I am a bit confused by your reasoning of how the two numbers are related. Why do you think they got the 50% number from that study? That study wasn't for NSCLC patients treated with Xalkori.
I assume that the Xalkori numbers were from Xalkori studies or from studies that Ariad had done. To make the comparison you are making, you would need to know the rates of brain metastasis in ALK+ primary tumors versus the other subtypes of NSCLC.
If the ALK incidence rate is about 5%, and if the study found only half that rate in brain mets, that suggests that ALK+ tumors don't metastasize to the brain as much as other lung tumors, or that when primary tumors that have ALK translocations in at least part of the tumor, the metastatic tumor has either lost or a ALK- subclone from the primary tumor seeded the metastasis. My guess is that the numbers are still small, and that the ALK+ % in brain mets will be similar to the AKL+ % in primary tumors.
My guess is that the ALK+ brain mets came from ALK+ tumors, because ALK translocation is thought to be the driver in the tumors it is in --> which explains the efficacy of drugs targeting ALK in ALK+ tumors. This suggests ALK+ is probably an early event present in the majority of the primary tumor, and thus ALK+ mets probably have maintained the ALK+ state. Of course, you would need patient matched controls to determine this. In any case, no matter how you get to the 50% of ALK+ patients treated with Xalkori having brain mets, it is clear that a drug able to inhibit the growth of brain mets would be beneficial in these patients.