Myelodysplastic neoplasms (MDS) is a clonal disorder that starts with the initiation of somatic mutations that occur in the genome of the multipotent hematopoietic stem cell (HSC). Mutations that provide growth and survival benefit at the level of the HSC and enhance self-renewal lead to the accumulation of clonal hematopoiesis over time, resulting in abnormal progenitor and precursor cells. Given the selective survival advantage of these initiation events over wild-type cells, these somatic mutations are termed “driver mutations”. Advanced high-throughput sequencing technologies have led to the discovery of recurrent chromosomal abnormalities, mutations that alter the expression of individual genes, and epigenetic abnormalities [16,18,19].

Approximately half of the patients with MDS harbor recurrent chromosomal abnormalities affecting copy number alteration (e.g., deletion, monosomy, or trisomy) or, more rarely, leading to a structural change (e.g., balanced translocation or inversion). The most common MDS-defining chromosomal abnormalities are deletion 5q (10–15%), monosomy 7/deletion 7q (10%), trisomy 8 (10%), and deletion 20q (5%) [20]. Up to 30% of MDS patients exhibit a complex karyotype (≥3 cytogenetic abnormalities), which is associated with a higher risk of progression to AML and a very poor prognosis. Complex karyotypes with more than three abnormalities has been found to be distinct from those with three abnormalities and is associated with even inferior outcome with a median OS of 1.5 vs. 0.7 years [11]. The aforementioned abnormalities, in addition to deletion 12p/addition (12p), isochromosome (17q), monosomy 17/addition or deletion of 17p, and/or idic(X)(q13), define MDS per the international consensus classification, and when detected in AML, would make a diagnosis of AML with myelodysplasia-related cytogenetic abnormalities [9].

Deletion 5q was first described almost 50 years ago and is the most common cytogenetic abnormality that is present in 10–15% of MDS patients, with a higher incidence in therapy-related or secondary MDS [21,22]. If isolated or in the setting of one additional cytogenetic abnormality apart from monosomy 7/del(7q), deletion 5q is associated with a favorable prognosis with a 15% probability to develop transformation in AML after 5 years. Patients often present with various degrees of refractory cytopenia and blast count <5%, and isolated 5q typically presents in older women with a median age of >70 years. Isolated del(5q) has a reported 6-year OS rate of 67% and PFS of 53%, and it has been shown to have high response rates to treatment with lenalidomide for which it has regulatory FDA approval [21,23]. However, when deletion 5q is present in the context of excessive blasts or other cytogenetic abnormalities, it has a sixfold higher rate of progression to AML [24]. Multiple gene loci present on the long arm of chromosome 5 contribute to the clinical picture. For instance, haploinsufficiency of ribosomal protein S14 (RPS14) contributes to the development of dyserythropoeisis via activation of the p53 pathway [25]. Haploinsufficiency of casein kinase 1 alpha 1 (CSNK1A1) plays a role in the initiation of clonal expansion via deregulation of the WNT/beta-catenin pathway [26].

The second most common cytogenetic abnormality in MDS is monosomy 7 or deletion 7q, which occurs in 10% of patients with de novo MDS and in about 50% of patients with treatment-related MDS [27]. Monosomy 7 is more prevalent than deletion 7q and is associated with a worse prognosis. Both monosomy 7 and deletion 7q are caused by completely different mechanisms of chromosome dis-segregation in mitosis versus chromosome rearrangement, respectively. High-throughput sequencing technologies have identified gene mutations associated with chromosome 7 anomalies, including SAMD9, SAMD9L, EZH2, and MLL3 [27,28,29,30].

Trisomy 8 is found in approximately 10% of MDS patients and is considered an intermediate risk with a median OS of 6 years [31]. It is thought to be a secondary or late event in the MDS transformation. While the precise mechanism in the tumorigenesis process remains unclear, gain of chromosome 8 has been shown to confer more resistance to apoptosis by upregulation of antiapoptotic genes present on chromosome 8 as well as overexpression of the MYC oncogene [32].

Deletion 20q occurs in 5% of MDS patients and it often appears as a major clone at diagnosis. While it is considered to have a favorable risk prognosis, the development of deletion 20q as a minor clone during stages of the disease can precede disease progression and is therefore associated with poor prognosis [33]. ASXL1 mutation co-occurs in 30% of the patients with deletion 20q and negatively impacts prognosis [34].

