Gastric cancer (GC) presents high mortality worldwide because of delayed diagnosis. Currently, exosome-based liquid biopsy has been applied in diagnosis and monitoring of diseases including cancers, whereas disease detection based on exosomes at the metabolic level is rarely reported. Herein, the specific aptamer-coupled Au-decorated polymorphic carbon (CoMPC@Au-Apt) is constructed for the capture of urinary exosomes from early GC patients and healthy controls (HCs) and the subsequent exosome metabolic pattern profiling without extra elution process. Combining with machine learning algorithm on all exosome metabolic patterns, the early GC patients are excellently discriminated from HCs, with an accuracy of 100% for both the discovery set and blind test. Ulteriorly, three key metabolic features with clear identities are determined as a biomarker panel, obtaining a more than 90% diagnostic accuracy for early GC in the discovery set and validation set. Moreover, the change law of the key metabolic features along with GC development is revealed through making a comparison among HCs and GC at early stage and advanced stage, manifesting their monitoring ability toward GC. This work illustrates the high specificity of exosomes and the great prospective of exosome metabolic analysis in disease diagnosis and monitoring, which will promote exosome-driven precision medicine toward practical clinical application.

https://doi.org/10.1021/acsnano.2c05355