How does it not also drive regular cells to death?
AKA, drug belonging to [drug class] that kills [neuronal tissue] will also kill [bad neuronal tissue] through same mechanism as [general drug class].
The key question is "if putting a shit-ton of opioids (which already cross into the brain) in cancer patients doesn't already meaningfully treat this condition, why do we imagine that delivering the same class of drug past the blood-brain barrier suddenly will?"
Having worked in a breast cancer lab once upon a time, I can tell you: because run-of-the-mill cell biology doesn't get grant funding if you don't attach "early cancer diagnostics" or "potential cancer treatment" to the grant application.
It appears that as they say now that they have identified the specific mechanism of cell death that can start looking at ways to target the tumor cells without flooding the rest of the brain with loperamide or other opiates. It will be a good trick but better than the other (non) options for glioblastoma.