Acute Myeloid Leukemia: New Multiomics Molecular Signatures and Implications for Systems Medicine Diagnostics and Therapeutics Innovation

Abstract

Acute myeloid leukemia (AML) is a common, complex, and multifactorial malignancy of the hematopoietic system. AML diagnosis and treatment outcomes display marked heterogeneity and patient-to-patient variations. To date, AML-related biomarker discovery research has employed single omics inquiries. Multiomics analyses that reconcile and integrate the data streams from multiple levels of the cellular hierarchy, from genes to proteins to metabolites, offer much promise for innovation in AML diagnostics and therapeutics. We report, in this study, a systems medicine and multiomics approach to integrate the AML transcriptome data and reporter biomolecules at the RNA, protein, and metabolite levels using genome-scale biological networks. We utilized two independent transcriptome datasets (GSE5122, GSE8970) in the Gene Expression Omnibus database. We identified new multiomics molecular signatures of relevance to AML: miRNAs (e.g., mir-484 and miR-519d-3p), receptors (ACVR1 and PTPRG), transcription factors (PRDM14 and GATA3), and metabolites (in particular, amino acid derivatives). The differential expression profiles of all reporter biomolecules were crossvalidated in independent RNA-Seq and miRNA-Seq datasets. Notably, we found that PTPRG holds important prognostication potential as evaluated by Kaplan–Meier survival analyses. The multiomics relationships unraveled in this analysis point toward the genomic pathogenesis of AML. These multiomics molecular leads warrant further research and development as potential diagnostic and therapeutic targets.

Introduction

Acute myeloid leukemia (AML) is a hematologic malignancy of hematopoietic progenitor cells (“blasts”) (Yao et al, 2021). The prevalence of substantial heterogeneity and patient-to-patient variability in treatment outcomes, as well as a lack of comprehensive diagnostics that can foresee disease prognosis and the effects of therapy, impede rational AML treatment. AML is a heterogeneous cancer with many molecularly distinct subtypes. Most AML patients coexpress numerous mutations that evolve, making AML a dynamic disease that cannot be effectively categorized into a predefined homogeneity (Yu et al, 2020). Therefore, the current cytogenetics-based classification is limited.

Systems science-driven biomarker discovery and development for AML has been hitherto limited mostly to single omics inquiries. Analysis of the genomic and epigenomic landscape in AML led to the identification of genetic risk groups for AML, which in turn provided the release of the recommendations for risk stratification (Döhner et al, 2017). Molecular responses to therapeutics have been elucidated at the transcriptomic level, and gene expression signatures for responders and nonresponders have been identified (Raponi et al, 2008, 2007).

More recently, mathematical modeling and machine learning approaches came to the forefront to predict survival outcomes of AML patients (Hoffmann et al, 2020; Karami et al, 2021). On the other hand, multiomics techniques that reconcile and integrate data streams from different levels of the biological hierarchy, from genes to proteins to metabolites, offer promising potential for AML discoveries toward systems medicine.

We report, in this study, a systems medicine and multiomics approach to integrate the AML transcriptome data and reporter biomolecules at the RNA, protein, and metabolite levels using genome-scale biological networks. The aim of the present study was to harness the genome-scale biomolecular networks (metabolic, signaling, and transcriptional regulatory) to decipher the interconnected structure of signaling, regulatory, and metabolic processes within responder and nonresponder AML patients with long and short overall survival.

Materials and Methods

Transcriptome datasets and data preprocessing

This work does not include clinical trials. All the data is collected from GEO database.

The design of the present multiomics integrative analyses was directed at deciphering how patients display individual differences in response to treatment and progression of the disease. Datasets that are pertinent to these aims and those that include adult patient groups and at least 25 samples in each dataset were accepted as inclusion criteria.

To analyze gene expression profiles in AML, we screened datasets in Gene Expression Omnibus (GEO) (Barrett et al, 2012) and ArrayExpress (Athar et al, 2019). We used two independent transcriptome datasets from two consecutive studies, GSE5122 (Raponi et al, 2007) and GSE8970 (Raponi et al, 2008), in the GEO database.

The GSE8970 dataset contains both treatment response and overall patient survival data, whereas the GSE5122 dataset contains only treatment response data in addition to gene expression profiles. Therefore, both datasets were used in the comparative analysis of treatment response groups, that is, responders and nonresponders, to identify gene signatures associated with treatment response. On the other hand, the GSE8970 dataset was used to identify genes associated with overall survival. Considering patients with overall survival ±10% of the median of the dataset, patients were divided into two groups, that is, high survival and low survival, and analyzed comparatively.

The farnesyltransferase inhibitor, tipifarnib, was used to treat patients in both datasets. “Treatment response” was defined as patients who had an objective response (complete remission, complete remission with incomplete platelet recovery, or partial remission) or a hematologic response (decrease of >50% of leukemic blast cells in the bone marrow). “Stable disease” was defined as no hematologic response but no disease progression (Raponi et al, 2007). For the purpose of the study, patients with stable disease were considered neither responders nor nonresponders and were therefore not included in the analysis. GSE5122 comprises 58 patients, 5 of whom were removed according to our criteria for treatment response. GSE8970 consists of 34 patients, 8 of whom were removed based on their response status. After preprocessing the data in the GSE5122 and GSE8970 datasets, the male-to-female ratio was 26:27 and 18:8, respectively, and the mean age was 60 and 73, respectively.

Identification of differentially expressed genes

The differentially expressed genes (DEGs) were identified from the normalized expression values by using the Linear Models for Microarray Data (LIMMA) package (version 3.45) (Smyth, 2015) based on a preconstructed statistical analysis approach (Kori et al, 2016). Raw data in each dataset were standardized using the Robust Multi-Array Average expression measure (Bolstad et al, 2013) and Affy package (Gautier et al, 2004) as implemented in the R/Bioconductor (ver.3.6.3) (Gentleman et al, 2004).

To identify DEGs associated with treatment response, comparative analyses of treatment response groups, that is, responders and nonresponders, were performed on both datasets. Similarly, high-survival and low-survival patient groups in GSE8970 dataset were compared to identify DEGs associated with overall survival outcomes.

The Benjamini–Hochberg method was used to control the false discovery rate. Fold changes determined the down and upregulation pattern of each DEG, and at least log FC >1.5 (upregulation) or log FC < −1.5 (downregulation) and adjusted p-value <0.05 were accepted as significant. The information on gene products was obtained from GeneCards: The Human Gene Database (Safran et al, 2010).

Functional enrichment analysis

The functional annotation and enrichment analyses were performed using the ConsensusPathDB database to reveal the biological functions and processes associated with DEGs (Kamburov et al, 2011). Gene Ontology (GO) (The Gene Ontology Consortium, 2015) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) (Kanehisa et al, 2014) were used as resources for molecular functions, biological processes, and pathways. p-Values were obtained through Fisher's Exact test. Benjamini–Hochberg's correction was used as the multiple testing correction technique, and enrichment results with adjusted p < 0.05 were considered statistically significant.

Identification of reporter metabolites associated with AML

The statistically significant changes in gene expression profiles in each dataset were mapped onto the Human Metabolic Reaction (HMR 2.0) (Mardinoglu et al, 2014) model using the reporter metabolites algorithm (Patil and Nielsen, 2005) implemented in the BioMet Toolbox (ver.2.0) (Garcia-Albornoz et al, 2014) to identify reporter metabolites. Benjamini–Hochberg's method was used for the correction of the p-values. Metabolites Biological Role (MBRole) database (ver.2.0) (López-Ibáñez et al, 2016) was used to determine the enrichment of reporter metabolites in metabolic pathways. Adjusted p < 0.05 was considered statistically significant.

