This is a good review of NKC. Though given the fact that NK cells adoptive transfer is not bound to requirement to be dervived from self unlike CD8 cells used in immunotherapy makes them a cheaper choice. This also overcomes the difficulty of lower proportion you described. That is because you can use iPSc to make NK cells. Li et al last month actually demonstrated this in mice. Infact they made CAR NK cells from iPSc just like those CAR T cells used in adoptive immunotherapy. Both the cells were derived from Humans , but NK cells dervived from human iPSc showed least side effects , plus they were equally as potent as CAR T cells. (Though I won't call it best test for potency as a better comparison would have been with mice derived CAR T, but as far as non requirement of being derived from self is concerned it was a good test). Also using CAR modified NK cells might be a better idea than just proliferating NK cells in vitro and injecting them.

IMO the major drawback of NK cells at least in comparison with other formats of immunotherapy lies in their rapid inactivation by TRegs and may require extra IL2 modulation. Then there are some reports on context dependent role of NK cells in graft rejection scenario. Which may be counterintuitive to li et al finding. Nevertheless, as pre-clinical trials are already in progress that would come out soon.