Table 5.

Pros and cons of scoring strategies for NK alloreactivity assessment.

Models	NK cells of the donor alloreactive toward host cells when:	Pros	Cons
Ligand–ligand model	Recipient is lacking an MHC class I ligand that is present in the donor	No KIR typing requiredEasy to use (online algorithm)	Simple approximation of educational modelApproximate estimation of the mismatches if using the IPD database (does not take into account HLA-Bw4 epitopes related to HLA-A and HLA-C or HLA-A3/11 epitopes)
Receptor–ligand model	At least one KIR gene expressed in the donor does not recognize any of the MHC molecules of the recipient	KIR typing only at genic resolution for donors
Educational models	Donor has educated NK cells—i.e., KIR and its cognate MHC ligand—but the recipient lacks the cognate KIR MHC ligand	KIR typing only at genic resolution for donorsMost comprehensive model for NK alloreactivity	Delicate process that can be overridden in certain conditions, e.g., high inflammation surroundings such as in aHSCT
Haplotypes	Donor has at least one KIR B haplotype	KIR typing only at genic resolution for donorsEasy to use (online algorithm)
Gene–gene model	KIR gene is present in the donor but absent in the recipient	Easy to use	Far from any biological underlying process
Allelic polymorphisms	A specific D/R interaction is present	Directly targets a functional gene difference	Multitude of models with variable relevance
			Allelic KIR genotyping (time and cost)
			Complex