Identification of reporter receptors, transcription factors, and miRNAs

The adapted version of the reporter features algorithm (Kori and Arga, 2018) was used to identify reporter receptors, transcription factors (TFs), and miRNAs. The combinatorial human transcriptional regulatory interaction network and Human Transcriptional Regulation Interactions database (HTRIdb) (Bovolenta et al, 2012) were used to obtain experimentally validated TF–target gene interactions, and to create a TF–target gene network.

Likewise, the experimentally confirmed miRNA–target gene interactions were acquired from our previous study (Gov and Arga, 2016; Kori and Arga, 2018) and miRTarbase (release 6.0) (Chou et al, 2016) to generate the miRNA–target gene network. To reconstruct a receptor–protein interaction network, the proteins with receptor activity (GO: 0004872) were screened in the PANTHER (Mi et al, 2019), DAVID (Huang et al, 2007), and GeneCodis (Tabas-Madrid et al, 2012) databases, and the physical protein interactions of these receptors were retrieved from the BIOGRID database (v.3.5.167) (Chatr-Aryamontri et al, 2017).

The p-values were converted to z-scores and integrated with the molecular interaction networks to determine a score for each biomolecule (receptor, transcription factor, or miRNA) based on the z-scores of its network neighbors using the inverse cumulative distribution. Then, the scores from a standard normal distribution were converted to p-values, and statistically significant features (p < 0.05) were designated as reporter biomolecules.

Crossvalidation of the reporter biomolecules

The prognostic power of reporter biomolecules (i.e., hubs, TFs, receptors, and miRNAs) was analyzed at the transcriptome level using an independent RNA-Seq and miRNA-Seq dataset from The Cancer Genome Atlas (TCGA). The TCGA-AML dataset consists of 149 samples with their clinical information (including overall survival data). Considering the expression profiles of the biomolecules of interest, patients were classified into low- and high-risk groups according to their prognostic indices.

Multivariate survival analyses and risk assessments were performed using the SurvExpress tool (Aguirre-Gamboa et al, 2013). Boxplots were used to show differences in expression levels between the risk groups. The t-test was used to estimate the statistical significance of the differences. Kaplan–Meier plots were used to determine the survival signatures of reporter biomolecules. As a cutoff, a log-rank p-value <0.05 was considered statistically significant in all analyses.

Results

The transcriptomic codes of AML

We retrieved gene expression profiles and associated clinical data of two datasets (GSE5122 and GSE8970) from the GEO database. AML patients were classified into two groups, responders and nonresponders, based on their response to tipifarnib treatment, and comparative analyzes were performed on both datasets to identify DEGs associated with treatment response. On the other hand, patients in the GSE8970 dataset were divided into high and low survival groups based on their overall survival, and comparative analyzes were performed to identify DEGs associated with overall survival. In both cases, upregulated and downregulated DEGs were identified.

As a result of comparative analyzes, we identified 1472 upregulated and 1135 downregulated DEGs associated with treatment response. In addition, 583 upregulated and 323 downregulated genes were found to be associated with overall survival of AML patients.

The up and downregulated genes were classified based on their activities and functions. The functional annotations of the upregulated DEGs associated with the treatment response were significantly enriched with cell cycle checkpoints, apelin signaling pathway, MAPK signaling pathway, RHOH GTPase cycle, and leukocyte activation. In contrast, genes associated with rheumatoid arthritis, glucose metabolism, hemostasis, and neutrophil degranulation were downregulated (Fig. 1). The enrichment analyses of upregulated DEGs associated with overall survival indicated signaling receptor binding, immune system development, response to the hormone, and hemoglobin binding. Enrichment of downregulated DEGs was carried out with the establishment of localization, enzyme binding, pancreatic adenocarcinoma pathway, response to wounding, and advanced glycation endproducts-receptor for advanced glycation endproducts (AGE–RAGE) pathway (Fig. 1).

FIG. 1.

Gene set enrichment analysis of the treatment response (GSE5122 and GSE8970) and survival (GSE 8970) datasets. The black bars represent upregulation of the pathway or process, whereas the gray bars represent downregulation.

The metabolic codes of AML

The genome-scale human metabolic network (HMR 2.0) was used to identify reporter metabolites with the integration of transcriptome data of each dataset. The pathway enrichment analyses were performed using MBROLE 2.0 to understand the metabolic activities of reporter metabolites better.

The most remarkable metabolic pathways were amino acid metabolisms such as alanine, asparagine, and glutamine metabolism; cysteine and methionine metabolism; valine, leucine, and isoleucine biosynthesis; and arginine and proline metabolism. Also, arachidonic acid metabolism was associated with several metabolites, such as 10,11-dihydro-20-dihydroxy-LTB4, 10,11-dihydro-20-trihydroxy-LTB4, 12(S)-HETE, 12-oxo-20-dihydroxy-LTB4, 12-oxo-20-hydroxy-LTB4, 12-oxo-20-trihydroxy-LTB4, 20-hydroxy-5S-HETE, 20-hydroxy-LTB5, 20-OH-10,11-dihydro-LTB4, 5(S)-HETE, leukotriene B4, and leukotriene B5 in enrichment analysis of reporter metabolites (Fig. 2).

FIG. 2.

A conceptual summary of reporter metabolites highlighted as potential molecular signatures in AML. Black represents detected metabolites; gray indicates the metabolites which were not detected in the present study. AML, acute myeloid leukemia.

The receptor codes of AML

In reviewing the literature, no data were found on the reporter receptors of AML. It is crucial to identify reporter receptors that initiate transcriptional responses, providing valuable data on identifying effective biomarkers and drug targets. Differential expression patterns of their physically interacting partners were used to establish the relevance of receptors. The results showed that five reporter receptors (ACVR1, CSF3R, EGFR, PTPRC, PTPRG) (Table 1) were common in all datasets.

Table 1.

Reporter Receptors and Transcription Factors of Acute Myeloid Leukemia Datasets (p < 0.05)

