Defective Natural Killer (NK) Receptor Expression and Effector Function in Acute Myeloid Leukemia Partially Normalize At Remission and Predict Response to Chemotherapy.

Abstract 2591

Despite favorable initial responses to induction chemotherapy, most patients with acute myeloid leukemia (AML) will relapse. We hypothesized that just as cancers evade immunosurveillance by suppressing the immune response (“immunoediting”), remission and relapse in AML may be determined by similar immune interactions. After stem cell transplantation natural killer (NK) cells exert powerful allogeneic graft vs. leukemia effects. To explore immunosurveillance by autologous NK cells and immunoediting by AML blasts, we prospectively analyzed NK surface phenotype and function in AML patients at presentation and following remission induction. Using multi-color flow cytometry, we analyzed the surface expression of natural cytotoxicity receptors (NCRs), killer immunoglobulin receptors (KIRs) and C-type lectins directly ex-vivo in 32 consecutive patients at presentation and following complete remission (CR) in 12 patients for whom remission samples were available. Results were compared with 15 healthy controls. NK effector function in AML was measured against K562 leukemia targets and autologous AML blasts by CD107a degranulation and interferon-gamma (IFNγ) and TNF-alpha (TNFα) production.

NK cells from 32 patients with AML at diagnosis had an abnormal phenotype compared to controls, with downregulation of the activatory receptor NKp46 (MFI 187± 15 vs. 266± 24, p=0.007) and upregulation of the inhibitory receptor NKG2A (mean 45%± 4.2 vs. 32%± 2.7, p=0.046). Moreover, AML-NK cells were defective in their effector function with significantly reduced CD107a degranulation (5% vs. 11%, p=0.0002), TNFα (1% vs. 3%, p=0.008) and IFNγ production (1% vs. 5%, p=<0.0001). In the 12 patients who achieved remission following induction chemotherapy NKp46 expression normalized from an MFI 122 at presentation to 242 (p=0.01); however, NKG2A expression continued to increase (45% vs. 61%, p=0.008). At remission AML-NK cells displayed CD107a degranulation levels comparable to that of healthy donor NK cells (9% vs. 11%, p=0.4). In contrast, TNFα and IFNγ production only partially normalized (2% vs. 3%, p=0.3) and (3% vs. 5%, p=0.04), respectively.

AML patients with NKG2A overexpression (> median of 32.6%) at diagnosis were significantly less likely to achieve CR post chemotherapy compared to those with lower NKG2A expression (78% vs. 32% p= 0.041). Furthermore, patients who failed to respond to induction chemotherapy had significantly reduced NK effector function at diagnosis compared to normal controls (mean TNFα production 1% cf. 5% p=0.019).

We then sought for evidence of immunoediting by AML on NK cell phenotype and function. Co-incubation of healthy donor NK cells with primary AML blasts for 24 hours resulted in significant reduction in TNFα production (p=0.02), IFNγ production (p=0.01) and a trend to reduced CD107a degranulation (p=0.07) against K562 leukemia targets. Our results indicate that AML blasts can produce long-lasting changes in NK cell subsets and impair their effector function and favouring immune escape from NK cell control. This work justifies larger prospective analyses to relate NK-based prognostic factors to classical predictive factors for remission and relapse, and should guide the development of NK cell based immunotherapy to improve outcome in AML.