Accurate Cancer Screening and Prediction of PD-L1-Guided Immunotherapy Efficacy Using Quantum Dot Nanosphere Self-Assembly and Machine Learning.

Accurate quantification of exosomal PD-L1 protein in tumors is closely linked to the response to immunotherapy, but robust methods to achieve high-precision quantitative detection of PD-L1 expression on the surface of circulating exosomes are still lacking. In this work, we developed a signal amplification approach based on aptamer recognition and DNA scaffold hybridization-triggered assembly of quantum dot nanospheres, which enables bicolor phenotyping of exosomes to accurately screen for cancers and predict PD-L1-guided immunotherapeutic effects through machine learning. Through DNA-mediated assembly, we utilized two aptamers for simultaneous ultrasensitive detection of exosomal antigens, which have synergistic roles in tumor diagnosis and treatment prediction, and thus, we achieved better sample classification and prediction through machine-learning algorithms. With a drop of blood, we can distinguish between different cancer patients and healthy individuals and predict the outcome of immunotherapy. This approach provides valuable insights into the development of personalized diagnostics and precision medicine.