Reporter biomolecules (receptors and TFs)	Name	Functioning in human diseases
ACVR1	Activin A receptor type 1	Mutations in the ACVR1 can be detected in endometrial cancers (Zhao and Pritchard, 2016).
CSF3R	Colony-stimulating factor 3 receptor	It is frequently associated with abnormalities of RUNX1, CBFB, CEBPA, and NPM1 genes in AML (Braun et al, 2019; Zhang et al, 2018).
EGFR	Epidermal growth factor receptor	Its expression in AML patients is associated with a poor prognosis (Nath et al, 2020).
PTPRC	Protein tyrosine phosphatase receptor type C	Increased PTPRC expression was connected with a poor prognosis (Ruela-de-Sousa et al, 2010).
PTPRG	Protein tyrosine phosphatase receptor type G	It is frequently deleted in the lung (Galvan et al, 2015) and renal cell carcinoma (Kastury et al, 1996).
AR	Androgen receptor	High expression was also associated with favorable overall survival of adenoid cystic carcinoma, liver hepatocellular carcinoma, AML, ovarian cancer (Hu et al, 2020).
E2F4	E2F transcription factor 4	Essential modulator of AML proliferation and differentiation through regulation of the MAPK signaling pathway in AML (Feng et al, 2020).
ESR1	Estrogen receptor 1	Constituted an independent outcome predictor in AML (Hess et al, 2008).
ETS1	ETS proto-oncogene 1	The proto-oncogene regulated the expression of chemokine and cytokine genes and was necessary for activating the Ras/ERK pathway (Plotnik et al, 2014).
FOXA1	Forkhead box A1	Abnormally expressed in AML patients with FLT3-TKD and NRAS-PM mutations (Neben et al, 2005).
FOXP3	Forkhead box P3	Higher expression was detected in bone marrow samples of AML patients (Wang et al, 2020).
GATA1	GATA-binding protein 1	Regulates differentiation and maturation of erythroid cells and megakaryocytes, and epigenetic deregulation of GATA1 causes the development of AML with NPM1 and FLT-ITD mutations (Drissen et al, 2010; Sportoletti et al, 2019).
GATA2	GATA-binding protein 2	Deletion caused AML blast cell differentiation, increased apoptosis in leukemic stem cells (lscs), enriched for a proapoptotic and myeloid differentiation signature in lscs (Menendez-Gonzalez et al, 2019).
GATA3	GATA-binding protein 3	Involved in multiple tumor-related pathways in B-ALL to impact leukemogenesis and endothelial-specific knockout of GATA3 leads to impairment of hematopoietic stem cell generation (Hou et al, 2017).
PRDM14	PR/SET domain 14	Highly expressed and correlated with poor survival in breast cancer (Casamassimi et al, 2020) and partner with CBFA2T3 on DNA and participates in T-ALL development (Tracey et al, 2019).
TFAP2C	Transcription factor AP-2 gamma	Transcriptionally activates p21 expression, decreases clonogenic survival, and delays breast cancer cell growth (Li et al, 2006).
YBX1	Y-Box-binding protein 1	Involved in erythroid cell development and highly expressed in myeloid leukemic cell lines, and downregulation was observed during myeloid differentiation (Bhullar and Sollars, 2011).

ALL, acute lymphocytic leukemia; AML, acute myeloid leukemia.

The regulatory codes of AML

The TFs and miRNA, which are regulatory elements in transcriptional expression, are associated with the development and progression of various diseases and play vital roles in biological processes. Thus, it is necessary to understand the mechanism of those regulatory elements in different physiological and disease conditions. Regarding the study, we identified the reporter transcriptional regulators by utilizing the combinatorial human transcriptional regulatory interaction network. According to the results, the 12 TFs were detected in all datasets (Table 1). Additionally, the reporter miRNAs were examined, and the 16 miRNAs were common in all datasets (Table 2).

Table 2.

Reporter Micro-RNAs Associated with Acute Myeloid Leukemia Datasets (p < 0.05)

miRNA	Functioning in human diseases
let-7b	Overexpression downregulates AML1-ETO oncogene expression in t(8;21) AML by targeting its 3′UTR (Johnson et al, 2021).
miR-106b-5p	Potential diagnostic biomarker for ischemic stroke and was ranked as one of the most prominently overexpressed miRNAs in AML (Verboon et al, 2016).
miR-122-5p	Suppresses cell migration and invasion in gastric cancer through downregulating DUSP4 (Xu et al, 2018).
miR-124-3p	Circ-HIPK2 acted as a sponge for miR-124-3p in APL cells (Li et al, 2018)
miR-16-5p	Expression positively associated with VEGF in AML (Li et al, 2017).
miR-17-5p	Downregulated in m t(15;17)-positive APL patients (Marcucci et al, 2011), whereas it was significantly overexpressed AML with t(11q23)/MLL (Li et al, 2008).
miR-181a-5p	The elevated expression was connected with longer overall and disease-free survival, indicating as a possible prognostic marker in AML patients treated with intensive induction chemotherapy and auto-stem cell transplantation (Seipel et al, 2020).
miR-192-5p	Suppressed the progression of lung cancer bone metastasis through targeting TRIM44 (Zou et al, 2019).
miR-20a-5p	Promotes colorectal cancer invasion and metastasis (Cheng et al, 2016).
miR-20b-5p	Upregulated and induces the malignant behavior of breast cancer stem cells (Xia et al, 2020) and overexpression is detected in CLL as an indicator of a favorable prognosis (Papageorgiou et al, 2018).
miR-26a-5p	Downregulated, could be a good prognostic marker in patients with bladder cancer (Miyamoto et al, 2016).
miR-335-5p	Tumor suppressor, represses estrogen receptor α in breast cancer cells, implying that it may be involved in estrogen signaling abnormalities (Martin et al, 2017).
miR-484	Diagnostic specificity for CLL (Brown et al, 2017) and high expression was remarkably correlated with a favorable prognosis in ALL (Schotte et al, 2012) and highly overexpressed in the serum of patients with early breast cancer (Zearo et al, 2014).
miR-519d-3p	Overexpression inhibits breast cancer cell growth and motility and arrests those cells into the G0/G1 cell cycle (Li et al, 2018).
miR-92a-3p	Downregulated in AML and negatively correlated with the MCL1 that is highly implicated in the pathogenesis of AML (Gado et al, 2019).
miR-93-5p	Increased in the exosomes of patients with ovarian cancer (Zhang et al, 2019).

CLL, chronic lymphocytic leukemia.

Prognostic performance of the reporter biomolecules

The analysis of the prognostic power of reporter biomolecules using RNA-Seq or miRNA-Seq datasets obtained from independent studies resulted in the datasets' samples divided into low- and high-risk groups. The box plots were used to demonstrate the variations in gene expression levels between the risk groups. Evaluation of prognostic capabilities was performed using the log-rank test and Kaplan–Meier plots based on survival results. The differential expression profiles and valuable prognostic power of the reporter receptor CSF3R (p = 0.04, hazard ratio = 1.52), PTPRG (p = 0.01, hazard ratio = 2.57), and AR (p = 0.01, hazard ratio = 1.9), are shown in Figure 3.

FIG. 3.

The crossvalidation results for reporter biomolecules. The Kaplan–Meier curves demonstrating the prognostic power of (A) CSF3R, (C) PTPRG, and (E) AR. Box plots representing the expression levels of (B) CSF3R, (D) PTPRG, and (F) AR between the low- and high-risk groups. The total size of each group is shown at the top right corner, and the number of censored samples are marked with +.

Discussion

Recent progress in understanding AML pathogenesis has paved the way for developing novel targeted therapies that contribute to moving beyond the current standard treatment and promising effective antileukemic activity. Identifying efficient predictive biomarkers and therapeutic targets could improve the sensitivity and specificity of diagnostic/prognostic methods and develop novel therapeutics and effective drug repurposing strategies. Although there have been numerous studies using healthy control and patient dataset comparison to analyze the mechanism behind the disease, we used a different approach in this study. This research aims to evaluate the datasets from the treatment response perspective to understand better the complexity of drug response and the overall survival outcome of patients during the disease treatment in the scope of the multiomics approach.

Evaluation of response to treatment requires that the patient's condition can be scientifically assessed at baseline and after a specified treatment period. During therapy, treatment response and survival outcomes are significant unpredictable obstacles. Therefore, we have attempted in this study to make comparisons between groups of patients representing differences in either response to treatment or overall survival, rather than making comparisons between healthy individuals and patients. The samples from two datasets were collected from patients before tipifarnib treatment. Differences in expression level in samples taken from patients before treatment will allow us to detect biomarkers that may give us an essential clue in treatment and make more effective pinpointing of overall survival status during the treatment.

Separating patients who can respond from those who are unlikely to respond to rationally planned targeted therapies would help ensure that the correct patients receive the proper treatment, resulting in improved patient care and, in turn, higher response rates and survival rates. Furthermore, mechanisms of action or resistance could be elucidated, and potential new targets for antineoplastic therapy could be identified through the results of this research.

