Functional biomaterials and machine learning approaches for phenotyping heterogeneous tumor cells and extracellular vesicles

Abstract

Heterogeneity in cancer is known to be a contributor to the formation of metastatic lesions, poor prognosis, and ultimately undermines therapeutic efficacy. This same tumor heterogeneity is reflected in circulating tumor cells (CTCs) and tumor derived extracellular vesicles (EVs), offering a less invasive snapshot into tumor status. The isolation and phenotypic characterization of CTCs and EVs has attracted significant interest in recent years as they offer great potential for deciphering the molecular basis of disease progression and the development of precision therapies. Next-generation biomaterials and advanced machine learning paradigms are transforming how we decipher phenotypic heterogeneity in these circulating biomarkers and help provide new insights into tumor biology and therapy resistance. In this review, we first briefly describe biomaterial-based platforms for the isolation of CTCs and EVs, followed by a detailed discussion of the pivotal role of phenotypic profiling and molecular identification. Finally, we provide a review of emerging biomaterial-based approaches that enable selective sorting, profiling, and detection of CTCs and EVs. In this process, we categorize the most widely utilized biomaterials into polymer-based materials, quantum dots and multifunctional magnetic nanospheres, fluorescent antibodies, Surface Enhanced Raman Spectroscopy (SERS) vectors, DNA-based multi-aptamer probes, cell-imprinted substrates, and silver nanoclusters. We then explore the application of machine learning algorithms in biomarker profiling of CTCs, extending to EVs. Furthermore, we provide a comprehensive analysis of relevant clinical studies and critically examine future challenges and research trajectories in this rapidly evolving field.