Without taking one side or another in this debate, I note that the paper on EGFRvIII was an in vitro experiment and not in vivo data. There is a lot of cellular cross-talk in the human body such that anything done in a lab is immediately suspect because it is impossible to model all the paracrine and endocrine effects. Similarly, and especially so in immunotherapy, animal species have enough differences in how their immune systems function that great care must be taken when interpreting animal models as well.
Unfortunately, the only way to get accurate data on cellular therapies is to try them on humans afflicted with disease in the context of a well-controlled, randomized trial.