By individual analyses of each transcriptome dataset, hundreds of genes were statistically identified and further analyzed. The analysis indicated that 1472 DEGs were upregulated, and 1135 DEGs were downregulated in treatment response datasets. Furthermore, 582 upregulated DEGs and 346 downregulated DEGs were detected in the overall survival dataset. The upregulated genes of treatment response datasets were shown to encode proteins in various molecular functions such as cell cycle checkpoints, RHOH GTPase cycle, MAPK signaling pathway, apelin signaling pathway, and immune system. In contrast, downregulated genes of those datasets were observed to be associated with the generation of rheumatoid arthritis, glucose metabolism hemostasis, neutrophil degranulation, peptide hormone metabolism, and T-cell receptor (TCR) signaling.

The enrichment analyses demonstrated that most of the upregulated DEGs in the overall survival dataset were associated with signaling receptor binding, response to hormone, hemoglobin binding, heme biosynthesis, metabolism of porphyrins, glutamate binding, activation of AMPA receptors, synaptic plasticity, and calcium signaling pathway. Whereas, the downregulated DEGs were related to the pancreatic adenocarcinoma pathway, the AGE–RAGE pathway, signaling by neurotrophin tyrosine kinases (NTRKs), glioblastoma signaling pathways, and the TGF-β signaling pathway.

The reporter metabolites and substantially enriched metabolic pathways associated with the most critical transcriptional changes were discovered. Amino acid and arachidonic acid metabolism with several metabolites had come into prominence according to reporter metabolite analysis. Most hematologic malignancies have been linked to the metabolism of particular amino acids such as cysteine, glutamine, arginine, and branched-chain amino acids (Tabe et al, 2019).

The previous study showed that l-asparagine treatment inhibited cell proliferation, induced apoptosis in the AML cell lines such as U937, HL-60, and KG-1a, and could be a promising new treatment option for AML (Chen et al, 2020). Glutaminase inhibitor, CB-839, targets the glutamine pathway, prevents glutathione production, and FLT3 inhibitor corporate to induce reactive oxygen species and inhibit AML cell growth through activation of apoptosis (Gregory et al, 2018). BCT-100, a pegylated human recombinant arginase, causes rapid degradation of intracellular and extracellular arginine concentrations, leading to the inhibition of AML blast proliferation and a reduction in AML engraftment (Mussai et al, 2015).

Many kinds of research showed that an engineered human cyst(e)inase enzyme is highly effective in cysteine and cystine degradation in many cancers, such as AML, chronic lymphocytic leukemia (CLL), acute lymphocytic leukemia (ALL), and multiple myeloma (Konopleva et al, 2018). Phosphoglycerate dehydrogenase controls the serine biosynthesis pathway, and its inhibitor WQ-2101 sensitizes AML cells with FLT3 mutation to cytarabine (Bjelosevic et al, 2021). These findings disclose new perspectives into how FLT3 mutations reprogram metabolism in AML and a combination therapy strategy to improve AML treatment.

Certain lipids were shown to be lower in AML cells than in normal leukocytes in many studies. Recent experiments have discovered wide-ranging alterations in the plasma and bone marrow lipidomes of AML patients (Tabe et al, 2020). Sphingolipids, phosphocholines, triglycerides, and cholesterol esters in AML patients were declined in the plasma (Pabst et al, 2017). However, gamma-linolenic acid 18:3 n-6 and 8,11,14-eicosatrienoic acid 20:3 n-6 were increased, although many prostaglandins, such as PGE2 and 15-keto-PGF2α were decreased in the plasma analyses (Pabst et al, 2017).

The findings of this study figured out the importance of arachidonic acid metabolism highlighted with 12 metabolites following amino acid pathways. The plasma arachidonic acid and precursor's correlation with unfavorable prognostic risk was also observed. The importance of the arachidonic acid cascade in AML can be explained by the association of high arachidonic acid concentrations in AML plasma, especially with high BM blasts and high peripheral blasts. Moreover, PGF2α was found to be elevated in AML with low BM and peripheral blast and in the favorable prognostic risk patients (Pabst et al, 2017).

Due to their genetic heterogeneity, myeloblasts and leukemic stem cells may be highly dependent on specific metabolic pathways in AML diseases (Mesbahi et al, 2022). Metabolic alteration in AML cells has been spotlighted as a potential therapeutic target, and promising findings have been shown in preclinical leukemia models using reagents that target metabolic pathways. The activation and the significant roles of the amino acid and arachidonic acid pathway in AML carcinogenesis were demonstrated by many clinical studies, and these results confirm our findings in that perspective.

The reporter features algorithm was adapted to identify reporter molecules, such as receptors, transcription factors, and microRNAs. In this study, five proteins (namely, ACVR1, CSF3R, EGFR, PTPRC, and PTPRG) were identified as reporter receptors in AML. The previous studies reported frequent mutations of CSF3R associated with abnormalities of RUNX1, CBFB, CEBPA, and NPM1 genes in AML (Braun et al, 2019; Zhang et al, 2018). In addition, EGFR expression in AML patients has been associated with a poor prognosis (Nath et al, 2020); however, the exact mechanism of EGFR in AML progression is still unknown and has to be deeply investigated in further studies.

In AML and ALL, increased PTPRC expression is related to a poor prognosis (Ruela-de-Sousa et al, 2010) and elevated in cells expressing an oncogenic FLT3 mutant FLT3-ITD (Arora et al, 2012). PTPRG is frequently deleted in the lung (Galvan et al, 2015) and renal cell carcinoma (Kastury et al, 1996) and hypermethylated in breast cancer (Sherry et al, 2010), childhood ALL (Xiao et al, 2014), and cutaneous T cell lymphoma (van Doorn et al, 2005).

While the expression and function of PTPRG have been the focus of studies and well explained in CML (Drube et al, 2018), PTPRG functions in the AML were not studied. ACVR1/BMPR1 pathway was found to be preferentially engaged to induce MIXL1 in hematopoietic stem cells or progenitors but was not proposed as a prognostic marker for AML (Raymond et al, 2014). Moreover, in the same study, ACVR1/BMPR1 signaling is suggested to stimulate endogenous MIXL1 expression in KG1, OCI-ML2, ML3, and K562 cell lines, and LDN-193189, a ACVR1/BMPR1 inhibitor, is shown to target these cells.

However, MIXL1 expression was unaffected by LDN-193189 in U937 and HL60 cells, suggesting either a MIXL1-independent BMP induction or nonexistence of this pathway in HL60 and U937 cells that are resistant to LDN-193189 (Raymond et al, 2014).

Consequently, to our knowledge, the functional association of receptors, PTPRG and ACVR1, with AML is being proposed for the first time in this study, which needs in vitro or in vivo validation. The differential expression of ACVR1, EGFR, PTPRC, PTPRG, and CSF3R between high and low-risk groups was also crossvalidated here using an independent dataset (TCGA-AML), and the prognostic power of PTPRG and CSF3R was demonstrated (Fig. 3). We showed that the expression of PTPRG and CSF3R are associated with a low risk of AML. Therefore, PTPRG and CSF3R warrant further mechanistic and functional investigation and may have great potential for predicting both treatment response and long overall survival rate for AML patients.

Since transcriptional regulation plays such an essential role in cellular homeostasis, pathogenesis that can result from disturbances in transcriptional regulatory mechanisms is not a surprise. AR, E2F4, ESR1, ETS1, FOXA1, FOXP3, GATA1, GATA2, GATA3, PRDM14, TFAP2C, and YBX1 were common in all datasets, and these transcription factors may play an essential role in the regulation of AML.

