Breast cancer is the most common cancer among women worldwide and a leading cause of cancer-related deaths. Recurrence and survival rates vary by BC subtype, with triple-negative BC (negative estrogen receptor [ER], progesterone receptor [PR], and Human Epidermal Growth Factor Receptor 2 [HER2]) being the most aggressive. Mammographic breast density, indicating the amount of epithelial and stromal elements in the breast, is an independent marker for predicting BC risk. Additionally, mammography's effectiveness in diagnosing cancer diminishes in women with very dense breasts, making lesion identification challenging due to masking phenomena. Studies have shown that high breast density is also linked to a higher risk of cancer in the contralateral breast and early-onset breast cancer. Dense stromal tissue can induce proliferation in the breast epithelium, indicating cross-talk between stromal cells and epithelial cells. This underscores the importance of a consistent method for assessing and monitoring breast density to optimize breast cancer screening strategies.

The American College of Radiology (ACR) introduced the BI-RADS system for categorizing mammographic density, typically determined visually. The 4th edition of BI-RADS categorized breast density by fibroglandular tissue percentage, while the 5th edition emphasized qualitative analysis, increasing subjectivity and variability among radiologists. This variability affects risk assessment accuracy and result reproducibility. Standardized protocols, adequate training, and intra-operator variability assessment are crucial for enhancing consistency in breast density assessments.

Artificial intelligence (AI) is increasingly used to measure breast density, improving objectivity and reproducibility. Commercial AI tool using quantitative methods perform automated measurements of volumetric breast density and classify them similarly to BI-RADS categories, calculating volumetric breast density as the fibroglandular tissue volume percentage of total breast volume. Given the impact of breast density on both the risk of developing breast cancer and the challenges in its detection, it was studied how breast density might affect treatment outcomes, particularly in the context of neoadjuvant chemotherapy. Neoadjuvant treatment with chemotherapy is widely used for triple-negative and HER2+ invasive ductal breast cancer, facilitating tumor downstaging, breast-conserving surgery, and potentially achieving a pathological complete response. Monitoring neoadjuvant treatment response is crucial for evaluating treatment effectiveness and guiding therapeutic decisions. Despite extensive research, there is not a specific biomarker that can predict appropriately the neoadjuvant chemotherapy response, highlighting the need for personalized oncological treatment.

The primary outcome measure of this study was to determine how the density of breast, classified using the ACR classification, changed before and after NAT in women with TN and/or HER2+ invasive ductal breast cancer. The second aim of the study was to assess the inter-reader variability in breast density measurements varied between two radiologists with different level of experiences and between the most experience radiologist and an automatic tool for breast density classification based on an AI tool.