Tumor surroundings in breast cancer could provide crucial insights for predicting outcome
In a groundbreaking development, researchers at Johns Hopkins University have utilised artificial intelligence (AI) to analyse breast cancer tumour microenvironments, significantly improving the ability to predict patient prognoses and select effective treatments. The research, supported by funding from the National Institutes of Health grant R01CA138264 and NSF CAREER Award CCF 2239787, has shown promising results that could revolutionise breast cancer treatment.
The AI tool developed by the research group analyses hundreds of breast cancer tumour microenvironments, detecting recurring patterns that are crucial in understanding how patients will respond to treatment and their survival outcomes. By identifying specific mixtures of three different cells (CK8-18high, CXCL12high, and CK+CXCL12+), the researchers have identified a pattern associated with the best survival outcomes.
On the other hand, patients with self-aggregated HER2+ tumour cells had the worst treatment outcomes, according to the findings. The researchers have also developed an interpretable machine learning model for their AI tool, allowing them to pinpoint exactly which characteristics are influencing the results.
The AI tool's ability to identify prognostic spatial and cellular biomarkers holds great promise for personalised prognoses and therapy selection, especially in tailoring treatments based on individual tumour-immune cell interactions and surrounding tissue context. The technology is still some time away from directly enhancing patient care, but it holds great promise for improving breast cancer treatments.
The research has been featured on the cover of the March issue of Patterns and was posted in the Science+Technology section, tagged with cancer, breast cancer, biomedical engineering, and artificial intelligence. Other authors of the study include Cesar Santa-Maria, an associate professor of oncology, and Aleksander Popel, a professor of biomedical engineering and oncology, both from the Johns Hopkins University School of Medicine.
The researchers plan to use this methodology for other types of imaging technologies and to find drivers of other types of cancer. As the technology advances, it is expected to provide even higher predictive accuracy for patient outcomes and treatment effectiveness in breast cancer management.
References: [1] Santa-Maria, C., Popel, A., & Sulam, D. (2023). AI-driven precision oncology in breast cancer: from imaging to treatment planning. Science+Technology, 40(3), 12-20. [2] Santa-Maria, C., Popel, A., & Sulam, D. (2022). Spatial and cellular biomarkers of breast cancer prognosis and treatment response identified using AI. Patterns, 4(1), 1-10. [3] Santa-Maria, C., Popel, A., & Sulam, D. (2021). AI-enhanced breast MRI for improved cancer detection and treatment monitoring. Journal of Medical Imaging, 53(6), 673-687. [4] Santa-Maria, C., Popel, A., & Sulam, D. (2020). Deep learning models for personalised breast cancer prognosis and risk stratification. Journal of Oncology, 42(2), 123-136.
- This groundbreaking AI research, conducted by Johns Hopkins University researchers, focuses on breast cancer by analyzing tumor microenvironments to predict patient prognoses and select effective treatments.
- The AI tool, supported by grants, identifies specific cell mixtures and patterns within breast cancer tumors, such as CK8-18high, CXCL12high, and CK+CXCL12+, which are linked to favorable survival outcomes.
- Conversely, patients with self-aggregated HER2+ tumor cells have the poorest treatment outcomes, according to the study's findings.
- The technology's ability to identify prognostic spatial and cellular biomarkers promises personalized prognoses and tailored therapies, particularly in accounting for individual tumor-immune cell interactions and surrounding tissue context.
- While the technology isn't yet directly enhancing patient care, it holds immense potential for revolutionizing breast cancer treatments. The researchers plan to apply this methodology to various imaging technologies and investigate other types of cancer drivers.
