Selection of Unrelated Allogeneic Hematopoietic Cell Donors Based on KIR3DL1 Allotypes Is Feasible and Results in Improved Disease-Free Survival in Transplant Recipients with MDS and AML

Shaffer, Brian C.; Heller, Glenn; Le Luduec, Jean-Benoit; Vahradian, Jack; Perales, Miguel; Giralt, Sergio A.; Tamari, Roni; Gyurkocza, Boglarka; Jakubowski, Ann A.; Papadopoulos, E B; Boudreau, Jeanette E; Kernan, Nancy A.; Hsu, Katharine C.

doi:10.1182/blood.V128.22.990.990

Abstract

Donor natural killer (NK) cells impart a potent anti-leukemia effect after allogeneic hematopoietic cell transplantation (allo HCT). The killer Ig-like receptors (KIR) play a central role in modulating NK effector function. Chief among them, the inhibitory KIR3DL1 exhibits a high degree of allelic polymorphism. We previously demonstrated that highly expressed KIR3DL1 alleles (KIR3DL1 allele groups *001 and *002) bind preferentially to HLA-Bw4 with I⁸⁰ to confer a high degree of inhibitory signaling. Similarly, poorly expressed KIR3DL1 alleles (KIR3DL1 allele groups *005 and *007) bind to HLA-Bw4 with T⁸⁰ to confer a strong inhibitory signal. In a retrospective cohort of 299 patients undergoing allo HCT for AML, we demonstrated that these highly inhibitory donor KIR3DL1/recipient HLA-Bw combinations (KIR3DL1-HIGH) resulted in a greater incidence of recipient relapse when compared to other donor/recipient KIR3DL1/HLA-Bw pairings with low or absent inhibitory signal (KIR3DL1-LOW/NO). (Giglio F et al, ASH Annual Meeting Abstracts, 2012) Here, we evaluated whether it was feasible to prospectively type and use KIR3DL1 allotypes in unrelated allo HCT donor selection, and whether this intervention would result in improved outcomes in patients undergoing transplant for AML and MDS. We performed PCR-SSP based intermediate resolution KIR3DL1 allele typing on all unrelated adult donors evaluated for patients with MDS and AML at our center from 2013 to present as previously described. (Boudreau J et al, PLoS One, 2014) KIR3DL1 status was provided to the treating physicians, who made the final donor selection.

A total of 941 prospective allo HCT donors underwent high resolution HLA typing and KIR3DL1 allotyping for 252 patients (median per patient = 4, range 1-12). Among all donors evaluated, 27% were found to be KIR3DL1-HIGH when considering the respective recipient HLA-Bw. Having multiple donors evaluated improved the likelihood of avoiding a KIR3DL1-HIGH donor: Among recipients with one donor evaluated, 27% had only KIR3DL1-HIGH donors available compared to 12% in recipients with 2-3 donors evaluated and 4% for recipients with >3 donors evaluated (P < 0.0001). Among all evaluated patients, 41% had both KIR3DL1-HIGH and KIR3DL1-LOW/NO donors available, indicating a potential selection based on KIR3DL1 was possible.

Among 252 patients evaluated, 115 proceeded to allo HCT (48 with MDS, 67 with AML). The median age at transplant was 62 years (range 3.6-78). Donors were HLA matched in 105 transplants and were single loci mismatched in 10 transplants. Conditioning was myeloablative in 73 (11 with total body irradiation). GVHD prophylaxis was with ex vivo CD34⁺ selection in 65 patients and with tacrolimus/methotrexate in the remaining patients. The median follow-up in transplanted patients was 13.1 months (range 0.5-43.3 months). On univariate analysis, the 2-year disease-free survival was 64% (95% confidence interval: 54-76%) in those with KIR3DL1-LOW/NO donors versus 39% (22-68%) in recipients with KIR3DL1-HIGH donors (P = 0.05). The incidence of relapse was similar but favored patients with KIR3DL1-LOW/NO donors (26% [15-37%] versus 35% [13-57%], P = 0.5). 2-year overall survival was similar between those with KIR3DL1-LOW/NO versus KIR3DL1-HIGH donors (75% [65-86%] versus 69% [52-91%], P = 0.43).

The time from the initiation of a formalized donor search to transplant did not differ between recipients with KIR3DL1-LOW/NO versus KIR3DL1-HIGH donors (median 81 v. 83 days, respectively, P = 0.97). Recipients with KIR3DL1-LOW/NO donors were CIBMTR Transplant Risk Score low = 53, intermediate = 6, and high = 33; whereas recipients of KIR3DL1-HIGH donors were low = 17, intermediate = 2, and high = 4 (P = 0.15). These results indicate that disease severity and transplant urgency did not influence whether a KIR3DL1-LOW/NO donor was able to be selected.

In summary, these prospective data on 115 patients from a single center support improved outcomes in patients with AML and MDS undergoing unrelated donor allo HCT using a donor with low or absent KIR3DL1 inhibition. A multi-center, prospective study to evaluate the prospective use of HLA and KIR genotype based selection of URDs for patients undergoing allo HCT for AML is underway (NCT02450708).

Figure 1.