Despite intense interests in developing blood measurements of Alzheimer’s disease (AD), the progress has been confounded by limited sensitivity and poor correlation to brain pathology. Here, we present a dedicated analytical platform for measuring different populations of circulating amyloid β (Aβ) proteins – exosome-bound vs. unbound – directly from blood. The technology, termed a mplified p lasmonic ex osome (APEX), leverages in situ enzymatic conversion of localized optical deposits and double-layered plasmonic nanostructures to enable sensitive, multiplexed population analysis. It demonstrates superior sensitivity (~200 exosomes), and enables diverse target co-localization in exosomes. Employing the platform, we find that prefibrillar Aβ aggregates preferentially bind with exosomes. We thus define a population of Aβ as exosome-bound (Aβ42+ CD63+) and measure its abundance directly from AD and control blood samples. As compared to the unbound or total circulating Aβ, the exosome-bound Aβ measurement could better reflect PET imaging of brain amyloid plaques and differentiate various clinical groups.

Exosomes are nanoscale vesicles distinguished by characteristic biophysical and biomolecular features; current analytical approaches, however, remain univariate. Here, we develop a dedicated platform for multiparametric exosome analysis—through simultaneous biophysical and biomolecular evaluation of the same vesicles—directly in clinical biofluids. Termed templated plasmonics for exosomes, the technology leverages in situ growth of gold nanoshells on vesicles to achieve multiselectivity. For biophysical selectivity, the nanoshell formation is templated by and tuned to distinguish exosome dimensions. For biomolecular selectivity, the nanoshell plasmonics locally quenches fluorescent probes only if they are target-bound on the same vesicle. The technology thus achieves multiplexed analysis of diverse exosomal biomarkers (e.g., proteins and microRNAs) but remains unresponsive to nonvesicle biomarkers. When implemented on a microfluidic, smartphone-based sensor, the platform is rapid, sensitive, and wash-free. It not only distinguished biomarker organizational states in native clinical samples but also showed that the exosomal subpopulation could more accurately differentiate patient prognosis.