Other less commonly present chromosomal abnormalities include -Y, del(3q), del(9q), 13/del(13q), del(11q)/t(11q), del(12p)/t(12p), 17/del(17p)/i(17q), +19/t(19), and idic(Xq13). The pathogenesis of more rare cytogenetic abnormalities is largely unknown, but the advancements in genetic technology may provide deeper insights into the pathophysiology and predict the prognosis of patients with MDS [35]. As an example, while monosomal karyotype (MK), defined by the presence of at least two separate autosomal monosomies or one monosomy plus one or more structural abnormalities, has been associated with adverse prognosis in AML, Schanz et al. revealed that MK was not independently associated with prognosis in MDS and that the distinction between MK+ and MK- did not add any prognostic information [36]. Their study had over 400 patients, half of whom were identified as having MK. Moreover, while the number of abnormalities (≤4 vs. >4 abnormalities) was associated with OS, occurring more frequently in MK+ compared to MK− MDS, the impact on prognosis was independent of the presence of MK. Therefore, complex karyotype rather than monosomal karyotype is incorporated in the IPSS-R risk model as a clinical variable.

While around half of the MDS patients harbor one or multiple chromosomal changes, more than 90% of the patients have mutations that alter the sequence and function of at least one oncogene or tumor suppressor gene. Using techniques including next-generation sequencing, only a few gene mutations were identified to be present in >10% of the patients, and most patients harbor 2–4 different gene mutations [16]. MDS patients also commonly have abnormal epigenetic profiles, leading to changes in their gene expression. Based on sequencing studies that calculate the VAF of genes, mutations in splicing factors and epigenetic modifiers were found to occur early in the evolution of MDS and mutations in transcription factors were found to occur either as early or late events. Hence, TET2, DNMT3A, SF3B1, ASXL1, TP53, and JAK2 are the most mutated genes underlying CHIP and CCUS [16].

Spliceosomes are formed by five small nuclear ribonucleoproteins (snRNP) and their associated proteins and are important in the process of removing noncoding regions of the mRNA before translation [37]. Alternative splicing occurs in >90% of the protein-coding genes, allowing the production of multiple mRNA transcripts and distinct protein isoforms. Mutations in genes encoding for spliceosomal proteins in MDS lead to aberrant 3′ splice side recognition contributing to tumorigenesis. The most frequently mutated genes are SF3B1, SRSF2, U2AF1, and ZRZR2. They are present in about 60% of MDS patients and they arise mutually exclusive of each other [38,39]. While mutational targets overall are largely similar between MDS and primary AML, mutations in the spliceosomes are overrepresented in MDS.

SF3B1 is the most frequently mutated spliceosome mutation in MDS, found in up to 30% of MDS patients [40,41]. The SF3B1 gene encodes for splicing factor 3b, which is a member of the U2 snRNP complex and is thought to be an initiating genetic event in MDS [41,42]. It is present in >80% of MDS patients with ring sideroblasts (MDS-RS) but rare in other MDS subtypes. SF3B1 is associated with favorable clinical outcomes with a low propensity to progress to AML. Based on the clinical implications in terms of risk stratification and therapeutic decision making, the 2016 WHO classification has included SF3B1-mutant MDS as a diagnostic criterion for MDS-RS with a positive predictive value of 98% [41]. Only 5% of the SF3B1 mutated patients have poor or very poor cytogenetic risk groups [40]. The median OS of patients with mutated SF3B1 is 79 months vs. 53 months in wild-type SF3B1, with progression to AML occurring in 7% of the patients after a median follow-up of more than nine years. In patients with the SF3B1 mutation and MDS-RS with either single or multilineage dysplasia, the median OS is 106 and 82 months, respectively. Notably, SF3B1 in the presence of more than one additional aberration (particularly RUNX1 mutation) or in combination with del(5q) is associated with a dismal prognosis [40]. Currently, luspatercept is approved for MDS-RS for transfusion-dependent anemia post erythropoietin-stimulating agents (ESAs) or for patients who are unlikely to benefit from ESA therapy [43].

The SRSF2 mutation is present in approximately 15% of MDS patients and, in contrast to the SF3B1 mutation, it is associated with a worse prognosis and high transformation rate to AML [44]. SRSF2 encodes for serine/arginine-rich splicing factor 2 protein, and mutated SRSF2 causes alteration of the mRNA recognition, resulting in mis-splicing of key transcriptional regulators [45]. The SRSF2 mutation often presents with dysplastic features of granulopoiesis and megakaryopoiesis. Gene mutation of U2AF1 occurs in 10–15% and results in RNA splicing dysfunction [46]. Studies agree that U2AF1 may predict poor prognosis with a higher risk of leukemic transformation, and U2AF1 is one of the prognostic gene mutations included in the IPSS-M [10]. ZRSR2 is less frequently present in MDS (5–10%) and alters the splice site recognition of the pre-mRNA. The impact of ZRSR2 on clinical outcome remains unknown [44].

