Early-stage multi-cancer detection using an extracellular vesicle protein-based blood test

Hinestrosa and colleagues purified extracellular vesicles from plasma with an alternating current electrokinetics platform and built a machine-learning classifier on vesicle protein measurements. In a case-control pilot of 139 stage I and II pancreatic, ovarian or bladder cancers against 184 controls, the area under the curve was 0.95, with sensitivity 71.2% (63.2 to 78.1) at 99.5% specificity; stage I sensitivity was 70.5%, reaching 95.5% in pancreatic, 74.4% in ovarian and 43.8% in bladder cancer.