Construction and validation of a SASP-related prognostic signature in patients with acute myeloid leukaemia

Abstract

Acute myeloid leukaemia (AML) is a common and highly aggressive haematological malignancy in adults. Senescence-associated secretory phenotype (SASP) plays important roles in tumorigenesis and progression of tumour. However, the prognostic value of SASP in patients with AML has not been clarified. The present study aims to explore the prognostic value of SASP and develop a prognostic risk signature for AML. The RNA-sequencing data was collected from the TCGA, GTEx and TARGET databases. Subsequently, differentially expressed gene analysis, univariate Cox regression and LASSO regression were applied to identified prognostic SASP-related genes and construct a prognostic risk-scoring model. The risk score of each patient were calculated and patients were divided into high- or low-risk groups by the median risk score. This novel prognostic signature included 11 genes: G6PD, CDK4, RPS6KA1, UBC, H2BC12, KIR2DL4, HSF1, IFIT3, PIM1, RUNX3 and TRIM21. The patients with AML in the high-risk group had shorter OS, demonstrating that the risk score acted as a prognostic predictor, which was validated in the TAGET-AML dataset. Univariate and multivariate analysis revealed the risk score was an independent prognostic factor in patients with AML. Furthermore, the present study revealed that the risk score was associated with immune landscape, immune checkpoint gene expression and chemotherapeutic efficacy. In the present study, we constructed and validated a unique SASP-related prognostic model to assess therapeutic effect and prognosis in patients with AML, which might contribute to understanding the role of SASP in AML and guiding the treatment for AML.

Keywords: acute myeloid leukaemia (AML); immune infiltration; prognosis; risk signature; senescence‐associated secretory phenotype (SASP).

FIGURE 1

Identification of differential expressed SASP‐related genes with prognostic value. (A) expression of differential expressed SASP‐related genes. (B) Univariate Cox regression analysis showed the hazard ratios of 88 SASP‐related genes correlated with AML prognosis (p < 0.05). (C) Intersection of differential expressed SASP‐related genes and prognosis related genes. (D, E) LASSO regression analysis narrowed down the genes to 11.

FIGURE 2

Prognostic analysis of the prognosis model in the training cohort. (A) The distribution of risk score, survival status and expression patterns of each patient. (B) The survival analysis of the two risk groups classified by the signature. (C) ROC curves predicting the 1‐, 3‐, 5‐year OS at 0.77, 0.79, 0.85, respectively.

FIGURE 3

Validation of the signature in the TARGET‐AML cohorts. (A) The distribution of risk score, survival status and expression patterns of each patient. (B) The survival analysis of the two risk groups classified by the signature. (C) ROC curves predicting the 1‐, 3‐, 5‐year OS at 0.64, 0.61, 0.57, respectively.

FIGURE 4

Construction and validation of the nomogram. Univariate (A) and multivariate (B) analyses showed that risk score was an independent prognostic factor in the TCGA‐AML cohort. (C) Nomogram predicting the 1‐, 3‐, 5‐year OS of AML patients. (D) The calibration plot predicting the probability of survival and actual survival rate at 1‐, 3‐, 5‐year OS. (E) ROC curves of nomogram in the TCGA‐AML cohort. (F) ROC curves of nomogram in the TARGET‐AML cohort.

FIGURE 5

Functional enrichment analysis. (A) Gene ontology (GO) biological process (BP) analysis. (B) Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis.

FIGURE 6

Immune cell infiltration and immunotherapy response of low‐ and high‐risk groups. (A) correlation of risk score and immune cell infiltration by ssGSEA. (B) RiskScore and ImmuneScore, StromalScore, ESTIMATEScore and TumorPurity correlation analysis. (C) Expression of PD‐1, PD‐L2, LAG3, CTLA4 and PD‐L1 in the two risk groups. (D) Chemotherapeutic Response in the two risk groups.