It wasn't that the central lab disagreed, it is that there were some lab data the investigators couldn't confirm when they went to poster, so they excluded those plus patients who had not received drug or not received a second scan.
In any case, that number isn't the important one for comparison. The important one is in Table 5, the retrospective analysis of patients who were also Tykerb failures is 23.9%. This trial is also Tykerb failures.
As to whether this trial was designed to be approvable? It's a non-SPA, single-arm, Phase II trial so any argument about it not being "designed" as approvable will always have traction. The trial was created by DNA to serve as an earlier approval pathway, however. They have always said if the data are "robust" they will take it to the FDA.
News 
Market Data 
Discover 