DNA methylation (CpG methylation) exerts a key role in normal differentiation and proliferation of the HSC. CpG islands are regions with a high frequency of CpG sites and are regulatory units present in promoter regions of 60–70% of the genes. Changes in DNA methylation contribute to altered gene expression without sequence mutations of the genomic DNA. Several genes that play a role in DNA methylation (e.g., TET2, DNMT3A, IDH1, and IDH2) are frequently mutated in MDS and cause global as well as gene-specific hypermethylation, resulting in silencing of tumor suppressor genes or genes involved in DNA repair. Like mutations involved in RNA splicing, mutations that affect DNA methylation typically occur early in the development of MDS.

As an example, de novo DNA methyltransferase (DNMT3A) is a member of the DNMT family that adds a methyl group to cytosine in CpG dinucleotides. DNMT3A is mutated in 15% of MDS patients and is frequently present in CHIP. It is associated with an increased risk of leukemic evolution and inferior prognosis [10]. Most DNMT3A mutations reside in the catalytic domain of the methyltransferase domain of DNMT3A, especially at the amino acid R882 locus, which results in reduced methyltransferase activity of the protein due to defective DNA binding and impaired CpG recognition [47]. It is mutated in approximately 20% of AML patients, in 5–10% of MDS cases, and in 60% of patients with CHIP [48,49]. Interestingly, while DNMT3A R882 mutations were found to be enriched in AML (~50% of all DNMT3A mutations), they are decreased in frequency in CHIP (~10%) and other myeloid neoplasms including MDS (~25–30%). Moreover, as observed with AML, MDS patients with R882 mutations are found to have a significantly worse overall prognosis and a more rapid progression to leukemia than patients with non-R882 DNMT3A mutations.

In contrast, TET2 leads to hypermethylation via a different mechanism and is mutated in 20–30% of MDS patients. It is responsible for alpha-ketoglutarate (α-KG)-dependent catalyzation of hydroxylation of 5-methyl-cytosine to hydroxymethyl-cytosine, promoting DNA methylation [50]. The prognostic implication of TET2 mutation remains unclear. IDH1 and IDH2 are enzymes that catalyze the oxidative decarboxylation of isocitrate to produce products required for the Krebs cycle, including α-KG. Mutations in IDH1 and IDH2 lead to the conversion of α-KG to an oncometabolite, 2-hydroxyglutarate, inhibiting α-KG-dependent enzymatic reactions such as TET2 DNA hydroxylation. Mutations in IDH1 and IDH2 occur in <10% of patients with MDS but are associated with an increased risk of transformation into AML. Both IDH1 and IDH2 have been shown to be associated with unfavorable prognosis, although the prognosis of IDH2 remains controversial. IDH1, IDH2, and TET2 are mutually exclusive but display an overlapping DNA hypermethylation signature [51]. Unlike TET2, IDH1 and IDH2 mutations are rare in CHIP. TET2 mutations have been shown to have a positive correlation with SRSF2 and ZRSR2 [17]. Targeted therapy using IDH1 and IDH2 inhibitors, including ivosidenib and olutasidenib for IDH1 and enasidenib for IDH2, have been approved for the treatment of AML and are being investigated in MDS [52,53,54].

Polycomb genes encompass a family of protein complexes that have been discovered to impact chromatin structure and histone modification, resulting in the repression of gene transcription [55]. Polycomb proteins function within two multi-subunit protein complexes: polycomb repressive complex 1 (PRC1) and PRC2 with mono-ubiquinate histone H2A lysine 1199 and methylate (di- and tri-) histone H3 lysine 27, respectively. The core PRC2 complex comprises four components, EZH1/2, SUZ12, EED, and RBAP46/48, while the composition of PRC1 complexes exhibits more variability. Mutations involved in these complexes have frequently been identified in MDS, including ASXL1, EZH2, KDM6A, SUZ12, and EED. ASXL1 physically interacts with EZH2 and influences PRC2 recruitment in HSCs, and mutated ASXL1 results in the loss of interaction with the PCR2 complex [56]. KDM6A is an enzyme that facilitates the demethylation of H3K27. Thus, loss-of-function mutations of PRC2 components are thought to deregulate the normal program of hematopoiesis through repression of transcription key genes in hematopoietic stem/progenitor cells, contributing to positive selection.

