GRAIL’s use of “staging” endpoints for multi-cancer screening is mathematically unsound, according to a meta-analysis in JAMA: https://jamanetwork.com/journals/jama/fullarticle/2817338?guestAccessKey=c6b0b8b9-a1d0-436b-bab5-dfd9275c5161 This paper has a lot of technical biostatistics, but the conclusion is pretty easy to comprehend.