High AR expression was reported to be associated with favorable overall survival of AML (Hu et al, 2020). E2F4 was found to be an essential modulator of AML proliferation (Feng et al, 2020). It was found that methylations of tumor suppressor genes and ESR1 constituted an independent outcome predictor in AML (Hess et al, 2008). Elevated levels of ETS1 were detected in AML, but the functional consequences of ETS1 overexpression in AML cells have not been observed (Lulli et al, 2010).

FOXA1 is often found to be abnormally expressed in AML patients with FLT3-TKD and NRAS-PM mutations (Neben et al, 2005). The decreased level of FOXP3 was found to delay the leukemia progression and prolong the survival of AML mice (Wang et al, 2020). The mechanisms in the deletion or deregulation of GATA1 (Drissen et al, 2010) and GATA2 (Menendez-Gonzalez et al, 2019) were found to be associated with the development of AML. However, GATA3 was found to involve multiple tumor-related pathways in B-ALL to impact leukemogenesis. Endothelial-specific knockout of GATA3 leads to impairment of hematopoietic stem cell generation; its role is not identified in AML (Hou et al, 2017).

PRDM14 acts in the maintenance of embryonic stem cells and the reacquisition of pluripotency in primordial germ cells (Tracey et al, 2019). PRDM14 was noticeably expressed and correlated with poor survival in breast cancer (Casamassimi et al, 2020) and can partner with CBFA2T3 on DNA and participate in T-ALL development (Tracey et al, 2019). Although PRDM14 has been widely studied in many cancers, such as lung (Zhang et al, 2013), breast (Taniguchi et al, 2017), and lymphoblastic leukemia (Dettman and Justice, 2008), the association of AML with PRDM14 was not specified. miR-10a is strongly overexpressed in AML cells and modulates TFAP2C by downregulating (Vu et al, 2020). The expression of YBX1 was found to be upregulated in myeloid leukemia cells, coupled with reduced proliferation and impaired leukemic capacity in AML cell lines (Feng et al, 2021).

Our study highlighted PRDM14 and GATA3 as potential effective biomarkers in a patient's treatment response and survival conditions during the disease and can be a therapeutic target in AML. Moreover, we performed survival analyses to test the prognostic capability of the reporter TFs through Kaplan–Meier curves, log-rank p-values, and hazard ratios. The prognostic power of AR was demonstrated in the RNA-Seq dataset obtained by an independent AML study (Fig. 2).

The results of reporter miRNA analysis represented various miRNAs; 16 reporter miRNAs were associated with treatment response. Previous studies have shown that most of the reporter miRNAs are associated with carcinogenesis and have altered expression levels in different cancers, including AML, CLL, ALL, gastric cancer, lung cancer, colorectal cancer, breast cancer, bladder cancer, and ovarian cancer. These include: let-7b (Johnson et al, 2021), miR-106b-5p (Verboon et al, 2016), miR-122-5p (Xu et al, 2018), miR-124-3p (Li et al, 2018b), miR-16-5p (Li et al, 2017), miR-17-5p (Marcucci et al, 2011), miR-181a-5p (Seipel et al, 2020), miR-192-5p (Zou et al, 2019), miR-20a-5p (Cheng et al, 2016), miR-20b-5p (Papageorgiou et al, 2018; Xia et al, 2020), miR-26a-5p (Miyamoto et al, 2016), miR-335-5p (Martin et al, 2017), miR-92a-3p (Gado et al, 2019), and miR-93-5p (Zhang et al, 2017).

Overexpression of miR-519d-3p inhibits breast cancer cell growth and motility and arrests those cells into the G0/G1 cell cycle. By targeting LIMK1, which is a serine/threonine kinase, it plays a crucial role in actin and microtubule dynamics (Li et al, 2018a). However, the association of miR-519d-3p in AML was not found. miR-484 has been implicated in diagnostic specificity for CLL (Brown et al, 2017), and its high expression was remarkably correlated with a favorable prognosis in ALL (Schotte et al, 2012). Besides, miR-484 has already been associated with various cancers, but the association of miR-484 was not identified in AML previously. As a result, the results indicated miR-519d-3p and miR-484 deserve more mechanistic and functional study and highlighted their potential as promising biomarkers for both therapy response and overall survival outcomes in AML.

Conclusions

AML is a complex disease resulting from recurrent cytogenetic and mutational changes in leukemic blasts and epigenetic alterations in progenitor cells. To understand disease progression, omics data must be analyzed at multiple levels in the context of systems biomedicine. In this study, we present molecular codes of AML at RNA (mRNA, miRNA), protein (receptor, TF, enzyme), and metabolite levels using an integrative multiomics approach. The applied methodology yielded significant evidence of novel molecular signatures in AML, such as PTPRG and ACVR1 as receptors, PRDM14 and GATA3 as TFs, miR-519d-3p and miR-484 as miRNAs, which require further validation studies in vitro and in vivo. The amino acid and arachidonic acid metabolisms were also identified as potential therapeutic targets.

In addition, the high prognostic performance of CSF3R, PTPRG, and AR was demonstrated. These biological molecules not only demonstrate the interaction of AML with specific biological processes, but they also have the potential to be used as system-level biomarkers for AML in the context of disease prognosis. Reliable biomarkers that can accurately predict the prognosis of patients with AML will be helpful for therapeutic outcomes. This multiomics study lays the foundation for future research into better therapeutic trials in the treatment of AML.

Footnotes

Author's Contributions

N.K.: Conceptualization, Formal analysis, Writing—Original draft preparation, Review and Editing, Visualization, Supervision. M.K.: Formal analysis, Writing—Original draft preparation. B.T.: Conceptualization, Formal analysis, Supervision. K.Y.A.: Conceptualization, Writing—Original draft preparation, Review and Editing, Supervision. B.K.Y.: Review and Editing, Supervision. O.A.D.: Conceptualization, Writing—Original draft preparation, Review and Editing, Supervision. All authors have read and agreed to the published version of the article.

Author Disclosure Statement

The authors declare they have no conflicting financial interests.

Funding Information

The present study was enabled by a research grant from the Health Institutes of Turkey (TUSEB) with grant number 2019-TA01-4065.

Abbreviations Used

References

Aguirre-Gamboa

, Gomez-Rueda

, Martínez-Ledesma

, et al. SurvExpress: An online biomarker validation tool and database for cancer gene expression data using survival analysis. PLoS One 2013; 8(9):e74250; doi: 10.1371/journal.pone.0074250.

Arora

, Köthe

, van den Eijnden

, et al. Expression of protein-tyrosine phosphatases in acute myeloid leukemia cells: FLT3 ITD sustains high levels of DUSP6 expression. Cell Commun Signal 2012; 10(1):19; doi: 10.1186/1478-811X-10-19.

Athar

, Füllgrabe

, George

, et al. ArrayExpress update—From bulk to single-cell expression data. Nucleic Acids Res 2019; 47(D1):D711–D715; doi: 10.1093/nar/gky964.

Barrett

, Wilhite

, Ledoux

, et al. NCBI GEO: Archive for functional genomics data sets—Update. Nucleic Acids Res 2012; 41(D1):D991–D995; doi: 10.1093/nar/gks1193.

Bhullar

, Sollars

. YBX1 expression and function in early hematopoiesis and leukemic cells. Immunogenetics 2011; 63(6):337–350; doi: 10.1007/s00251-011-0517-9.

