing targeted therapies for thousands of patients who were previously misclassified. This development, highlighted at the 2025 American Society of Clinical Oncology (ASCO) Annual Meeting, could reshape the future of breast cancer care by ensuring more patients receive the treatments best suited to their condition.
HER2, or human epidermal growth factor receptor 2, is a protein found on the surface of some breast cancer cells that drives tumor growth. Traditionally, breast cancers have been classified as HER2-positive or HER2-negative, with targeted therapies like trastuzumab (Herceptin) reserved for those with high HER2 expression. However, recent research has shown that even cancers with low or ultra-low HER2 levels can benefit from newer targeted therapies, such as Enhertu (trastuzumab deruxtecan), a powerful antibody-drug conjugate. The challenge lies in accurately identifying these low-expression cases, as conventional diagnostic methods often misclassify them as HER2-negative, leaving patients without access to potentially life-prolonging treatments.
Historically, studies suggest that at least 55% of breast cancers classified as HER2-negative are actually HER2-low, and another 10% are HER2-ultralow. These misclassifications have limited treatment options for many patients. AI is now stepping in to address this gap, offering hope to those who might otherwise miss out on effective therapies.
At the 2025 ASCO Annual Meeting, researchers presented compelling evidence that AI can significantly improve the accuracy of HER2 classification. In a study involving 105 pathologists across 10 countries, AI-assisted tools reduced the misclassification of HER2-low and HER2-ultralow cases by 24.4%. The study, led by Dr. Marina De Brot, used AI to analyze immunohistochemistry (IHC) slides, which are stained to detect HER2 protein levels in tumor samples. By comparing pathologists’ readings to a central reference standard, the AI tool demonstrated its ability to identify subtle HER2 expressions that human eyes might miss.
The AI system, such as the one developed by Indica Labs, categorizes HER2 expression into more nuanced levels—0, 1+, 2+, and 3+—moving beyond the binary positive/negative framework. This precision is critical because even minimal HER2 expression can make a patient eligible for therapies like Enhertu, which has shown remarkable results in clinical trials. For example, the DESTINY-Breast06 trial found that patients with HER2-low breast cancer treated with Enhertu had a 36% lower risk of disease progression compared to those on standard chemotherapy, with an average progression-free period of 13.2 months versus 8.1 months.
This enhanced detection capability is a game-changer. “The introduction of therapies targeting low HER2 levels has made accurate classification more important than ever,” said Dr. De Brot in an ASCO press release. “AI tools can help pathologists ensure patients are correctly identified for treatments that could extend their lives.”
The implications of AI-driven HER2 detection are profound, particularly for patients with hormone receptor-positive (HR+), HER2-low, or HER2-ultralow metastatic breast cancer. These cancers, which account for about 70% of all breast cancer cases in the U.S., are often harder to treat because they don’t respond as well to traditional therapies. Enhertu, approved by the FDA for HER2-low breast cancer, has already shown promise in extending survival and delaying disease progression. By improving diagnostic accuracy, AI could ensure that more patients qualify for this and similar treatments.
For instance, the DESTINY-Breast09 trial demonstrated that Enhertu, when combined with pertuzumab, significantly improved progression-free survival in patients with HER2-positive metastatic breast cancer. Extending this drug’s reach to those with lower HER2 levels could benefit an estimated 3,000 to 4,000 additional patients annually in the U.S. alone.
The human impact of these advancements cannot be overstated. Consider the case of patients like Linda Kelly, a 67-year-old woman with advanced breast cancer that spread to her bones and chest wall. After being diagnosed with HR-positive, HER2-negative cancer, she was enrolled in a trial for capivasertib, a targeted therapy for patients with specific genetic mutations. While not directly related to HER2 detection, her story underscores the importance of precise diagnostics in identifying eligible patients for targeted treatments. Linda credits the drug with giving her “extra years of life” to spend with her family, highlighting the potential for AI to make similar stories possible for HER2-low patients.
AI’s role in breast cancer care doesn’t stop at diagnosis. Researchers are planning multicenter studies to embed AI tools in routine diagnostics, measuring their impact on treatment allocation and time to therapy. These studies aim to streamline workflows for pathologists, allowing them to focus on complex cases while AI handles initial screenings. The ComPath Academy, an AI training platform, has already shown promise in improving pathologists’ concordance in HER2 scoring, further enhancing diagnostic reliability.
While the potential of AI in breast cancer care is immense, challenges remain. Harmonizing AI tools across different platforms, staining protocols, and practice settings is a significant hurdle. Pathologists will need training to integrate these tools seamlessly into their workflows, and healthcare systems must invest in infrastructure to support AI adoption. Additionally, ensuring equitable access to these technologies is critical, as disparities in healthcare access could limit the benefits for underserved populations.
There’s also the question of overdiagnosis, a concern raised in studies about AI in breast cancer screening. While AI can detect subtle HER2 expressions, it must be carefully calibrated to avoid identifying cancers that may not require aggressive treatment, which could lead to unnecessary procedures. Researchers are addressing this by validating AI tools against diverse populations to ensure they are both accurate and generalizable.
The integration of AI into breast cancer diagnostics represents a pivotal moment in oncology. By detecting low and ultra-low HER2 levels with greater precision, AI is helping to unlock targeted therapies for a broader group of patients, offering hope to those who might have been overlooked by traditional methods. As Dr. Robin Zon, ASCO president, noted, “With the increasing use of anti-HER2 therapies, patients now have access to potentially life-changing medications.”
In the U.S., where breast cancer remains the most common cancer among women, with over 300,000 new cases annually, these advancements could save countless lives. As AI tools continue to evolve and integrate into clinical practice, they promise to make breast cancer treatment more personalized, effective, and accessible. For patients, this means not just longer lives, but better quality of life—a goal that makes every step forward in this field truly meaningful.
America forgotten ghost towns are a unique glimpse into the past. Once full of life,…
NH lake named one of the best in the U.S. for swimming by Reader’s Digest…
Looking for a peaceful escape surrounded by nature? The best hidden gardens and arboretums in…
San Francisco is a food lover’s paradise. With famous eateries like Tartine Bakery, Swan Oyster…
Small historic towns are often overlooked in favor of big cities, but they offer something…
Colorado is a hiker’s dream. From towering mountain peaks to quiet alpine lakes, this state…