Integrated Multi-Cohort Analysis of Whole-Genome and Transcriptome Data Reveals Interplay between Metabolism and Methylation in Myelodysplastic Syndrome (MDS)

1. Introduction

Myelodysplastic syndrome (MDS) increases the likelihood that individuals will develop diverse undifferentiated hematologic malignancies [1]. While MDS is primarily found in people over 65, it accounts for less than 5% of childhood cancers [2]. Despite the fact that out of every 100,000 individuals under 70, 40–50 have MDS, emphasizing its higher frequency in older age groups, the age-standardized incidence of MDS, a rare illness, ranges from 1.3 to 4.3 per 100,000 person-years and rises steadily with age [3,4]. The current research indicates that 80–90% of individuals with MDS present repetitive changes in multiple genes [5]. Therefore, investigating hereditary changes that engender leukemia and lymphoma is crucial to improving disease outcome predictions. Advances in next-generation sequencing (NGS) have deepened our understanding of the role that genetic changes play in malfunctioning blood formation and the prognosis for MDS patients. For instance, MDS often features primary and secondary genetic changes, with around 1,500 mutations throughout the genome [6]. These genetic changes result in complex interactions that may affect survival rates in certain MDS cases. Identifying these inherent changes can guide the development of tailored treatment approaches for MDS patients.

One of cancer’s hallmarks is a change in metabolism, which is crucial for cellular processes that instigate carcinogenesis [7]. Hematopoietic stem cells (HSCs), the foundation of hematopoiesis, orchestrate various metabolic needs and states when they mature into advanced myeloid and lymphoid cells [3]. HSCs’ metabolic versatility is evident in their high energy demand during growth and maturation, allowing them to transition from a glycolysis-focused metabolism to a mitochondria-centered one [8]. In the 1920s, Warburg et al. identified tumor tissues that exhibited increased aerobic glycolysis and accelerated lactate discharge. This finding, which suggests information about glucose absorption, lactate emission, and oxygen levels, is also pertinent to MDS [3,9]. A transformative approach to MDS entails identifying metabolic gene markers that are prognostically valuable. The nuanced relationships between frequently altered genes and metabolic systems provide a lens into how these genes may counteract cancer development by effectively adjusting various aspects of metabolism.

The mutations that initiate MDS in HSCs are shaped by clinical features, cellular factors, and genetic makeup [10]. For example, the French-American-British (FAB) system from 1982 labeled MDS “refractory anemia” and segmented it into five types according to cellular morphology and the myeloid blast count [11]. This system was predominant for about 20 years. However, the generic divisions of the initial system led the World Health Organization (WHO) to revamp it and emphasize the crucial role of genetic mutations in diagnosing the condition [12]. In 2001, the WHO presented its first categorization based on HSC mutations that initiate MDS [13]. The ensuing updates from 2008 to 2016 enriched this framework by incorporating clinical, morphological, immunophenotypic, and genetic aspects of the condition [13]. By 2022, the WHO had further refined the MDS classifications. The current divisions hinge on genetic variations and morphological features, emphasizing genetic disruptions of clonal hematopoiesis. This detailed system supports accurate diagnosis and risk assessment, allowing for personalized treatment approaches [10].

In our integrated multidisciplinary study, we emphasized the mutational burden and the transcriptomic profile landscape found in MDS patients to evaluate the relevance of the genes and overly active molecular pathways that are linked to MDS metabolism. We pinpointed numerous genes with mutations that are recognized as modules of mutually exclusive mutations, which could amplify cellular metabolism in individuals with MDS. Ultimately, we combined gene mutational data with gene expression data to identify metabolism-associated signature genes related to DNA methylation in MDS, providing potential insights into the biology of the condition.

2. Methodology

This work followed a retrospective, cross-sectional study design using the cBio Cancer Genomics Portal tool and the Gene Expression Omnibus (GSE) to gather clinical, mutational, and transcriptomic information on MDS patients from various datasets and studies [14,15,16,17,18,19]. The genomic data for evaluating changes in potentially useful genes came from 4,260 MDS samples from joint studies that reported cellular genomics data in 2011 held at the University of Tokyo and in Wellcome Trust Sanger Institute and others in ClinicalTrials.gov number NCT00146120, comparative study in the United States, and Wellcome Trust Sanger Institute, Hinxton, United Kingdom all in 2020 [20,21,22,23]. The transcriptomic data found on the GSE database consists of three independent cohorts (GSE114922 in Wellcome Trust Centre for Human Genetics, United Kingdome, GSE63569 in University of Oxford, United Kingdome, and GSE183328 in CIMA, Spain) held in 2015, 2018, and 2022, respectively [14,15,16,17]. All obtained data were parallel with mutations in MDS of clinical significance.