Bjelosevic

, Gruber

, Newbold

, et al. Serine biosynthesis is a metabolic vulnerability in FLT3-ITD-driven acute myeloid leukaemia. Cancer Discov 2021; 11(6):1582–1599; doi: 10.1158/2159-8290.CD-20-0738.

Bolstad

, Irizarry

, Astrand

, et al. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 2013; 19(2):185–193; doi: 10.1093/bioinformatics/19.2.185.

Bovolenta

, Acencio

, Lemke

HTRIdb: An open-access database for experimentally verified human transcriptional regulation interactions. BMC Genomics 2012; 13(1):405; doi: 10.1186/1471-2164-13-405.

Braun

, Okhovat

, Coblentz

, et al. Myeloid lineage enhancers drive oncogene synergy in CEBPA/CSF3R mutant acute myeloid leukemia. Nat Commun 2019; 10(1):5455; doi: 10.1038/s41467-019-13364-2.

10.

Brown

, Croce

, Nana-Sinkam

. Clinical and therapeutic applications of microRNA in cancer. In: Translating MicroRNAs to the Clinic. Academic Press: Boston, 2017; pp. 17–37.

11.

Casamassimi

, Rienzo

, di Zazzo

, et al. Multifaceted role of PRDM proteins in human cancer. Int J Mol Sci 2020; 21(7):24648; doi: 10.3390/ijms21072648.

12.

Chatr-Aryamontri

, Oughtred

, Boucher

, et al. The BioGRID interaction database: 2017 update. Nucleic Acids Res 2017; 45(D1):D369–D379; doi: 10.1093/nar/gkw1102.

13.

Chen

, Zhang

, Zeng

, et al. Antiproliferative effects of L asparaginase in acute myeloid leukemia. Exp Ther Med 2020; 20(3):2070–2078; doi: 10.3892/etm.2020.8904.

14.

Cheng

, Zhao

Tang H, et al. MicroRNA-20a-5p promotes colorectal cancer invasion and metastasis by downregulating Smad4. Oncotarget 2016; 7(29):45199–45213; doi: 10.18632/oncotarget.9900.

15.

Chou

, Chang

, Shrestha

, et al. miRTarBase 2016: Updates to the experimentally validated miRNA-target interactions database. Nucleic Acids Res 2016; 44(D1):D239–D247; doi: 10.1093/nar/gkv1258.

16.

Dettman

, Justice

. The zinc finger SET domain gene Prdm14 is overexpressed in lymphoblastic lymphomas with retroviral insertions at Evi32. PLoS One 2008; 3(11):e3823; doi: 10.1371/journal.pone.0003823.

17.

Drissen

, Guyot

, Zhang

, et al. Lineage-specific combinatorial action of enhancers regulates mouse erythroid Gata1 expression. Blood 2010; 115(17):463–3471; doi: 10.1182/blood-2009-07-232876.

18.

Drube

, Ernst

, Pfirrmann

, et al. PTPRG and PTPRC modulate nilotinib response in chronic myeloid leukemia cells. Oncotarget 2018; 9(10):9442–9455; doi: 10.18632/oncotarget.24253.

19.

Döhner

, Estey

, Grimwade

, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood 2017; 129(4):424–447; doi: 10.1182/blood-2016-08-733196.

20.

Feng

, Xie

, Han

, et al. YBX1 is required for maintaining myeloid leukemia cell survival by regulating BCL2 stability in an m6A-dependent manner. Blood 2021; 138(1):71–85; doi: 10.1182/blood.2020009676.

21.

Feng

, Li

, Du

, et al. E2F4 functions as a tumour suppressor in acute myeloid leukaemia via inhibition of the MAPK signalling pathway by binding to EZH2. J Cell Mol Med 2020; 24(3):2157–2168; doi: 10.1111/jcmm.14853.

22.

Gado

, Mousa

, Badawy

, et al. Assessment of the diagnostic potential of miR-29a-3p and miR-92a-3p as circulatory biomarkers in acute myeloid leukemia. Asian Pac J Cancer Prev 2019; 20(12):3625–3633; doi: 10.31557/APJCP.2019.20.12.3625.

23.

Galvan

, Colombo

, Frullanti

, et al. Germline polymorphisms and survival of lung adenocarcinoma patients: A genome-wide study in two European patient series. Int J Cancer 2015; 136(5):E262–E271; doi: 10.1002/ijc.29195.

24.

Garcia-Albornoz

, Thankaswamy-Kosalai

, Nilsson

, et al. BioMet Toolbox 2.0: Genome-wide analysis of metabolism and omics data. Nucleic Acids Res 2014; 42(W1):W175–W181; doi: 10.1093/nar/gku371.

25.

Gautier

, Cope

, Bolstad

, et al. Affy-analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 2004; 20(3):307–315; doi: 10.1093/bioinformatics/btg405.

26.

Gentleman

, Carey

, Bates

, et al. Bioconductor: Open software development for computational biology and bioinformatics. Genome Biol 2004; 5(10):R90; doi: 10.1186/gb-2004-5-10-r80.

27.

Gov

, Arga

. Interactive cooperation and hierarchical operation of microRNA and transcription factor crosstalk in human transcriptional regulatory network. IET Syst Biol 2016; 10(6):219–228; doi: 10.1049/iet-syb.2016.0001.

28.

Gregory

, Nemkov

, Reisz

, et al. Glutaminase inhibition improves FLT3 inhibitor therapy for acute myeloid leukemia. Exp Hematol 2018;58:52–58; doi: 10.1016/j.exphem.2017.09.007.

29.

Hess

, Erram

, Berkhof

, et al. Concurrent methylation of promoters from tumor associated genes predicts outcome in acute myeloid leukemia. Leuk Lymphoma 2008; 49(6):1132–1141; doi: 10.1080/10428190802035990.

30.

Hoffmann

, Thiede

, Glauche

, et al. Differential response to cytotoxic therapy explains treatment dynamics of acute myeloid leukemia patients: Insights from a mathematical modelling approach. J R Soc Interface 2020; 17(170):20200091; doi: 10.1098/rsif.2020.0091.

31.

Hou

, Liao

, Zhang

, et al. Regulatory network of GATA3 in pediatric acute lymphoblastic leukemia. Oncotarget 2017; 8(22):36040–36053; doi: 10.18632/oncotarget.16424.

32.

, Fang

, Xu

, et al. The androgen receptor expression and association with patient's survival in different cancers. Genomics 2020; 112(2):1926–1940; doi: 10.1016/j.ygeno.2019.11.005.

33.

Huang

, Sherman

, Tan

, et al. DAVID Bioinformatics resources: Expanded annotation database and novel algorithms to better extract biology from large gene lists. Nucleic Acids Res 2007; 35(Suppl 2):169–175; doi: 10.1093/nar/gkm415.

34.

Johnson

, Davis

, Zhou

, et al. MicroRNA let-7b downregulates AML1-ETO oncogene expression in t(8;21) AML by targeting its 3′UTR. Exp Hematol Oncol 2021; 10(1):8; doi: 10.1186/s40164-021-00204-7.

35.

Kamburov

, Pentchev

, Galicka

, et al. ConsensusPathDB: Toward a more complete picture of cell biology. Nucleic Acids Res 2011; 39(Suppl 1):D712–D717; doi: 10.1093/nar/gkq1156.

36.

Kanehisa

, Goto

, Sato

, et al. Data, information, knowledge and principle: Back to metabolism in KEGG. Nucleic Acids Res 2014; 42(D1):D199–D205; doi: 10.1093/nar/gkt1076.

37.

