Distinct association of RUNX family expression with genetic alterations and clinical outcome in acute myeloid leukemia

2.1 GEPIA dataset

DNMTs expression in bone marrow cells of AML patients and normal controls (normal bone marrow samples from healthy donors) were analyzed by the Gene Expression Profiling Interactive Analysis (GEPIA) web (http://gepia.cancer-pku.cn/), whose data from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) projects [11].

2.2 Patients from TCGA

All the 173 de novo AML patients with RUNXs expression data included in this study were obtained from TCGA datasets [12]. The whole-genome sequencing or whole-exome sequencing, along with RNA and microRNA sequencing and DNA-methylation analysis were used to analyze the gene expression, mutation and methylation in these de novo AML patients [12]. Clinical and molecular characteristics were obtained, including, age, sex, white blood cell (WBC) counts, peripheral blood (PB) blasts, bone marrow (BM) blasts, French-American-British (FAB) subtypes, and the frequencies of genetic mutations as presented in Table 1. After induction chemotherapy for all 173 AML patients, consolidation treatment included chemotherapy for 100 patients and hematopoietic stem cell transplantation (HSCT) for 73 patients.

2.3 Bioinformatics analyses

The details for the identification of microRNAs targeting RUNX3 were reported as our previous study [13, 14].

2.4 Statistical analyses

SPSS 22.0 were used for statistical analyses and figures creation. Mann-Whitney’s U test was used for the comparison of continuous variables, whereas Pearson Chi-square analysis or Fisher exact test was applied for the comparison of categorical variables. The prognostic effect of RUNXs expression was evaluated using disease-free survival (DFS) and overall survival (OS) analyzed though Kaplan-Meier analysis. The two-tailed P value 0.05 in all statistical analyses was defined as statistically significant.

3.1 RUNXs expression in AML

Using the online web GEPIA (http://gepia.cancer-pku.cn/detail.php) [11], we found that all RUNXs (RUNX1, RUNX2 and RUNX3) expression in AML patients from TCGA was significantly increased as compared with The Genotype-Tissue Expression (GTEx) normal bone marrow (BM) samples ( 0.001, Fig. 2.4).

3.2 Clinical characteristics of AML patients with aberrant RUNXs expression

To explore the associations of RUNX1/2/3 expression with clinical characteristics in AML, we further divided these AML patients into two groups based on median level of RUNX1/2/3 expression, respectively. The comparison of clinical characteristics of the AML patients between two groups was summarized in Table 1. High RUNX1 expression was significantly associated with higher peripheral blood (PB) blasts ( P = 0.000), higher incidence of French-American-British (FAB)-M0/M1 subtypes ( P = 0.038 and 0.037, respectively), lower incidence of FAB-M5 subtype ( P = 0.001), and higher frequency of inv(16)/t(16;16) ( P = 0.009). Overexpression of RUNX2 was correlated with lower age ( P = 0.019), higher frequency of FAB-M3 subtype ( P = 0.000), higher incidence of inv(16)/t(16;16) or t(15;17) ( P = 0.001 and 0.000), lower frequency of normal and complex karyotype ( P = 0.022 and 0.029, respectively). Additionally, high RUNX3 expression was markedly related to lower white blood cells (WBC) ( P = 0.000), lower PB blasts ( P = 0.012), higher percentage of FAB-M0/M3 ( P = 0.009 and 0.001, respectively) and t(15;17)/[-7/del(7)]/others/complex karyotypes ( P = 0.003, 0.007, 0.028 and 0.001, respectively), lower incidence of FAB-M4 ( P = 0.012) and t(8;21) or inv(16)/t(16;16) karyotypes ( P = 0.014 and 0.001, respectively).

OPEN IN VIEWER

We further observed the association of RUNXs expression with gene mutations in AML. Although there were less mutational cases ( 5) in several groups, the comparison of genetic mutations of the AML patients between two groups was presented in Table 1. Cases with high RUNX1 expression presented a lower frequency of CEBPA mutation ( P = 0.018), whereas a bit higher incidence of RUNX1 and WT1 mutations ( P = 0.063 and 0.057, respectively) than those with low RUNX1 expression. For RUNX2, high RUNX2 expression was associated with higher incidence of RUNX1, WT1 and KIT mutations ( P = 0.016, 0.009 and 0.064, respectively), and lower frequency of NPM1, TP53 and DNMT3A mutations ( P = 0.001, 0.001 and 0.051, respectively). For RUNX3, overexpression of RUNX3 was significantly correlated with RUNX1, TP53, CEBPA and KRAS mutations ( P = 0.000, 0.005, 0.048 and 0.064, receptively), whereas low expression of RUNX3 was markedly correlated with FLT3 and NPM1 mutations ( P = 0.000 and 0.000, respectively).

Regarding the significant associations, we further compared the RUNX1/2/3 expression level between two groups with and without these gene mutations, which were presented Fig. 2.

3.4 The effect of abnormal RUNXs expression on survival in AML

We next evaluate the prognostic effect of RUNXs expression on survival. By Kaplan-Meier analysis, patients with high RUNX1 expression showed nearly same overall survival (OS) and disease-free survival (DFS) time compared with those with low RUNX1 expression among both whole-cohort and non-M3 AML (Fig. 3a, $P>$ 0.05). Similar results were also observed between groups with low and high RUNX2 expression (Fig. 3b, $P>$ 0.05). However, cases with high expression of RUNX3 presented shorter OS and DFS than those with low expression of RUNX3 among non-M3 AML patients (Fig. 3c, P = 0.024 and 0.058, respectively).

OPEN IN VIEWER

Because RUNX3 overexpression predicted poor clinical outcome in AML, we intended to investigate whether patients with RUNX3 overexpression could benefit from hematopoietic stem cell transplantation (HSCT). We compared OS and DFS between patients with and without HSCT among both high and low RUNX3 expression groups. In high RUNX3 expression groups, patients received HSCT had markedly longer OS and DFS than patients without HSCT among both all AML (Fig. 4, P = 0.002 and 0.007, respectively) and non-M3 AML (Fig. 4, P = 0.000 and 0.000, respectively). In low RUNX3 expression groups, there were no significant differences in OS and DFS between HSCT and non-HSCT groups among both all AML (Fig. 4, P = 0.069 and 0.191, respectively) and non-M3 AML (Fig. 4, P = 0.080 and 0.233, respectively).

OPEN IN VIEWER

To gain insights into the biological function of RUNX3 in AML, we first compared the transcriptomes of high and low RUNX3 expression groups. A total of 835 differentially expressed genes were identified including 653 positively correlated genes and 182 negatively correlated genes (FDR 0.05, | log2 FC | $>$ 1.5; Fig. 5A and B; Supplementary Materials 1). Positively correlated genes such as SETBP1, CD7, and BCL2L10 were reported as proto-oncogenes in AML [15, 16, 17]. Furthermore, the Gene Ontology analysis was also showed in Fig. 5C.

OPEN IN VIEWER

We also compared microRNAs expression between high and low RUNX3 expression groups. A total of 69 differentially expressed microRNAs were identified including 62 positively correlated genes and 7 negatively correlated genes (FDR 0.05, | log2 FC | $>$ 1.5; Fig. 5D; Supplementary Materials 1). MiR-137 and miR-124 were negatively correlated with RUNX3 expression, and reported as an anti-leukemia role in AML biology [18, 19]. Moreover, positively correlated microRNAs such as miR-125b and miR-100 were seen as oncogenic microRNAs in AML [20, 21].