NPM1 and DNMT3A mutations are associated with distinct blast immunophenotype in acute myeloid leukemia

Abstract

The immune system is important for elimination of residual leukemic cells during acute myeloid leukemia (AML) therapy. Anti-leukemia immune response can be inhibited by various mechanisms leading to immune evasion and disease relapse. Selected markers of immune escape were analyzed on AML cells from leukapheresis at diagnosis (N = 53). Hierarchical clustering of AML immunophenotypes yielded distinct genetic clusters. In the absence of DNMT3A mutation, NPM1 mutation was associated with decreased HLA expression and low levels of other markers (CLIP, PD-L1, TIM-3). Analysis of an independent cohort confirmed decreased levels of HLA transcripts in patients with NPM1 mutation. Samples with combined NPM1 and DNMT3A mutations had high CLIP surface amount suggesting reduced antigen presentation. TIM-3 transcript correlated not only with TIM-3 surface protein but also with CLIP and PD-L1. In our cohort, high levels of TIM-3/PD-L1/CLIP were associated with lower survival. Our results suggest that AML genotype is related to blast immunophenotype, and that high TIM-3 transcript levels in AML blasts could be a marker of immune escape. Cellular pathways regulating resistance to the immune system might contribute to the predicted response to standard therapy of patients in specific AML subgroups and should be targeted to improve AML treatment.

Keywords: AML; DNMT3A; NPM1; TIM-3; immunophenotype.

Figure 5.

Survival analysis in groups according to selected immunophenotype markers. The analyses of the overall survival (OS, left column) and of the relapse-free survival (RFS, right column) in groups with low versus high levels of the indicated marker were performed using GraphPad Prism software, the obtained p-values for survival difference between groups are given in the graphs. The cutoff values were optimized with regard to the resulting p-values for OS.

Figure 1.

Results of hierarchical clustering of AML samples according to immune escape-related markers. (A) Surface amounts of HLA class I, HLA-DR, CLIP, PD-L1, and TIM-3 were measured using flow cytometry and used as the input for clustering analysis. (B) The graphs show the individual values in samples from the two clusters (cluster 1, cluster 2) defined in the tree diagram (A). The group denoted as “intermed” includes all the remaining samples except for the outlying ID 18. TIM-3 mRNA values were not used for hierarchical clustering, but are shown for comparison. Mann-Whitney test was performed to evaluate differences between cluster 1 and cluster 2, and p-values were found to be less than 0.0001 for all of the parameters. (C) Relative incidence of recurrent mutations in the above defined groups (N = 16, 18, and 18 for cluster 1, intermediate group, and cluster 2, respectively).

Figure 2.

Comparison of the measured values in groups according to NPM1 or DNMT3A mutation status. The cohort was subdivided according to NPM1/DNMT3A mutation, and the values of the indicated surface marker were compared using the non-parametric Mann-Whitney test. CD47 values (panel F) were compared using two-tailed t-test. The resulting p-values are indicated on top of the corresponding graphs. The horizontal bars show the medians, and the star symbol denotes a statistically significant difference between groups.

Figure 3.

HLA-related markers in NPM1/DNMT3A groups. (A) HLA class I and HLA-DR surface expression in patient subgroups according to NPM1 and DNMT3A mutations. Difference between groups were assessed using Mann-Whitney test, the resulting p-values are given in the graphs. (B) Correlations between HLA class I, HLA-DR, and CLIP. Open symbols: samples with wild-type NPM1, closed symbols: samples with mutated NPM1. (C) Samples with more than 85% HLA-DR-positive blasts were divided according to NPM1 or DNMT3A mutation and CLIP amounts were compared using the Mann-Whitney test. The resulting p-values are indicated in the graph.

Figure 4.

Correlation of TIM-3 mRNA with TIM-3 protein, PD-L1, and CLIP surface positivity of AML blasts. (A) Correlation of TIM-3 trancript amount with the fraction of TIM-3-positive cells determined by flow cytometry in mononuclear cell preparations. (B) The fraction of cells expressing the invariant CLIP peptide versus TIM-3 mRNA. (C) The fraction of cells expressing the inhibitory receptor PD-L1 versus TIM-3 mRNA. Spearman correlation test was performed to assess the correlation, the obtained correlation coefficients (r) and p-values are given in the graphs.

Figure 6.

NK cell-mediated lysis of primary AML cells with low or high HLA class I expression. The cytotoxic assay with NK cells from a healthy donor was performed in triplicates at two different effector to target (E:T) ratios as indicated. The means of the triplicates are given in the Figure for samples with low HLA levels (N = 7, open symbols) or with normal HLA levels (N = 5, closed symbols). The bars indicate means and s.d. of the values shown in each group. In the positive control (K562 cells, not shown in the graph), the percentage of lysed cells reached 94% under the same experimental conditions at E:T ratio 10:1.