Thanks for the cite (BATTLE) - it provides some possible hints (hinting only because: a) they were checking 11 biomarkers so need a strong p value to be despositive, b) they are checking against a control arm with erlotinib, an EGFR inhibitor, and EGFR inhibitors are known to accelerate disease in some patients (see below cite)). FWIW there are several papers of similar fidelity, but nothing very definitive.
See Gefinib trial for an example of a completely definitive study on fact that Gefitinib actually accelerates disease in some patients (I am not expecting such a definitive study in VEGFR patients - but just looking for something more definitive than seems to have been done.)
Market Data 
Markets 
AI
