Hi Amy, Thanks for your comments. In answer: (1) This works best if it achieves a relatively large scale, like 1000 patients per year. At this scale, a biomarker at 2% prevalence (low) would get 20 patients a year, which is reasonable for Phase 1/1b. For a higher prevalence biomarker (10%), 1000 patients a year could nicely support a randomized Phase 2 trial. (2) This really depends on the relative prevalence of the biomarkers. Some could have as low as 2% while others 30-40%. 95% is a high bar for genetic-only biomarkers, but if the study included others, you could get there reasonably with 5 or more arms. (3) As ballpark figures, recruiting, consenting, and CT-guided biopsies together cost $2000-$4000, depending on the site. CLIA assays cost $500 - $2000, depending on the assay. Per-patient costs on a trial, however, are usually around $80,000 without the biopsies and assays. So, if collectively you can get the screen-fail rate below 50%, all the additional costs of running a biomarker-driven trial relative to an unselected trial drop to a ~10% premium, which is well worth it. (4) The reason I said "likely come from philanthropic sources" is that the largest hurdle to this idea really is buy-in from the companies. No one wants to pay for the initial setup because you are already taking a risk as an early adopter. So, if this could be covered by a donation, it would get the ball rolling.