Decoding Intratumoral Heterogeneity in Breast Cancer: The Evolution From Radiomics to Topological Quantification for Predicting Lymphovascular Invasion.

Lymphovascular invasion (LVI) is a critical independent prognostic factor in invasive breast cancer and is strongly associated with early recurrence and distant metastasis. Accurate preoperative assessment of LVI is therefore essential for treatment planning. However, conventional needle biopsy is limited by spatial sampling error and may not adequately capture intratumoral heterogeneity. In this context, magnetic resonance imaging (MRI) has evolved from a morphologic modality into a noninvasive tool for characterizing tumor heterogeneity. This Scale for the Assessment of Narrative Review Articles (SANRA)-guided narrative review delineates the technological evolution of MRI-based heterogeneity assessment for preoperative LVI prediction in breast cancer. We first summarize advances in radiomics, emphasizing the transition from whole-tumor analysis to optimized volume-of-interest strategies and delta-radiomics for capturing dynamic hemodynamic changes. We then review spatially resolved approaches, including habitat imaging and peritumoral analysis, which may improve prediction by partitioning tumors into biologically distinct subregions and characterizing the tumor microenvironment. Finally, we discuss the intratumoral heterogeneity quantification score, a topological metric that measures spatial fragmentation within tumors. Evidence from diffusion-weighted imaging, diffusion kurtosis imaging, intravoxel incoherent motion, positron emission tomography-derived metabolic heterogeneity, scanner harmonization, and cross-study reproducibility is also synthesized. Together, this progression from static feature extraction to dynamic, spatially explicit quantification provides a framework for precise preoperative risk stratification and supports the development of personalized therapeutic strategies in breast cancer.