ASXL1 is the most frequently mutated gene in this category and is mutated in 15–20% of MDS cases [38,57]. Additionally, ASXL1 is frequently mutated in CHIP as well as in 5–10% of AML cases. Mutations in ASXL1 are more likely to coexist with other mutations, except with SF3B1, DNMT3A, and IRF1 mutations for which they have negative correlations. ASXL1 negatively impacts OS and increases the risk of relapse. The second most mutated gene that affects chromatin modification is EZH2, which is mutated in 5–10% of MDS patients. TET2, RUNX1, and ASXL1 are most frequently mutated together with EZH2 [38,58]. EZH2 is located at 7q36.1, and in both AML and MDS, this region is frequently affected by loss of chromosome 7 or deletion 7q, which is associated with adverse outcomes. Indeed, studies have shown that, like ASXL1, EZH2 is an independent unfavorable prognostic factor with progression to AML [15,59].

Another group of mutations in MDS is the group of transcription factors. Transcription factors bind to specific DNA sequences, and mutations have been reported to cause impairment of differentiation and maintenance of the HSC. Somatic mutations are present in 10–15%, and common genes are RUNX1, BCOR, ETV6, GATA2, and CIX1. In addition, they can be present as germline mutations responsible for familial MDS/AML [60].

The RUNX1 transcription factor is a critical regulator of hematopoiesis [61]. Mutation of RUNX1 disrupts the core-binding factor complex, leading to alteration of gene transcription. RUNX1 accounts for about 10% of MDS cases (the third most frequently mutated gene in MDS) and is typically a subclonal mutation associated with unfavorable clinical outcomes and advanced disease. It is a common abnormality in therapy-related MDS [62]. Mutated RUNX1 is frequently accompanied by additional mutations of the genes ASXL1, SRSF2, TET2, SF3B1, and EZH2 and often co-exists with del(7)/del(7q) [61].

BCOR is a transcription factor that is a component of the PRC and encodes for a corepressor of BCL6. BCOR mutation is present in 5% of MDS patients and commonly co-occurs with RUNX1 and DNMT3A mutations. Although the type of mutation may be important, BCOR mutations are associated with unfavorable outcomes [63,64]. ETV6 is only present in less than 5% of the patients, and more than 30 fusion partner genes have been identified in a broad spectrum of hematologic malignancies, mainly T-ALL. In the IPSS-M, ETV6 is significantly associated with worse OS and progression to AML after adjustment for IPSS risk groups (p = 0.04) [10,15]. GATA2 belongs to the GATA family of zinc finger transcription factors that are important for hematopoietic stem cell maintenance and differentiation. It is frequently associated with familial MDS but also occurs as a somatic mutation [65].

The cohesin ring is a conserved multimeric protein complex that is involved in sister chromatic cohesion during cell division, DNA repair, and transcription regulation. It is composed of two structural maintenance heterodimers, SMC1A and SMC3, that form a close loop with RAD21 and STAG1/STAG2 proteins. Moreover, they bind other regulatory molecules including NIPBL, PDS5b, and CTCF [66,67]. Mutations in the cohesin proteins lead to loss of function and have been identified in 10% of MDS patients as well as in other hematologic myeloid malignancies. Mutations lead to the loss of cohesin binding sites on chromatin that allow access to transcription factors. Particularly, STAG2, which is present in about 5% of the patients, has been associated with predicted poor survival [68].

Mutations involved in signal transduction are less commonly associated with MDS compared to AML. Overall, they occur in 5–10% of the patients, with each individual mutation present in <5% of the cases. Activating mutations of tyrosine kinase and/or serine/threonine kinase results in constitutive activation of the JAK-STAT or RAS-MAPK pathway. Examples include JAK2, CBL, NRAS, and NF1 of which CBL, a tumor suppressor with E3 ubiquitin ligase activity, is associated with reduced OS [69,70]. For PTPN11, JAK2, and NF1, no survival impact has been observed so far.

TP53 is a tumor suppressor and a transcription factor and is the most frequently mutated gene in cancer [71]. It is mutated in approximately 10% of de novo MDS cases and in 25% of therapy-related MDS cases as well as in 20% of MDS patients with the 5q-deletion [72]. Moreover, about 50% of TP53-mutated patients have a complex karyotype. TP53 mutation is associated with high-risk MDS, rapid transformation to AML, early relapse, and poor OS. TP53 mutations define a separate disease entity per the 2022 ICC and also define a unique subgroup within patients with a complex karyotype [9,73]. Although most studies look at the impact of the presence/absence of mutated TP53, few studies have investigated the allelic status. Recently, Bernard et al. compared single-hit mutated TP53 to multi-hit mutated TP53 in a large cohort of >3300 MDS patients [74]. Their study found that multi-hit (i.e., meaning that cells have lost both copies of TP53) was more frequently found in patients with a complex karyotype, had fewer co-occurring mutations, and was associated with shorter OS and transformation into AML. Moreover, they also noted that single-hit TP53 MDS patients were indistinguishable from nonmutant TP53 MDS in terms of outcome and response to therapy.