Plasmonic alloys enhance metabolic fingerprints for rapid diagnosis and classification of myocardial infarction

Abstract

Although cardiac troponin (cTn) is a recommended clinical biomarker of myocardial infarction (MI), it is inefficient for MI diagnosis requiring serial cTn measurements and is inaccurate for MI subtype identification. Advanced bioanalytical platforms commonly rely on materials with tailored structure and composition. Here, we construct porous PtAu alloys to effectively extract serum metabolic fingerprints (SMFs) via laser desorption/ionization mass spectrometry (LDI-MS), achieving accurate diagnosis and classification of MI by machine learning of SMFs. The PtAu alloys demonstrate enhanced metabolite detection, superior to the monometallic nanoparticles and organic matrix. It is attributed to the porous structure, enhanced photocurrent response, electromagnetic field, and photothermal conversion. Machine learning of SMFs yields diagnostic models with the area under curves (AUCs) of 0.941–1 for 604 subjects from multiple centers in a serum test, overcoming the clinical inefficiency for serial cTn measurements. In particular, our platform achieves accurate discrimination among patients with type 1 MI, type 2 MI, and myocardial injury, with a maximum AUC of 0.905, outperforming the cTn biomarker. Notably, the diagnosis and classification for MI can be finished within ∼30 min. Our platform has the potential to reduce time spent in the emergency department and improve treatment for MI.