In CML, the preclinical data has proved very reliable in predicting clinical outcome. There are literally hundreds of known mutations, and each has a slightly different impact on resistance to each of the TKIs. Further, when you get multiple mutations, sometimes they seem to synergize their resistance and other times they don't. Added to that, there are typically multiple clones running around, so a single analysis of the dominant clone won't always be predictive of what drug level to use. Finally there are the resistant patients that don't have known mutations - those are actually some of the hardest patients to treat.
So I stand by my view that there is conceptually no reasonable way to determine what the minimum effective dose is for a particular patient other than by titration. Now I grant you could reasonably argue about what the most efficient titration scheme should be, and I suppose the FDA might demand some trials to establish that. But such trials would very hard and long given you simply aren't dealing with large numbers of potential patients.
Peter
Market Data 
Markets 
AI
