Researchers are making strides in identifying patients with hormone receptor (HR)-positive, HER2-negative early breast cancer who face a very low risk of distant recurrence. By combining artificial intelligence (AI), digital pathology, and traditional clinical factors, clinicians can now gain a more precise understanding of which patients are likely to benefit from specific treatments.
At the 43rd Annual Miami Breast Cancer Conference, data revealed that integrating independent prognostic markers—including digital imaging and gene expression—offers a more detailed risk assessment than standard tools alone. W. Fraser Symmans, MD, a leading expert from The University of Texas MD Anderson Cancer Center, highlighted the importance of advancing both prognostic and predictive biomarkers for HR-positive/HER2-negative breast cancer.
One major challenge in this patient population is the so-called “prognostic paradox.” Some high-risk tumors respond well to chemotherapy yet still carry worse long-term outcomes due to aggressive residual disease. Symmans noted that while predicting overall prognosis has become easier with genomic tools, anticipating individual treatment responses remains complex without a clear biological connection to therapy.
Recent studies validate AI-driven models that combine clinical factors with digital pathology features, such as mitotic rate and nuclear characteristics. In a study analyzing 633 patients from the CANTO registry and the UNIRAD trial in France, an AI-based score identified 20% of high-risk patients who had only a 4.6% rate of distant recurrence over nine years. Similarly, integrating the 21-gene Recurrence Score with the Sensitivity to Endocrine Therapy (SET2,3) Index improved five-year recurrence predictions, highlighting the benefit of combining multiple independent prognostic tests.
The SET index, particularly SET2,3 and SET ER/PR variants, has shown promise in predicting which patients may benefit from extended endocrine therapy or specific chemotherapy regimens. Patients with high endocrine activity experienced a 7.1% absolute improvement in ten-year breast cancer-free intervals when treated with extended letrozole therapy. In contrast, patients with low endocrine activity derived minimal benefit, illustrating that prognostic risk and endocrine sensitivity are distinct factors that must be considered together.
Advances in chemoprediction also point to personalized treatment strategies. Data from CALGB 9741 and GEICAM/9906 trials suggest that patients with low endocrine activity may benefit more from dose-dense chemotherapy or taxane-based regimens. Observational data from the FLEX registry further confirmed that higher-risk biological profiles consistently correlate with better chemotherapy responses.
Symmans emphasized that while prognosis is relatively straightforward, predicting treatment response is far more complex. The emerging approach of combining AI-based histopathology with genomic and endocrine activity markers represents a major step toward precision medicine in breast cancer. Clinicians, however, are advised to interpret these tools cautiously until they are fully validated and integrated into routine practice. By leveraging these multimodal signatures, oncologists can develop more tailored treatment plans, potentially improving outcomes for patients with HR-positive, HER2-negative breast cancer.
