I don't follow that logic. Sure, it stands to reason that
if the drug can only kill infected cells, and the human hasn't already lost enough cells to kill them, then the drug won't kill the human.
But let's suppose that you can't measure in advance the number of cells that are infected, nor can the drug actually tell the difference between infected and not-infected cells (due to the aforementioned genetic abnormality.) What happens when the drug kills too many normal cells? The human dies.