We analyzed these 4,260 MDS patient samples with the cBio platform. This comprehensive analysis allowed us to identify and examine mutation patterns in this group of patients [18,19,20,21,22,23]. For the RNA-seq meta-analysis, we selected three independent cohorts with raw data published in the GSE. We selected only MDS RNA-seq data from the same biological source (CD34+ hematopoietic stem cells) and technical sequencing platform (Illumina). Raw sequencing reads were downloaded from the GSE database with the SRA toolkit. These reads were then mapped to the human genome (hg38). The resulting BAM files were filtered to remove duplicates, leaving only primary aligned reads. Gene expression count matrices were then generated with a gene annotation GTF file downloaded from Gencode release 44 (GRCh38.p14). The DESEQ2 package was used for data normalization and differential gene expression was reviewed afterward. To identify differentially expressed genes, we established comparisons by contrasting MDS samples with healthy samples. Genes were labeled significant if they had an adjusted p-value of less than 0.1 and a log2 fold change of more than 1.5. Upregulated genes were then subjected to pathway enrichment analysis using the reactome database. Scatter plots were generated using the ggplot package.

3. Results

In numerous studies, biomarkers, especially DNA methylation, and mutations in genes such as isocitrate dehydrogenase (NADP(+)) 1 (IDH1) and isocitrate dehydrogenase (NADP(+)) 2 (IDH2) have been spotlighted as fundamental indicators of MDS [24,25]. Such genetic indicators can modulate patients’ reactions to chemotherapy drugs, including decitabine [26]. Considering the complex nature of MDS, recognizing that our comprehension of the routes that these notably mutated genes define remains limited is vital. Our meta-study systematically integrated various genes to reveal the pathways that are relevant to MDS patients. Using the cBio cancer genomics portal tool, we assessed 4,260 MDS patient samples drawn from diverse datasets and articles. This examination highlighted genes with prevalent mutations, such as tet methylcytosine dioxygenase 2 (TET2), DNA methyltransferase 3 alpha (DNMT3A), ASXL transcriptional regulator 1 (ASXL1), splicing factor 3b subunit 1 (SF3B1), serine and arginine-rich splicing factor 2 (SRSF2), RUNX family transcription factor 1 (RUNX1), and nucleophosmin 1 (NPM1) (Figure 1) [18,19,20,21,22,23,27,28,29,30,31,32,33].

Interestingly, we found that the DNMT3A gene mutations displayed mutual exclusivity with multiple other vital genes, as Table 1 shows. This mutual exclusivity pattern suggests that these genes contribute to or manage a common biological function, possibly metabolism. Therefore, concurrent mutations in multiple genes could harm cell systems. Further analysis showed that the DNMT3A gene is linked to DNA methylation, the SRSF2 gene is involved in RNA splicing, the ASXL1 gene contributes to chromatin modification, the JAK2 gene is essential for signal processing, and the TP53 gene is fundamental to both tumor inhibition and transcription factor function. Significantly, except for JAK2, all of these genes are correlated with an unfavorable prognosis [10]. Given their critical roles, many of these genes will likely be integrated into the standard clinical assessment of MDS patients.

Our comprehensive meta-analysis, which pooled independent RNA-seq datasets from various MDS studies, amplified the sample size, thereby enhancing its statistical power and our capacity to pinpoint genes that are differentially expressed and unique to MDS. Using this approach helped us identify novel differentially expressed genes by contrasting the transcriptomes of healthy individuals and those with MDS. The data showed that the interleukin 1 alpha (IL1A), oxidized low-density lipoprotein receptor 1(OLR1), and adrenoceptor alpha 2B (ADRA2B) genes in [34,35,36] were among the most characteristically downregulated genes in the MDS genotype. Notably, the erythroferrone (ERFE) [37] and sulfotransferase family 4A member 1 (SULT4A1) [38] are the most upregulated genes in MDS cells (Figure 2A), which suggests extensive metabolic programming. The pathways enrichment analysis using the Reactome database revealed that, as our mutational analysis showed, the DNA methylation pathway was enriched, as Figure 2B and C show, suggesting that it plays a massive role in gene regulation and epigenetic modifications.

Moreover, Figure 2D shows the ERFE and SULT4A1 genes as novel upregulated genes in the metabolic process, identified with the GO term GO:0008152. According to this analysis, the ERFE gene positively regulates glucose imports and the fatty acid metabolic process, and the SULT4A1 gene contributes to mitochondria-involving biological processes and steroid metabolic processes. Therefore, these results suggest extensive changes to the DNA methylation pathways, which can result from a defective metabolism in MDS.

This study underscores the therapeutic potential of addressing the most commonly mutated genes that were found in 4,260 MDS patient samples gathered from diverse datasets and academic articles. These specific genes could disrupt vital signaling pathways and molecular functions, including metabolism, that play central roles in the well-being and growth of individuals with MDS. Notably, mutations in the DNMT3A gene accounted for a significant fraction (as much as 18%) of the mutations found in MDS cases [39], corroborating our extensive multicohort findings. Mutated DNMT3A has a connection with GSH levels that oppose chemotherapy. A definitive association between DNMT3A mutations and GSH levels in MDS remains to be determined, but GSH plays an essential role as an antioxidant regulating numerous cell activities [40].

