Construction and validation of a necroptosis-related prognostic signature in acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML), an uncommonly low 5-year survival and high mortality rate, is a potentially catastrophic diagnosed subtype of leukemia. The development of new prognostic markers is urgently needed to guide its treatment. Necroptosis is a newly defined biological process for regulating cell death, and previous studies have confirmed that the abnormality of the physical function can lead to multiple malignancies. Here, we performed necroptosis-related genes (NRGs) to build a predictive model in the Cancer Genome Atlas (TCGA)-AML patients, thus exploring the correlation between the NRG prognosis signature (NRG score) of this model and immune infiltration, pathway activity, clinical features, and immunotherapy. Besides, we computed the statistical measure Spearman rank correlation between the NRG score and the Log IC50 values of therapeutic agents. Subsequently, we divided the TCGA-AML cohort into 2 groups, one with high scores and the other with low scores depending on the model score. AML patients with high NRG scores exhibited a lower estimated overall survival (OS) rate than those with low NRG scores, which was confirmed in the validation set. The prognostic value of the constructed NRG signature to the AML, independent of other variables, was demonstrated by uni- and multivariate stepwise regression analysis. When comparing the infiltrating states of specialized cells associated with immune system from the 2 groups, B cells naive, Plasma cells, and monocytes represented significant differences among various subgroups of samples. Moreover, the 30 hallmark-related pathways related to necroptosis characteristics were remarkably different between the high/low NRG score groups. And patients showed remarkable NRG score distribution in clinical features of bone marrow lymphocyte, category, and FAB classifications. Besides, we found that the BIRB0796, VX680, Vorinostat, and Axitinib positively related with NRG score, whereas CI. 1040, PD. 0325901, Z.L LNle. CHO, and AZD6244 negatively correlated with the NRG score. These drugs may provide a reference for subsequent treatment.

Figure 1.

The distribution of necroptosis at the genomic level. (A) The level of NRGs mutation rate in AML. (B) Diagram of chromosome distribution of NRGs. AML = acute myeloid leukemia, NRGs = necroptosis-related genes.

Figure 2.

The results of the enrichment analysis of NGRs in various clinical features. (A) and (B) The difference in the enrichment of NRGs across different clinical characteristics. (C) and (D) Heatmaps panels of the NRGs expression profiling in different clinical features, with red indicating high-expression genes and blue indicating low-expression genes. (E) and (F) Z-score heat maps of differential methylation probes in patients with different clinical characteristics. NRGs = necroptosis-related genes.

Figure 3.

Kaplan–Meier curve of genes associated with necroptosis (A, PLA2G4A; B, H2AZ2; C, CASP1; D, CAMK2D; E, TLR4; F, STAT5B; E, PYCARD; H, GLUD1). HR = hazard ratio, OS = overall survival.

Figure 4.

Prediction of OS in AML patients and the veracity of NRG score considering other clinical/pathological information using the Cox regression hazard method. (A)The graphical representation of the survival probabilities of AML patients with high-(Red line)/low-risk (Blue line) scores in the TCGA-AML cohort. (B) The graphical representation of proportional hazards model of the prognostic value for the risk score and clinical/pathological parameters with prognostic potential. AML = acute myeloid leukemia, NRGs = necroptosis-related genes, OS = overall survival, TCGA = The Cancer Genome Atlas.

Figure 5.

The necroptosis-related signature was validated using the GSE12417 and GSE146173 datasets. Graphical representations of the probability of survival time were plotted for patients with high and low scores in GSE12417 (A) and GSE146173 datasets (C). Identification of independent prognostic factors to assess the score and clinicopathological parameters with prognostic potential in the GSE12417 cohort (B) and the GSE146173 dataset (D).

Figure 6.

The relationship between necroptosis-related patterns and the TIME. Panel A shows a boxplot of the infiltration state of 22 TIICs in relation to necroptosis-related patterns (A). The graphical representation of the relationship between the NRG score and immune cell infiltration for the 2-TIME groups (B). “*” represents the P value that was <.05, “**” represents the P value that was <.01, “***” represents the P value that was <.001, and non-significant results were denoted as “ns.” NRGs = necroptosis-related genes, TIICs = tumor-infiltrating immune cells.

Figure 7.

The graphical representation of the hallmark enrichment difference analysis.

Figure 8.

Distribution of AML scores in clinical features. Category (A), abnormal lymphocyte (B), bone marrow basophil (C), bone marrow blast cell (D), bone marrow lymphocyte (E), FAB (F), Age (G), Gender (H). AML = acute myeloid leukemia.

Figure 9.

The association between log (IC50) of 8 drugs and risk score.

Figure 10.

The log (IC50) distribution of 8 drugs in the different NRG score groups. NRGs = necroptosis-related genes.

Figure 11.

The correlation between the somatic mutation and the NRG score was analyzed separately for the high-risk group (A) and low-risk group (B). NRGs = necroptosis-related genes.