Karami

, Akbari

, Moradi

, et al. Survival prognostic factors in patients with acute myeloid leukemia using machine learning techniques. PLoS One 2021; 16(7):e0254976; doi: 10.1371/journal.pone.0254976.

38.

Kastury

, Ohta

, Lasota

, et al. Structure of the human receptor tyrosine phosphatase gamma gene (PTPRG) and relation to the familial RCC t(3;8) chromosome translocation. Genomics 1996; 32(2):225–235; doi: 10.1006/geno.1996.0109.

39.

Konopleva

, Lorenzi

, Ghotbaldini

, et al. Contribution of amino acid metabolism to hematologic malignancies. Blood 2018; 132(Suppl 1):SCI-10; doi: 10.1182/blood-2018-99-109469.

40.

Kori

, Arga

. Potential biomarkers and therapeutic targets in cervical cancer: Insights from the meta-analysis of transcriptomics data within network biomedicine perspective. PLoS One 2018; 13(7):e0200717; doi: 10.1371/journal.pone.0200717.

41.

Kori

, Gov

, Arga

. Molecular signatures of ovarian diseases: Insights from network medicine perspective. Syst Biol Reprod Med 2016; 62(4):266–282; doi: 10.1080/19396368.2016.1197982.

42.

, Song

, Wu

, et al. MiR-519d-3p suppresses breast cancer cell growth and motility via targeting LIM domain kinase 1. Mol Cell Biochem 2018a;444(1):169–178; doi: 10.1007/s11010-017-3241-4.

43.

, Goswami

, Domann

. AP-2γ induces p21 expression, arrests cell cycle, and inhibits the tumor growth of human carcinoma cells. Neoplasia 2006; 8(7):568–577; doi: 10.1593/neo.06367.

44.

, Zhu

, Wang

, et al. Profiling of microRNAs in AML cells following overexpression or silencing of the VEGF gene. Oncol Lett 2017; 13(1):105–110; doi: 10.3892/ol.2016.5412.

45.

, Ma

, Tan

, et al. Profiling and functional analysis of circular RNAs in acute promyelocytic leukemia and their dynamic regulation during all-trans retinoic acid treatment. Cell Death Dis 2018b;9(6):651; doi: 10.1038/s41419-018-0699-2.

46.

, Lu

, Sun

, et al. Distinct microRNA expression profiles in acute myeloid leukemia with common translocations. Proc Natl Acad Sci U S A 2008; 105(40):15535–15540; doi: 10.1073/pnas.0808266105.

47.

López-Ibáñez

, Pazos

, Chagoyen

MBROLE 2.0—Functional enrichment of chemical compounds. Nucleic Acids Res 2016; 44(W1):W201–W204; doi: 10.1093/nar/gkw253.

48.

Lulli

, Romania

, Riccioni

, et al. Transcriptional silencing of the ETS1 oncogene contributes to human granulocytic differentiation. Haematologica 2010; 95(10):1633–1641; doi: 10.3324/haematol.2010.023267.

49.

Marcucci

, Mrózek

, Radmacher

, et al. The prognostic and functional role of microRNAs in acute myeloid leukemia. Blood 2011; 117(4):1121–1129; doi: 10.1182/blood-2010-09-191312.

50.

Mardinoglu

, Agren

, Kampf

, et al. Genome-scale metabolic modelling of hepatocytes reveals serine deficiency in patients with non-alcoholic fatty liver disease. Nat Commun 2014; 5(1):3083; doi: 10.1038/ncomms4083.

51.

Martin

, Conger

, Yan

, et al. MicroRNA-335-5p and -3p synergize to inhibit estrogen receptor alpha expression and promote tamoxifen resistance. FEBS Lett 2017; 591(2):382–392; doi: 10.1002/1873-3468.12538.

52.

Menendez-Gonzalez

, Vukovic

, Abdelfattah

, et al. Gata2 as a crucial regulator of stem cells in adult hematopoiesis and acute myeloid leukemia. Stem Cell Rep 2019; 13(2):291–306; doi: 10.1016/j.stemcr.2019.07.005.

53.

Mesbahi

, Trahair

, Lock

, et al. Exploring the metabolic landscape of AML: From haematopoietic stem cells to myeloblasts and leukaemic stem cells. Front Oncol 2022;12:807266; doi: 10.3389/fonc.2022.807266.

54.

, Muruganujan

, Ebert

, et al. PANTHER version 14: More genomes, a new PANTHER GO-slim and improvements in enrichment analysis tools. Nucleic Acids Res 2019; 47(D1):D419–D426; doi: 10.1093/nar/gky1038.

55.

Miyamoto

, Seki

, Matsushita

, et al. Tumour-suppressive miRNA-26a-5p and miR-26b-5p inhibit cell aggressiveness by regulating PLOD2 in bladder cancer. Br J Cancer 2016; 115(3):354–363; doi: 10.1038/bjc.2016.179.

56.

Mussai

, Egan

, Higginbotham-Jones

, et al. Arginine dependence of acute myeloid leukemia blast proliferation: A novel therapeutic target. Blood 2015; 125(15):2386–2396; doi: 10.1182/blood-2014-09-600643.

57.

Nath

, Bhattacharyya

, Sarma

, et al. The prognostic impact of epidermal growth factor receptor (EGFR) in patients with acute myeloid leukaemia. Indian J Hematol Blood Transfus 2020; 36(4):749–753; doi: 10.1007/s12288-020-01274-z.

58.

Neben

, Schnittger

, Brors

, et al. Distinct gene expression patterns associated with FLT3- and NRAS-activating mutations in acute myeloid leukemia with normal karyotype. Oncogene 2005; 24(9):1580–1588; doi: 10.1038/sj.onc.1208344.

59.

Pabst

, Kortz

, Fiedler

, et al. The plasma lipidome in acute myeloid leukemia at diagnosis in relation to clinical disease features. BBA Clin 2017;7:105–114; doi: 10.1016/j.bbacli.2017.03.002.

60.

Papageorgiou

, Kontos

, Tsiakanikas

, et al. Elevated miR-20b-5p expression in peripheral blood mononuclear cells: A novel, independent molecular biomarker of favorable prognosis in chronic lymphocytic leukemia. Leuk Res 2018;70:1–7; doi: 10.1016/j.leukres.2018.04.014.

61.

Patil

, Nielsen

Uncovering transcriptional regulation of metabolism by using metabolic network topology. Proc Natl Acad Sci U S A 2005; 102(8):2685–2689; doi: 10.1073/pnas.0406811102.

62.

Plotnik

, Budka

, Ferris

, et al. ETS1 is a genome-wide effector of RAS/ERK signaling in epithelial cells. Nucleic Acids Res 2014; 42(19):11928–11940; doi: 10.org/10.1093/nar/gku929.

63.

Raponi

, Harousseau

, Lancet

, et al. Identification of molecular predictors of response in a study of tipifarnib treatment in relapsed and refractory acute myelogenous leukemia. Clin Cancer Res 2007; 13(7):2254–2260; doi: 10.1158/1078-0432.CCR-06-2609.

64.

Raponi

, Lancet

, Fan

, et al. A 2-gene classifier for predicting response to the farnesyltransferase inhibitor tipifarnib in acute myeloid leukemia. Blood 2008; 111(5):2589–2596; doi: 10.1182/blood-2007-09-112730.

65.

Raymond

, Liu

, Liang

, et al. A role for BMP-induced homeobox gene MIXL1 in acute myelogenous leukemia and identification of type I BMP receptor as a potential target for therapy. Oncotarget 2014; 5(24):12675–12693; doi: 10.18632/oncotarget.2564.

66.

Ruela-de-Sousa

, Queiroz

KCS

, Peppelenbosch

, et al. Reversible phosphorylation in haematological malignancies: Potential role for protein tyrosine phosphatases in treatment? Biochim Biophys Acta 2010; 1806(2):287–303; doi: 10.1016/j.bbcan.2010.07.007.

67.

Safran

, Dalah

, Alexander

, et al. GeneCards Version 3: The human gene integrator. Database 2010;2010:baq020; doi: 10.1093/database/baq020.

68.

Schotte

, Pieters

, den Boer

. MicroRNAs in acute leukemia: From biological players to clinical contributors. Leukemia 2012; 26(1):1–12; doi: 10.1038/leu.2011.151.

69.

Seipel

, Messerli

, Wiedemann

, et al. MN1, FOXP1 and hsa-miR-181a-5p as prognostic markers in acute myeloid leukemia patients treated with intensive induction chemotherapy and autologous stem cell transplantation. Leuk Res 2020;89:106296; doi: 10.1016/j.leukres.2020.106296.

70.

Sherry

, Yasuro

, Suling

, et al. Function and regulatory mechanisms of the candidate tumor suppressor receptor protein tyrosine phosphatase gamma (PTPRG) in breast cancer cells. Anticancer Res 2010; 30(6):1937–1946.

71.

Smyth

GK.

limma: Linear models for microarray data. In: Bioinformatics and Computational Biology Solutions Using R and Bioconductor. Statistics for Biology and Health. Springer: New York, 2015; pp. 397–420.

72.

Sportoletti

, Celani

, Verasano

, et al. GATA1 epigenetic deregulation contributes to the development of AML with NPM1 and FLT3-ITD cooperating mutations. Leukemia 2019; 33(7):1827–1832; doi: 10.1038/s41375-019-0399-7.

73.

Tabas-Madrid

, Nogales-Cadenas

, Pascual-Montano

GeneCodis3: A non-redundant and modular enrichment analysis tool for functional genomics. Nucleic Acids Res 2012; 40(W1):W478–W483; doi: 10.1093/nar/gks402.

74.

Tabe

, Konopleva

, Andreeff

Fatty acid metabolism, bone marrow adipocytes, and AML. Front Oncol 2020;10:e00155; doi: 10.3389/fonc.2020.00155.

75.

Tabe

, Lorenzi

, Konopleva

Amino acid metabolism in hematologic malignancies and the era of targeted therapy. Blood 2019; 134(13):1014–1023; doi: 10.1182/blood.2019001034.

76.

Taniguchi

, Hoshino

, Moriya

, et al. Silencing PRDM14 expression by an innovative RNAi therapy inhibits stemness, tumorigenicity, and metastasis of breast cancer. Oncotarget 2017; 8(29):46856–46874; doi: 10.18632/oncotarget.16776.

77.

The Gene Ontology Consortium. Gene Ontology Consortium: Going forward. Nucleic Acids Res 2015; 43(D1):D1049–D1056; doi: 10.1093/nar/gku1179.

78.

Tracey

, Brooke-Bisschop

, Jansen

, et al. The pluripotency regulator PRDM14 requires hematopoietic regulator CBFA2T3 to initiate leukemia in mice. Mol Cancer Res 2019; 17(7):1468–1479; doi: 10.1158/1541-7786.MCR-18-1327.

79.

van Doorn

, Zoutman

, Dijkman

, et al. Epigenetic profiling of cutaneous T-cell lymphoma: Promoter hypermethylation of multiple tumor suppressor genes including BCL7, PTPRG, and p73. J Clin Oncol 2005; 23(17):3886–3896; doi: 10.1200/JCO.2005.11.353.

80.

Verboon

, Obulkasim

, de Rooij

JDE

, et al. MicroRNA-106b∼25 cluster is upregulated in relapsed MLL-rearranged pediatric acute myeloid leukemia. Oncotarget 2016; 7(30):48412–48422; doi: 10.18632/oncotarget.10270.

81.

, Stölzel

, Wang

, et al. miR-10a as a therapeutic target and predictive biomarker for MDM2 inhibition in acute myeloid leukemia. Leukemia 2020; 35(7):1933–1948; doi: 10.1038/s41375-020-01095-z.

82.

Wang

, Feng

, Wang

, et al. Blocking migration of regulatory T cells to leukemic hematopoietic microenvironment delays disease progression in mouse leukemia model. Cancer Lett 2020;469:151–161; doi: 10.1016/j.canlet.2019.10.032.

83.

Xia

, Li

, Qiu

, et al. Oncogenic miR-20b-5p contributes to malignant behaviors of breast cancer stem cells by bidirectionally regulating CCND1 and E2F1. BMC Cancer 2020; 20(1):949; doi: 10.1186/s12885-020-07395-y.

84.

Xiao

, Lee

, Xiao

, et al. PTPRG inhibition by DNA methylation and cooperation with RAS gene activation in childhood acute lymphoblastic leukemia. Int J Cancer 2014; 135(5):1101–1109; doi: 10.1002/ijc.28759.

85.

, Gao

, Wang

, et al. MiR-122-5p inhibits cell migration and invasion in gastric cancer by down-regulating DUSP4. Cancer Biol Ther 2018; 19(5):427–435; doi: 10.1080/15384047.2018.1423925.

86.

Yao

, Li

, Huang

, et al. Leukemia stem cell-bone marrow microenvironment interplay in acute myeloid leukemia development. Exp Hematol Oncol 2021; 10(1):39; doi: 10.1186/s40164-021-00233-2.

87.

, Li

, Zhang

, et al. Clinical implications of recurrent gene mutations in acute myeloid leukemia. Exp Hematol Oncol 2020; 9(1):4; doi: 10.1186/s40164-020-00161-7.

88.

Zearo

, Kim

, Zhu

, et al. MicroRNA-484 is more highly expressed in serum of early breast cancer patients compared to healthy volunteers. BMC Cancer 2014; 14(1)200; doi: 10.1186/1471-2407-14-200.

89.

Zhang

, Xu

, Liu

MicroRNA profiling of plasma exosomes from patients with ovarian cancer using high-throughput sequencing. Oncol Lett 2019; 17(6):5601–5607; doi: 10.3892/ol.2019.10220.

90.

Zhang

, Zhang

, Feng

, et al. Investigating the microRNA-mRNA regulatory network in acute myeloid leukemia. Oncol Lett 2017; 14(4):3981–3988; doi: 10.3892/ol.2017.6686.

91.

Zhang

, Meng

, Dong

, et al. High expression of PRDM14 correlates with cell differentiation and is a novel prognostic marker in resected non-small cell lung cancer. Med Oncol 2013; 30(3):605; doi: 10.1007/s12032-013-0605-9.

92.

Zhang

, Wang

, Chen

, et al. CSF3R Mutations are frequently associated with abnormalities of RUNX1, CBFB, CEBPA, and NPM1 genes in acute myeloid leukemia. Cancer 2018; 124(16):3329–3338; doi: doi.org/10.1002/cncr.31586.

93.

Zhao

, Pritchard

. Inherited disease genetics improves the identification of cancer-associated genes. PLoS Genet 2016; 12(6):e1006081; doi: doi.org/10.1371/journal.pgen.1006081.

94.

Zou

, Zhu

, Lian

, et al. miR-192-5p suppresses the progression of lung cancer bone metastasis by targeting TRIM44. Sci Rep 2019; 9(1):19619; doi: 10.1038/s41598-019-56018-5.