Understanding the effect of DNMT3A mutations on metabolic processes in hematologic cancers might offer clarity regarding the cellular strategies that drive tumor growth, persistence, and resistance to chemotherapy. Elevated GSH levels may enhance tumor cells’ adaptability and resilience, potentially leading to drug resistance. Therefore, GSH concentration might present a potential metabolic weakness in MDS cases featuring DNMT3A mutations. The U.S. Food and Drug Administration (FDA) has sanctioned the use of hypomethylating agents (HMAs) to manage acute myeloid leukemia (AML) and MDS [41]. Ongoing clinical trials targeting MDS and their biological focuses are presented in Table 2.

4. Discussion

We delved into the genetic mutation patterns and gene expression profiles reported in independent MDS studies. This approach provided a greater understanding of the biological nature of MDS. Our investigation explored the link between mutations in genes related to MDS metabolism and the subsequent therapeutic outcomes, drawing on data from cBioPortal (MSK, 2020) and Myelodysplasia (UTokyo, Nature 2011). This approach differentiates our work from previous studies focusing on singular genes or a confined group of MDS genes. For example, the patterns of mutual exclusivity in specific genes, as presented in Table 1, underscore their pivotal roles in MDS. The DNMT3A gene, which registered the highest mutation frequency (Figure 1), also indicates potential interplay with other genes, elaborated in Table 1. Such observations illuminated the significant effect that the DNMT3A gene has on the MDS metabolic landscape, possibly including interlinkage with the enumerated genes in a common biological function. Moreover, our results imply that the cells of individuals with MDS might struggle to accommodate concurrent mutations in the genes specified in Table 1.

Interestingly, as mentioned previously, DNMT1 mutations feature prominently among the most frequent MDS mutations. This could account for the surge in clinical trials evaluating DNMT1 inhibitors. However, no recognized clinical trial has been designed to address the metabolic changes marked by elevated GSH levels associated with DNMT1 mutations. The metabolic dynamics of MDS, especially the potential therapeutic target presented by GSH, remain uncharted territory. Conversely, while TET2 gene modifications are prevalent in MDS, no clinical trials targeting these alterations have been registered. There is notable exclusivity between mutations in TET2 and IDH1/2. Studies have shown that introducing 2-hydroxyglutarate, a byproduct of IDH1/2 mutations, into TET2-deficient cells can trigger synthetic lethality [42]. This suggests that MDS patients with TET2 mutations might benefit from therapies based on the oncometabolite 2-hydroxyglutarate. Currently, no clinical trials are specifically aimed at MDS patients with TET2 mutations. Therefore, the development of inhibitors for TET2 mutations may pave the way for innovative therapeutic interventions.

Our study underscores the mutual exclusivity of TET2 and IDH1/2 mutations, hinting at a potential therapeutic target within the myeloid neoplasia dependency pathway. Recent research has begun to elucidate this relationship, revealing that the 2-hydroxyglutarate produced by IDH1/2 mutations can be lethal to TET-deficient cells. Furthermore, TET2 inhibitors demonstrate a selective effect on TET2 mutant hematopoietic precursor cells in vitro and in vivo [42]. These findings emphasize the metabolic influence of TET2-mutant cells, suggesting avenues for developing innovative therapeutic approaches. In addition, our analysis across multiple cohorts using the cBioportal database revealed NPM1 mutations in 16% of MDS patients. Metabolic assessments have also identified NPM1-mutant cases as a unique subset [43]. This discovery underscores a potential metabolic weakness in NPM1-mutant cells, offering additional possibilities for therapeutic interventions due to the implied, pronounced metabolic shift in NPM1-mutated cells.

Our gene expression meta-analysis showed that MDS patients’ transcriptome is associated with deregulation of DNA methylation, as our MDS mutational analysis suggested. In addition, the gene expression meta-analysis highlighted the deregulation of metabolism-related genes (ERFE and SULT4A1). This suggests a mechanism of metabolic control of the epigenome in MDS. Numerous studies have shown that metabolism has a profound effect on DNA methylation in cancer [44,45,46]. The ERFE gene, the primary erythroid regulator of hepcidin, was reportedly involved in alterations in cancer metabolism [47]. SULT4A1 is found within the cytosolic and mitochondrial sub-compartments of both mouse and human brains, hinting at a potential auxiliary function in mitochondrial activity [48]. A study by Hossain et al. indicated that SULT4A1 directly influences mitochondrial functionality and redox equilibrium [49]. These observations highlight the roles that SULT3A1 plays in metabolism. Exploring the MDS profile in the context of metabolic and DNA methylation mutations remains crucial for pinpointing innovative therapeutic interventions.

5. Conclusion

The inherent complexity and varied nature of MDS continue to obscure its origins, which remain to be fully understood. Moreover, exploring MDS patients’ mutation load, especially concerning metabolic processes and methylation, could illuminate their cellular molecular makeup. Such insights could further enhance our grasp on the roles of metabolism and DNA methylation in MDS and aid in devising patient-specific treatment strategies after diagnosis. In summary, this study offers a deeper understanding of how to identify novel MDS-associated metabolic genes and DNA methylation, the potential clinical applications and the possibility of improving therapeutic avenues for MDS patients, and a foundation for prospective clinical trials and investigations.