Skip to content

Breast cancer remains one of the most common malignancies in women and a leading cause of cancer-related mortality. Although notable progress was made throughout the 2010s, driven in particular by early screening and improved treatments, the heterogeneity of these tumors remains a central problem that shapes patient prognosis. Histological and molecular classifications have enabled clinical practice to evolve from a uniform approach toward more personalized management. This review provides an overview of current strategies and tools aimed at identifying, evaluating and validating new predictive biomarkers, that is, biomarkers capable of anticipating a patient's response to a given treatment.

The authors recall that the classification of breast tumors, initially based on gene expression, tumor characteristics and clinical course, has been refined by expression signatures such as PAM50, which relies on the expression level of 50 genes, or by whole-transcriptome analysis. Although more than ten subtypes have been described, only four are of clinical relevance: Luminal A, Luminal B, HER2-enriched tumors and triple-negative breast cancers (TNBC). Luminal tumors are characterized by the expression of estrogen and progesterone receptors and are distinguished by their proliferation level, assessed using the KI-67 marker. HER2-enriched tumors display amplification of the HER2 oncogene, whereas triple-negative cancers, the most aggressive, express none of these three markers. This classification directly guides the therapeutic choice: hormone therapy for tumors expressing hormone receptors, HER2-targeted therapy for tumors overexpressing this receptor, and conventional chemotherapy combining DNA-targeting agents and spindle poisons for triple-negative cancers.

Although 87% of patients benefit from these treatments, the emergence of resistance and subsequent relapses is a frequent complication and a major cause of death. Patients with apparently similar tumors do not respond identically to the same agents, making stratification based on prediction of treatment response essential. The identification of robust biomarkers offers a twofold benefit: improving clinical outcomes by selecting the appropriate treatment for each subpopulation, while also reducing toxicities and costs by avoiding the use of ineffective treatments.

The review details the approaches mobilized in this field of research — genomic, proteomic and machine learning-based — as well as innovative tools such as imaging and liquid biopsies. Finally, the authors summarize the predictive biomarkers currently available in breast cancer and discuss the companion biomarkers associated with targeted therapies, with a view toward precision medicine aimed at delivering the right treatment to the right patient and avoiding any unnecessary overtreatment.