Relapse is the most important cause of post-transplant mortality. The interaction between killer immunoglobulin-like receptors (KIRs) of donor natural killer (NK) cells and human leukocyte antigen (HLA)-class I molecules on recipient target cells may influence the outcome of allogeneic hematopoietic stem cell transplantation (HCT) by modulating NK cell alloreactivity. In addition to modulation by KIRs, NK cells also respond to CMV reactivation post-HCT to promote maturation and functional competence that could enhance their antileukemic effect. Since NK cells are postulated to play a prominent role in the setting of ex vivo T cell depleted HCT, we studied the relative contributions of KIR-mediated alloresponse and early CMV reactivation, in addition to measurable residual disease (MRD) status, on clinical outcomes.


A cohort of 109 consecutive patients who received myeloablative, HLA-identical sibling allogeneic peripheral blood HCT for hematological malignancies was studied at a single institution. All patients received a myeloablative preparative regimen (fludarabine 125 mg/m2, cyclophosphamide 120mg/kg, total body irradiation 400-1200 cGy based on age) followed by the infusion of a CD34+ selected graft from an HLA-identical sibling. Three models of donor KIRs were evaluated: a) KIR2DS1 positive (Venstrom, et al. NEJM 2012) seen in 41% of donors, b) "KIR3" - all 3 KIRs 2DL5A, 2DS1, and 3DS1 (Stringaris, et al. BBMT 2010) seen in 34% or, c) any KIR B genotype defining loci (KIR2DL5, 2DS1, 2DS2, 2DS3, 2DS5, or 3DS1) seen in 68%. CMV reactivation (65%) was described as a time-dependent covariate as well as dichotomized into early (< Day 60) vs other. Baseline multigene array MRD status was determined by RT-PCR (Goswami, et al. BMT 2015) and found to be positive in 6/21 (29%) of AML patients. Cox proportional hazards models were used. The cumulative incidence of relapse (CIR) was estimated and compared by the Gray's method to account for competing risks.


Median recipient age was 43 years and median HCT-CI score was 3 (range 2 - 4). 40% of the cohort had AML, 26% ALL, 19% MDS, 10% NHL/CLL and 5% CML. At a median follow up of ~ 5 years, cumulative incidence of relapse was 35.4%, overall survival (OS) was 48.3% (95% CI 39.2-59.5), and NRM was 26.5%. CMV reactivation was not significantly associated with CIR as a time-dependent covariate (HR 1.4, P =0.35) or as a dichotomized covariate (HR 1.7, P =0. 16). Donor KIR2DS1 (HR 1.8, P = 0.07), KIR3 positivity (HR 1.9, P =0.06), and any haplotype B (HR 1.5, P =0.25) were not significantly associated with CIR. In the multivariate models adjusted for age at HCT, disease risk and CMV reactivation. However, MRD status in AML subjects was highly predictive of future relapse (MRD+ CIR was 83% vs MRD- CIR 21%, HR 7.5, p = 0.0015)(Figure 1).


Pretransplant MRD (confined to AML subjects) strongly predicted future AML relapse in our models, and there was no significant association found with early CMV reactivation or with donor KIRs.

Figure 1.

Cumulative Incidence of Relapse by MRD Status.

Figure 1.

Cumulative Incidence of Relapse by MRD Status.

Close modal

No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution