Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Myelodysplastic Syndromes
The myelodysplastic syndromes (MDS) include a group of clonal hematopoietic disorders characterized by ineffective hematopoiesis and blood cytopenias, abnormal blood and bone marrow cell morphology ( Fig. 61.1 ), and a risk of clonal evolution including progression to acute myeloid leukemia (AML). The syndromes have a highly variable clinical course, manifesting in some patients as indolent asymptomatic cytopenias or only necessitating occasional transfusions over a course of many years, and in others as aggressive diseases that rapidly evolve into treatment-refractory acute leukemias and cause death from infection or other complications within a few months.
ELEMENTS OF MYELODYSPLASTIC SYNDROME.
Myelodysplastic syndromes are generally characterized by cytopenias (A) caused by ineffective hematopoiesis (B), which is related to multilineage dysplasia (C). (A) This patient presented with a white blood cell count of 1500/μL, hemoglobin 8.9 g/dL, and platelet count 47,000/μL. (B) The bone marrow was hypercellular, indicating ineffective hematopoiesis. (C) Evidence of trilineage dysplasia was apparent on the peripheral smear. Anisocytosis with macroovalocytes and poikilocytosis is seen in the red blood cells (C, top ). The latter included the somewhat uncommon finding of Cabot ring forms (right) . A large proportion of the granulocytes were severely hypogranular (C, middle left ) compared to some normal forms still in the circulation (right) . Platelets (C, bottom ) were decreased in number, and many were severely hypogranular ( middle , barely visible) compared with residual normal platelets (left) .
Despite this phenotypic variability, individual cases of MDS share many pathophysiologic mechanisms at the molecular and cellular levels, and our collective understanding of these mechanisms has evolved substantially in the last decade. At the same time, although MDS also shares features with AML and other myeloid malignancies, mounting evidence shows that it is appropriately conceived as a distinct class of diseases, and that the older conceptualization of MDS as “pre-leukemia” is overly simplistic.
While our collective understanding of MDS pathophysiology is improving, the condition remains difficult to treat, and outcomes for many patients with MDS are poor. This chapter describes our current understanding of the classification, pathobiology, clinical features, and treatment approaches for this heterogeneous group of disorders.
Central to an understanding of how MDS is diagnosed and classified is the concept of morphologic dysplasia. Although specific criteria for diagnosis of MDS as a distinct clinical syndrome exist and are discussed below, the term “myelodysplasia” refers more generally to an abnormal appearance of hematopoietic precursors during pathologic examination of bone marrow, which may result from many different causes (e.g., drug toxicity, nutritional deficiency (see chapter on nutritional deficiencies), viral infection). The process by which our understanding of MDS has evolved from that of a morphologic oddity associated with cytopenias to that of a distinct clonal disease process has been a century in the making, propelled forward at several points by paradigm shifts in pathologic and molecular characterization of hematologic diseases.
History
In describing the history of MDS as a distinctly categorized disease entity, it is perhaps most useful to separately describe the history of its three major conceptual components. The earliest recognized of these, not surprisingly, was that of cellular dysplasia in association with cytopenias. Although studies of blood morphology in anemic patients were reported in the 19th century not long after the quantitative hemocytometer debuted, the first detailed characterization of dysplasia was probably made by Giovanni di Guglielmo in Pavia in 1923, when he described abnormal erythroid forms in the marrows of patients with various types of cytopenias.
A second major concept, that of ineffective hematopoiesis, developed in the 1930s with the description of “refractory anemia” in patients unresponsive to iron pills or liver extract (the precursor to B 12 supplementation), and later to folic acid supplementation. In the 1950s, others expanded this to include similarly treatment-refractory leukopenia and thrombocytopenia. Importantly, however, ineffective hematopoiesis and marrow dysplasia were initially incompletely linked, and the ineffective hematopoiesis of many early “refractory anemia” patients was probably rooted in other disorders, such as anemia of inflammation due to advanced rheumatologic disease or non-myeloid neoplasia.
The third major component of MDS, and possibly the most prominent, is its conceptualization as a pre-leukemic state. In 1942, a group from France led by Paul Chevallier described a case of anemia evolving into leukemia as “odo-leukemia” (“odo” from the Greek for edge or threshold), which they attributed to benzene exposure. A similar disorder was termed “preleukaemic anemia” by a British hematologist, J.L. Hamilton-Paterson, in 1949. This term was then adopted and expanded by an American group at the University of Chicago including Leon Jacobson in the early 1950s, providing one of the clearest descriptions of a disorder clinically defined by both cytopenias and a risk of leukemia. Other groups published similar descriptions in the 1950s and 1960s under different terminologies.
In 1970, Bernard Dreyfus in Paris and colleagues coined the term refractory anemia with excess blasts (RAEB, described in French as “les anémies réfractaire avec excès de myeloblasts”) and later attempted to further characterize components of RAEB based on morphology and clinical characteristics. The term “hematopoietic dysplasia” was used in the early 1970s to describe a heterogeneous group of disorders distinct from AML, and the term was later simplified to “myelodysplasia.”
One of the key events in the history of myeloid disease classification was the formation of the French-American-British (FAB) Cooperative Group in 1976, which in that year issued a comprehensive and influential categorization of AML based on the 1975 classification of David Galton and John Dacie in London. The FAB classification persisted as the dominant descriptive system for AML until the early 2000s. The original FAB AML formulation from 1976 included two types of “dysmyelopoietic syndromes” that the group warned should not be confused with AML: RAEB and chronic myelomonocytic leukemia (CMML), which had first been described in the late 1960s. However, in 1982 the FAB recognized MDS as a separate group of diseases and accorded a distinct classification system, described in more detail below.
Classification
The currently accepted classification scheme for MDS was initially published by the World Health Organization (WHO) in 2001, revised in 2008, and revised again in 2016 ( Table 61.1 ). Like the classifications of the FAB group, the WHO classification is principally a morphology-based hematopathology system with limited inclusion of cytogenetic and molecular markers, and classifies disease chiefly based on the number of dysplastic lineages and the percentage of marrow blasts.
Table 61.1
2016 World Health Organization Classification of the Adult Myelodysplastic Syndromes
| MDS with Single Lineage Dysplasia (MDS-SLD) | |
| Dysplasia | ≥10% of cells from a single lineage |
| Blasts | <5% in marrow; <1% in peripheral blood; no Auer rods |
| MDS With Single Lineage Dysplasia and With Ring Sideroblasts (MDS-SLD-RS) | |
| Dysplasia | Isolated erythroid dysplasia |
| Blasts | <5% in marrow; <1% in peripheral blood; no Auer rods |
| Notes | ≥15% of erythroid precursors are ring sideroblasts, or if |
| SF3B1 mutation present, ≥5% of erythroid precursors are ring sideroblasts | |
| MDS With Isolated Del(5q) | |
| Dysplasia | Erythroid dysplasia prominent; granulocytic series variable; normal or increased numbers of megakaryocytes with hypolobated nuclei |
| Blasts | <20% (though usually much less) |
| Notes | Must include deletion of chromosome 5q inclusive of band 5q31 as sole chromosomal abnormality or with one other abnormality |
| MDS With Multilineage Dysplasia (MDS-MLD) | |
| Dysplasia | ≥10% of cells from two or more myeloid lineages |
| Blasts | <5% in marrow; <1% in peripheral blood; no Auer rods |
| Notes | Peripheral monocyte count must be <1 × 10 9 /L; ring sideroblasts may be present, as above for MDS-SLD-RS, in which case it is classified asMDS-MLD-RS |
| MDS With Excess Blasts (MDS-EB) | |
| Dysplasia | No specific requirement |
| Blasts | MDS-EB1: 5%–9% in marrow, <5% in peripheral blood, AND no Auer rods |
| MDS-EB-2: 10%–19% in marrow, 5%–19% in peripheral blood, OR Auer rods | |
| Notes | ≥20% blasts is now considered AML |
| Unclassifiable MDS (MDS–U) | |
| Dysplasia | May be minimal, or not meeting criteria for another subtype |
| Blasts | <5% in marrow; no Auer rods |
| Notes | In presence of clonal cytogenetic finding considered diagnostic of MDS |
a Note: excludes refractory cytopenias of childhood (RCC). MDS/myeloproliferative neoplasms such as chronic myelomonocytic leukemia (CMML), and therapy-related MDS/acute myeloid leukemia (AML).
The WHO classification was intended to replace the FAB classification system, which included 5 subtypes of MDS. Those 5 FAB subtypes were refractory anemia (RA, <5% blasts and any cytopenia pattern); refractory anemia with ring sideroblasts (RARS, defined by <5% myeloblasts and ring sideroblasts—formerly called “ringed” —representing ≥15% of erythroid precursors); RAEB (5% to 19% blasts in the marrow); RAEB in transformation to AML (RAEB-t, defined as 20% to 29% blasts in the marrow); and CMML (defined by >1 × 10 9 peripheral monocytes).
The WHO systems bear some similarities to the FAB scheme, but there are several notable changes. First, the WHO system recognizes that MDS may present with isolated leukopenia or thrombocytopenia, albeit uncommonly, hence the change from RA to “MDS with single lineage dysplasia” (MDS-SLD), and the addition of a second distinct class for patients with more than one affected cell line/multilineage dysplasia (MDS-MLD). Second, a mounting body of evidence suggested that MDS with concomitant myeloproliferative features such as CMML is clinically and biologically distinct from MDS without such features, especially when the white count is elevated, and thus warranted separate designation. Third, the unique clinical syndrome associated with low-risk MDS with isolated deletion of chromosome 5q, in combination with its particular responsiveness to treatment with lenalidomide, warranted its designation as a specific clinicopathologic entity. Fourth, in the 2016 edition, the WHO removed the obsolete “refractory anemia” terminology from the 1930s. Finally, the blast threshold for diagnosis of AML instead of MDS was changed with the 2001 classification from 30% to 20%, rendering the FAB designation of RAEB-t obsolete, and RAEB was divided into two categories reflecting different prognoses for patients with 5% to 9% versus 10% to 19% blasts.
The WHO revision improved on certain aspects of the FAB system, including refining prognostic categories. However, artificial boundaries such as the 10% cell threshold in a lineage to define dysplasia, 15% ring sideroblasts to define MDS-RS (unless SF3B1 mutation is present, in which case the threshold is lowered to 5% ring sideroblasts), and 5%/10%/20% blast cutoffs often prove themselves to be of variable relevance, both biologically and clinically. These thresholds fail to capture a number of variables important in MDS, including patient age and comorbidities, kinetics of disease, and cytogenetic and molecular genetic data. Newer classifications are likely to evolve as molecular characterization of MDS improves; for instance, in 2020, an International Working Group (IWG) proposed definition of SF3B1 mutant MDS as a distinct entity.
While the FAB and WHO subgroups carry some prognostic information (for instance, patients with FAB RAEB or WHO RAEB-2 are at higher risk of progression to AML than subgroups without excess blasts ), other systems specifically dedicated to estimating disease risk, such as the International Prognostic Scoring System and its Revised version (IPSS and IPSS-R; see below), are better suited to the task of risk stratification. Future iterations of classification systems will incorporate elements of genetic or clinical data that better group MDS patients based on disease biology, prognosis and natural history, or anticipated responsiveness to treatment modalities. For now, however, the WHO classification remains a key component of several descriptive tools applied to patients with MDS.
Epidemiology
Most available data suggest that MDS is one of the most common hematologic malignancies, although this claim has been difficult to validate until recently due to confusing terminology and incomplete reporting of cases to registries. Cases of MDS were not reported to the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) until 2001, and reporting was inconsistent for several years thereafter. Initial reports to the registry suggested an incidence of only about 10,000 new cases per year. Subsequent comparison of these data with Medicare and insurance claims for MDS suggested that a substantial proportion of MDS cases went unreported to SEER, with an estimated age-adjusted incidence of >5.3 per 100,000, compared to the SEER estimate of 3.3.
The reporting difference was especially stark in patients age 65 and older, where it was estimated that the actual incidence of MDS might actually be close to fourfold higher than what was captured in the database (i.e., 75 vs. 20 per 100,000). Some of the underreporting was likely related to specific criteria for entry into the database: for instance, myeloid malignancies could only be counted once, such that patients presenting with a new diagnosis of secondary AML were usually coded as AML, without mention of the antecedent MDS. Much of the underreporting, however, was likely due to the complexity of diagnosing and classifying MDS, as described above. The rate of reporting may be improving with time, with annual incidence in the SEER database in more recent years estimated at around 20,500 and an age-adjusted incidence of 4.9 per 100,000.
Even the most accurate database would probably underestimate the total global burden of MDS, which is likely present in a substantial percentage of older patients with idiopathic cytopenias who never undergo bone marrow analysis. Although some proportion of this uncaptured population likely has biologically indolent disease that would never require therapy, more thorough cross-sectional studies, particularly in older patients, would help clarify the distinction between this end of the MDS spectrum and more aggressive biology that brings patients to clinical attention.
As referenced above, MDS is by and large a disease of older adults and reflects the inevitable acquisition of genetic mutations by aging hematopoietic progenitor cells. In the United States, the median age at diagnosis is approximately 71 years, and in the absence of a congenital disorder/germline predisposition or exposure to radiation or cytotoxic chemotherapy for another disease, diagnosis prior to the age of 50 is rare. Recent data suggest that a substantial proportion of older adults harbor hematopoietic clones defined by the presence of mutations recurrently found in MDS and AML at a variant allele frequency (VAF) of at least 2%, and that this state of “clonal hematopoiesis of indeterminate potential” (CHIP) progresses to MDS at a rate of 0.5% to 1% per year (see Chapter 19 ). Smaller clones with a VAF < 1% are ubiquitous by middle age. CHIP most commonly is driven by mutations in DNMT3A, TET2 , or ASXL1 , with high allelic heterogeneity, and pre-disposes to cardiovascular events due to clonal monocytes causing NLRP3 inflammasome-driven inflammation in the endothelium, which results in accelerated atherogenesis.
MDS occurs in children much more rarely, at an estimated annual rate of 1 per 1 million. Most cases have excess blasts and evolve to AML over time; other subtypes such as lower-risk MDS including 5q minus syndrome are even less common. Several genetic syndromes confer an increased risk of MDS (see Chapter 30 ); these include Down syndrome, Fanconi anemia, dyskeratosis congenita and other telomeropathies, Shwachman-Bodian-Diamond syndrome, and germline mutations in GATA2 , RUNX1 , ETV6 , and DDX41 among others. In patients with Fanconi anemia, GATA2 mutations, or telomeropathies, MDS arising during young adulthood may be the initial presenting sign of inherited conditions that in other patients affect multiple organ systems and result in diagnosis in early childhood.
Most subtypes of MDS appear to be more common in men than women, with SEER data suggesting respective age-adjusted incidences of 6.7 versus 3.9 per 100,000. One exception to this is MDS with isolated del(5q), which most series show to be slightly more common in women. The reason for these sex differences is unclear; some have postulated a protective factor encoded on the X chromosome, or a similar protective factor on the Y chromosome, which is clonally lost in some cases of MDS, as well as in many older healthy men. Others have suggested that the increased incidence in men could be due to differences in occupational exposures, but this has never been clearly demonstrated. In fact, the only toxic chemical exposure definitively proven to cause MDS is benzene, which is now significantly less prevalent in industrial workplaces than it had been in the past. Other environmental exposures, including cigarette smoking, have been suggested to predispose to MDS. Agent Orange, a defoliant used by the US Army in the Vietnam War and frequently diluted with hydrocarbons to facilitate spraying, has been anecdotally linked to MDS risk, but given the lack of ongoing exposure in the population, a definitive causal connection is difficult to prove.
MDS in the United States and Western Europe are epidemiologically similar, but there are differences between these geographical regions and other parts of the world. In Asia and Eastern Europe, for instance, the average age at MDS diagnosis is younger, and the frequency of both severe cytopenias and risk of progression to leukemia may differ. Some of these differences are likely due to variance in environmental exposures. In Japan, for example, broad population-wide exposure to ionizing radiation from the atomic bombings of Hiroshima and Nagasaki in the 1940s continued to influence MDS incidence well into the 1990s. However, the cause of differences in MDS subtypes, such as the low incidence of MDS with ring sideroblasts in Japan compared with the West, remain unclear.
The two exposures most consistently associated with subsequent development of MDS are ionizing radiation and cytotoxic chemotherapy, and MDS arising in these settings, known as therapy-related MDS or therapy-related (treatment-related) MDS (t-MDS), is frequently characterized by TP53 mutations, multiple chromosomal abnormalities including complex karyotypes (most commonly defined as ≥3 clonal chromosomal anomalies), and frequent transformation to treatment-refractory AML. In the United States, radiation is most frequently encountered as treatment for other cancers, and radiation fields that include the hips or pelvis, the most active sites of hematopoiesis in adults, probably pose the greatest risk. Less commonly, exposure to radiation can occur as the result of occupational exposures (e.g., workers at nuclear reactors) or industrial accidents. Among chemotherapeutic agents, there is a substantial difference in the risk of subsequent MDS. In particular, alkylating agents (e.g., cyclophosphamide and melphalan), platinum-based agents, and fludarabine appear to carry the greatest risk, while the risk with other drugs is lower. PolyADP ribose polymerase (PARP) inhibitors also appear to increase the risk of therapy-related MDS, although many people who receive PARP inhibitors have also received other therapies such that the independent contribution of the PARP inhibitor may be unclear. The rapidly progressive AML seen in association with exposure to topoisomerase inhibitors (e.g., doxorubicin, etoposide) is not typically preceded by MDS.
Pathobiology
Our understanding of the complex pathobiology underlying MDS has evolved substantially since their first description. MDS is now recognized to arise from the acquisition of sequential mutations in hematopoietic stem cells (HSCs) that either confer clonal advantage, impair normal leukocyte function, or both. In some cases, this cell-autonomous expansion of a malignant clone is complemented by non-cell autonomous alterations in the marrow microenvironment and immune surveillance that allow the clone’s expansion. This section details our current understanding of these concepts.
Myelodysplastic Syndromes Cell of Origin
One of the central challenges in understanding the pathogenesis of MDS has been isolating the cell of origin and understanding that cell’s mechanisms of self-renewal and propagation, both of which are necessary for the establishment of a malignant clone. The capacity for self-renewal on its face implies that the origin cell is either a hematopoietic stem cell, and thus possesses intrinsic self-renewal capabilities, or is a more differentiated myeloid progenitor that acquired the ability to self-renew. Additional alterations may confer other characteristics associated with MDS, such as enhanced proliferation, resistance to apoptosis, specific types of morphologic dysplasia, or ineffective hematopoiesis and cytopenias, at which point MDS also becomes clinically evident.
The clonal nature of MDS was first established in the 1980s by studies that showed skewed inactivation of glucose-6-phosphate dehydrogenase (G6PD), an X-linked gene, in the hematopoietic cells of female MDS patients heterozygous for G6PD deficiency. More recent studies have used deep sequencing techniques to track the clonal evolution from MDS into AML and have confirmed that in these cases, the pre-existing MDS is as highly clonal as the resulting secondary AML.
Whether MDS is a disorder of stem or early progenitor cells has been more difficult to prove. Xenotransplant experiments using immunophenotypically defined HSPCs (classically CD34 + CD38 − Lin−) from MDS patients have not shown a striking proliferative or self-renewal advantage for the MDS cells compared to normal controls, and the degree to which these experiments accurately depict the clonal dynamics of MDS in humans is unclear. More recent studies combining immunophenotypic analysis with deep sequencing of clonal mutations in MDS cells have shown that the mutations appear to originate exclusively in the most primitive, stem-cell-like compartment, and others have shown that differential expansion of specific progenitor compartments may vary between different phenotypes and risk profiles of MDS. These studies have led to the conclusion that MDS is, in fact, a disorder of transformed HSCs. The conceptualization of MDS as a stem cell disorder partly explains why it is so refractory to most attempts at conventional therapy.
Genetic Alterations
Like other cancers, the core hypothesis underlying MDS pathogenesis is that the originating clone of MDS increasingly diverges from normal HSCs, and ultimately undergoes frank malignant transformation, through the sequential accumulation of acquired genetic abnormalities. Improvement in our understanding of how these abnormalities are acquired, how they interact with each other, and their impact on pathways affecting proliferation, self-renewal, and differentiation, has been one of the major advancements in the study of MDS over the last decade.
Several types of genetic abnormalities are found in MDS. The first to be recognized (beginning in the 1970s) were cytogenetic abnormalities on standard karyotyping, which are present in about 50% of patients (see Chapter 57 ). A second category of cryptic chromosomal aberrations, including microdeletions and copy number-neutral loss of heterozygosity (CN-LOH), are too small to be detected by karyotype but may be found with fluorescence in-situ hybridization (FISH) or single nucleotide polymorphism (SNP) arrays. The most common type of MDS-associated genetic abnormality, mutations in single genes, has been increasingly well-characterized, and clinical tests for recurrent mutations are becoming increasingly available—though not all mutations have equal clinical relevance. Finally, epigenomic alterations—global aberrations in histone and chromatin modification—are common in MDS, and are often, though not always, associated with mutations in genes involved in epigenetic regulation.
Germline Predisposition
Although MDS is largely a disease of older adults, several inherited genetic disorders predispose to its development as well. While most of these disorders are associated with development of bone marrow failure or MDS in childhood, some (including Shwachman-Diamond syndrome, Familial Platelet Disorder-AML due to germline RUNX1 mutations, and SAMD9 mutations ) can present in young adulthood, and a few (some telomeropathies and germline DDX41 mutations ) most commonly present in middle age or later (see Chapter 30 ). Similarly, although most are associated with other characteristic non-hematologic features, a few (such as some GATA2 mutation disorders, DDX41 mutations, and familial platelet disorder with propensity for AML [FPD-AML]) have only hematologic abnormalities. A partial summary of the most common germline predisposition disorders is given in Table 61.2 .
Table 61.2
Selected Germline Myelodysplastic Syndromes Predisposition Syndromes
| Disorder | Cellular Defect | Gene | Inheritance | Features |
|---|---|---|---|---|
| Fanconi anemia | DNA repair | FANC genes (many) | Variable | Other cancers, facial and skeletal dysmorphologies, short stature, other systemic disorders |
| Diamond-Blackfan anemia | Ribosomal function | Multiple ( RPL and RPS genes, others) | Mostly AD | Short stature, facial dysmorphologies, cardiac/renal abnormalities |
| Shwachman-Diamond syndrome | Ribosomal function | SBDS ( most common ), DNAJC21, EFL1 | AR | Short stature, exocrine pancreatic deficiency, immunodeficiency |
| Telomeropathies | Telomere maintenance | DKC1, TERT, TERC , other telomerase complex components | Variable |
|
| FPD-AML | Abnormal RUNX1 function | RUNX1 | AD | Thrombocytopenia, platelet dysfunction |
| MonoMAC, Emberger syndrome | Abnormal GATA2 function | GATA2 | AD | MonoMAC: monocytopenia, hearing loss, warts, atypical mycobacterial infections; Emberger: lymphedema |
| SAMD9 mutation | Abnormal SAMD9 function | SAMD9 | AD | Short stature, adrenal hypoplasia, infections |
| Li-Fraumeni syndrome | TP53 defects | TP53 | AD | Other cancers (breast, osteosarcoma, soft tissue sarcoma, CNS, adrenocortical carcinoma, etc.) |
| DDX41 mutation | Abnormal DDX41 function | DDX41 | AD | Other hematologic cancers (AML, CML, lymphoma); presents in middle/older age |
AD , Autosomal dominant; AR , autosomal recessive; FPD-AML , familial platelet disorder with propensity for AML.
Somatic Mutations
Of the types of genetic abnormalities found in MDS, acquired mutations in individual genes are the most recently recognized ( Table 61.3 ); they are also the most frequent, currently estimated to be present in at least 80% of MDS patients. Most mutations are acquired randomly, either spontaneously (e.g., deamination of methylated cytosine to thymine), during DNA replication prior to cell division, or during DNA repair; mutations accumulate with age.
Table 61.3
Genes Somatically Mutated in MDS
| Gene | Frequency (%) | Notes | |
|---|---|---|---|
| Originating | TET2 |
|
|
| DNMT3A | 8–13 | Often seen in older people without MDS (clonal hematopoiesis) | |
| ASXL1 |
|
|
|
| SF3B1 |
|
Strong association with RS | |
| SRSF2 |
|
Enriched in CMML | |
| U2AF1 |
|
Association with del(20q) | |
| TP53 | 10–12 | Association with complex karyotype, therapy-related disease, poor prognosis | |
| PPM1D | <5 | Association with therapy-related disease Not clearly a true driver | |
| Intermediate | RUNX1 | 10–15 | Can be somatic or germline |
| BCOR | 5–10 | Frequently associated with RAS pathway mutations | |
| STAG2 | 5–10 | Cohesin class mutations enriched in high-risk MDS and secondary AML | |
| EZH2 |
|
May be functionally involved in some cases of 7q− | |
| IDH1/2 | <5 | More frequent in AML | |
| ATRX | Rare | Associated with acquired thalassemia | |
| Transforming mutations | NRAS | 5–10 | Seen in progression to AML |
| KRAS | 5–10 | Seen in progression to AML | |
| NF1 | <5 | Can occasionally be germline | |
| PTPN11 | <5 | More common in JMML | |
| BRAF | Rare | Also seen in hairy cell leukemia |
AML , Acute myeloid leukemia; CMML , chronic myelomonocytic leukemia; GCPR , G-coupled protein receptor; IDH , isocitrate dehydrogenase; JMML , juvenile myelomonocytic leukemia; MDS , myelodysplastic syndromes; MPN , myeloproliferative neoplasm; RS , ring sideroblasts.
One of the central challenges of understanding the genetics of MDS has been determining which mutations contribute to the pathogenesis of the disease and which do not. While HSCs accumulate many mutations over their lifetimes, the majority are of no pathogenic consequence and are thus termed “passenger” mutations. Meanwhile, HSCs acquire nonsynonymous exonic mutations with translational consequences (“coding” mutations) at a rate of about one mutation per decade, though most of these will not affect clonal dynamics or cellular function. Since the incidence of MDS increases with age, HSCs from an average MDS patient might typically contain between 5 and 10 coding mutations, though the actual number varies widely between patients, against a background of hundreds of non-coding or synonymous single nucleotide variations (SNVs). Only a minority of coding mutations, the “drivers,” actually contribute to the development of the MDS.
Although the recurrently mutated genes in MDS can be characterized in several ways, one of the simplest strategies categorizes them based on their typical position in MDS ontogeny. In this model, mutations that occur early in MDS pathogenesis, such as those in splicing factors, methylation enzymes, and certain epigenetic modifiers, are capable of driving clonal expansion, but their other phenotypic consequences are variable. Mutations that occur later in MDS development may not be capable of driving clonal expansion on their own, but their effects may complement or amplify the originating mutations, and they may be associated with clinical changes, such as worsening of cytopenias. Finally, the acquisition of activating mutations, especially within the RAS pathway, drives transformation to AML. This section is a brief summary of our current understanding of some of the most important genes and pathways that are recurrently deranged in MDS.
Originating Mutations
The unifying feature of these mutations is that they occur early in MDS pathogenesis, which can be either inferred by high VAF indicating presence within the majority of cells in an MDS clone, or from analysis of sequential samples from single individuals. The same mutations are also commonly found in clonal hematopoiesis, an asymptomatic condition in which driver mutations of MDS and AML are found in the blood of individuals with normal blood counts and without known hematologic malignancies, which shows that they are capable of driving clonal expansion as sole lesions.
Some of the common originating MDS mutations, including TET2, DNMT3A , and ASXL1 , do not have significant phenotypic consequences beyond conferring a selective advantage to mutant HSCs. Splicing factor mutations (e.g., SF3B1, SRSF2, U2AF1 ), in contrast, are frequently associated with development of cytopenias and morphologic dysplasia on bone marrow examination. Somatic mutations in TP53 , which is responsible for integrating the cellular response to DNA damage, constitute a unique category of high-risk MDS that frequently, though not always, develops after exposure to cytotoxic therapy and is rarely associated with other cooperating mutations.
TET2
TET2 , which encodes a member of the Ten-Eleven Translocation gene family, is mutated in 20% to 30% of all MDS cases and in 40% to 50% of CMML. The TET2 protein is a methylcytosine oxygenase responsible for converting 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC) using iron and α-ketoglutarate (α-KG, produced by IDH1 and IDH2), and for further oxidizing 5hmC to 5-formyl- and 5-carboxycytosine. These reactions contribute to demethylation through base excision repair back to unmodified cytosine. Mutations in TET2 tend to be inactivating frameshift or nonsense mutations or specific missense substitutions predicted to lead to abrogation of protein function, and many patients are either compound heterozygotes or have UPD at chromosome 4q, leading to effective abrogation of TET2 function. Indeed, patients with TET2 mutations have been shown to have globally altered methylation profiles.
DNMT3A
The DNMT3A gene consists of 29 exons and encodes a 908-amino acid protein that, along with DNMT3B, is one of the two enzymes responsible for de novo CpG methylation independent of replication, whereas a third methyltransferase, DNMT1, is responsible for maintenance of baseline hemi-methylation during active replication. Only DNMT3A mutations, however, have been found to occur recurrently in myeloid malignancies, perhaps suggesting differential expression in hematopoietic cells. DNMT3A mutations in patients with hematologic malignancies, which include truncating mutations and SNVs in key functional domains, are associated with reduced enzymatic function. For unclear reasons, DNMT3A mutations are relatively less common in MDS (5% to 15%) than they are in AML (30% to 50%) or clonal hematopoiesis (60% to 80%).
ASXL1
ASXL1 codes for a polycomb chromatin-binding protein and is involved in epigenetic regulation of gene expression. It acts as a coactivator of the retinoic acid receptor and directly interacts with chemical modifiers of histones (e.g., NCOA1, a histone acetyltransferase, and LSD1, a histone demethylase). ASXL1 mutations occur in about 10% to 29% of total MDS and myeloproliferative neoplasm (MPN) patients and 40% of CMML patients. Pathogenic ASXL1 mutations are exclusively truncating variants in the C-terminal portion of the protein. In studies of MDS cohorts, investigators have observed that ASXL1 mutations are rarely associated with DNTM3A and JAK2 mutations but frequently co-occur with TET2 , SRSF2 , and RUNX1 abnormalities.
SF3B1
SF3B1 encodes the component of the U2 small nuclear riboprotein complex (snRNP) responsible for 3′ branch site recognition and is the most frequently mutated splicing factor gene. SF3B1 mutations can be found in about 20% to 30% of all MDS patients. They have a particularly strong association with ring sideroblast morphology, with mutations found in >60% to 85% of patients with MDS-MLD-RS, MDS-SLD-RS, or MDS/MPN-RS-T, as well as a substantial proportion of other myeloid neoplasms in which ring sideroblasts can be found. The most common mutations are heterozygous missense substitutions that change lysine to glutamate at codon 700 (K700E), with other less common hotspots in the same vicinity such as K666. SF3B1 mutations tend not to be associated with complex karyotypes or poor-prognosis mutations. They are clinically associated with isolated, transfusion-dependent anemia, preserved WBC and platelet counts, and a lower risk of progression to AML. In cross-sectional studies of MDS patients, SF3B1 mutations appear to confer a relatively good prognosis, and they are relatively less common in high-risk MDS cohorts, such as those undergoing allogeneic hematopoietic stem cell transplant.
SRSF2
SRSF2 encodes a member of the serine/arginine (SR)-rich family of pre-mRNA splicing factors that interacts with the U2 and U1 components of the spliceosome. After SF3B1 , it is the second-most commonly mutated splicing factor, with mutations present in 10% to 15% of MDS and 40% of CMML patients. Almost all mutations are heterozygous missense substitutions for proline at codon 95, while a small minority of in-frame insertions or deletions affects the same region. SRSF2 mutations co-occur with several other mutations, many of which are also frequently found in CMML, including TET2 , ASXL1 , RUNX1 , and STAG2 . In contradistinction to patients with SF3B1 mutations, patients with SRSF2 mutations tend to have more dysplasia in the granulocytic lineage and less in the erythroid lineage, and patients consequently tend to have a less prominent transfusion requirement, more enrichment in the RAEB subtypes, and, at least in some studies, a greater risk of progression to AML.
U2AF1
U2AF1 encodes an auxiliary factor in the U2 spliceosome that is responsible for recognizing the AG splice acceptor dinucleotide at the 3′ end of introns. U2AF1 mutations occur in about 10% to 15% of patients with MDS. Similar to other commonly mutated splicing factor genes, the two most common mutations are both heterozygous missense substitutions, again implying a dominant negative effect, and both appear to alter sequence specificity of pre-mRNA binding and splicing. The two mutations are in separate zinc finger DNA binding domains, one at codon 34 toward the N terminal of the protein, and one at codon 157 toward the C terminal.
TP53
TP53 mutations, which result in inactivation of the DNA damage response (DDR) pathway, define a unique evolutionary pathway in MDS. TP53 mutations occur in about 10% to 15% of MDS patients and are universally associated with a poor prognosis. TP53 mutations are often not accompanied by other cooperating point mutations, suggesting that they may be able to drive MDS development as sole lesions. However, they have been shown in cell culture and animal models to be disadvantageous relative to wildtype cells in the absence of chemotherapy or radiation. Consistent with these findings, they are enriched in cases of therapy-related MDS and are often associated with complex and monosomal karyotypes that would likely not be tolerated in cells with functioning DDR pathways. They also frequently co-occur with del(5q) and may represent a progression pathway for patients with 5q− syndrome, including those treated with lenalidomide.
PPM1D
PPM1D encodes a phosphatase that negatively regulates multiple components of the DDR pathway, including p53, ATM, ATR, CHK1, and CHK2, as well as the MAP kinase pathway and the NFKB pathway. Under normal conditions, its role is to restore homeostasis after periods of cellular stress. Pathogenic PPM1D mutations truncate the protein near the C-terminus and confer a gain of phosphatase function, thus downregulating the DDR and partially phenocopying TP53 mutations. PPM1D mutations are frequently found in the blood of adults who have previously received cytotoxic therapy. They have also been found in MDS patients, where they are strongly associated with prior therapy. In MDS, PPM1D is frequently co-mutated with TP53 and does not on its own appear to have a strong correlation either with prognosis or with other phenotypic features, such as complex cytogenetics. It is therefore not clear whether PPM1D mutations actually drive MDS development or whether they mark bystander HSC clones that expand in the context of cytotoxic therapy.
Intermediate Mutations
The common feature of these mutations is that, while they are frequent in MDS, they almost never appear as sole lesions, are almost never found in clonal hematopoiesis, and usually appear to have subclonal relationships to founding mutations based on allele fraction. It is thus likely that most of these mutations are incapable of driving clonal expansion on their own but play important cooperative roles in MDS pathogenesis.
EZH2
EZH2 encodes the catalytic subunit of protein repressive complex 2 (PRC2), which promotes the di- and tri-methylation of lysine 27 on histone 3 (H3K27). Locally, methylated H3K27 results in closed chromatin and transcriptional repression, and global H3K27 tri-methylation in particular is associated with reduced pluripotency and cellular senescence, suggesting a role for EZH2 in regulation of cell fate. EZH2 resides on the long arm of chromosome 7, and its loss has been hypothesized to be at least part of the reason 7q− is a deleterious cytogenetic occurrence in MDS, though not all 7q deletions affect the EZH2 locus. Mutations in EZH2 itself are found in 5% to 10% of patients with MDS and 20% to 30% of patients with CMML. Consistent with the model of EZH2 as a negative regulator of pluripotency and survival, and in contrast to the gain of function mutations in lymphoproliferative neoplasms, EZH2 mutations in MDS tend either to be inactivating frameshift or nonsense mutations, or missense mutations concentrated in the gene’s SET domain, which is critical for DNA binding.
RUNX1
RUNX1 (formerly known as AML1 ) is a transcription factor gene commonly mutated in MDS, and its biology is complex. It encodes the alpha subunit of the core binding transcription factor (CBF) and is involved in determining the lineage fate of hematopoietic stem cells. RUNX1 was initially identified as one of the genes involved in two different common pathogenic translocations: t(8;21), found in AML, and t(12;21), found in ALL. Subsequently, germline point mutations in RUNX1 were identified in autosomal dominant FPD-AML, and later as somatic events in both sporadic AML and MDS. Reflecting the complexity of RUNX1 biology, mutations occur throughout the gene, can be either monoallelic or biallelic, and can be frameshift insertions or deletions or nonsense or missense substitutions. However, most mutations appear to have an inactivating effect on RUNX1 function, either by affecting the DNA-binding RUNT domain or by disrupting the C-terminal protein interaction domain. Many of the remaining mutations outside these regions appear to affect RUNX1 interactions with epigenetic regulators like MLL, thereby affecting histone methylation.
BCOR
This gene encodes a corepressor of BCL-6, a transcription factor required for germinal B-cell development. However, it also has a non-canonical role as a component of the PRC1.1 complex, which monoubiquitinates the H2A histone component and leads to epigenetic silencing. Truncating BCOR mutations are found in 5% to 10% of MDS cases, likely due to disruption of PRC1.1 function. They frequently co-occur with TET2 mutations, a combination that in animal models has been shown to induce a rapidly fatal MDS due to derepression of key myeloid and HOXA genes, leading to myeloid skewing and clonal advantage, as well as simultaneous induction of p53 target genes in erythroblasts, which causes dyserythropoiesis and anemia.
ETV6
ETV6 encodes an ets-like transcription factor with mostly repressive activity. It is situated on the short arm of chromosome 12, and its role in hematologic malignancy has been best characterized by its involvement in recurrent translocations, including t(3;12)(q26;p13) and deletions of 12p. In MDS, however, both missense and inactivating frameshift point mutations have been described as well. ETV6 mutations are relatively rare events in MDS, occurring in at most 5% of cases ; recently, familial cases of MDS and AML due to inherited ETV6 mutations have also been described.
Cohesin Complex Genes
Genes encoding members of the cohesin complex family, including STAG2 , SMC3 , RAD21 , and SMC1A , are each mutated in a small minority of MDS cases, but collectively, they can be found in about 10% of MDS. They have multiple physiologic roles, including the promotion of chromosomal stability during mitosis, as well as the generation of chromatin looping structures that enable promoter interactions with distal enhancer elements. Their role in MDS pathogenesis is largely driven by disruption of these epigenetic functions. STAG2 mutations, which are the most common of this group, appear to have specific interactions with other myeloid transcription factors, including RUNX1 .
Transforming Mutations
Mutations in tyrosine kinase and growth factor receptor genes are typically associated with pro-proliferative signals and occur in a wide range of myeloid malignancies, including AML ( FLT3 ), myeloproliferative neoplasms ( JAK2 and MPL ), and mast cell disorders ( KIT ). While they do occur in MDS, they are usually late, subclonal events that often mark progression to secondary AML. Activating mutations in NRAS are the most frequent tyrosine kinase mutations found in MDS, but still only occur in about 5% of cases and as above tend to begin as subclonal, pro-proliferative events that frequently drive the transition to AML.
Karyotypic Abnormalities
Chromosomal abnormalities, larger-scale genetic aberrations that can be detected on either karyotype or FISH, are also common events in MDS (see Chapter 70 ). The most common types of abnormalities in MDS are deletions or duplications of very large chromosomal regions; as opposed to some other hematologic cancers, translocations and inversions are less common. Several of the most common karyotypic abnormalities have been shown to have prognostic value, and one—del(5q)—has enough unique biologic features to be its own subclassification within the WHO criteria.
Del(5q)
Interstitial deletion of the long arm of chromosome 5 (del(5q)) is the most common chromosomal abnormality in MDS, and del(5q) is the only karyotypically-defined subtype recognized by the WHO. Although the specific region affected varies between patients, there are two commonly deleted regions (CDRs), one on 5q31.1 and the other at 5q32–33.3, with most patients having a deletion that includes both CDRs. MDS patients in whom del(5q) is the sole karyotypic abnormality often display the “5q-minus syndrome,” which is clinically characterized by anemia, normal or elevated platelet count, female predominance, lower risk of transformation to AML, and a striking response to lenalidomide.
The clinical and molecular physiology of 5q− syndrome is complex, and multiple genes lost in the CDRs are responsible for different aspects of the 5q− phenotype. First, haploinsufficiency of RPS14, a ribosomal subunit gene located at 5q31.2, is responsible for the dyserythropoiesis seen in the syndrome, due to p53 expression that is mediated by induction of innate immune signaling. Second, deletion of a key microRNA, miR-145, is responsible for the megakaryocytic component of the phenotype. Separately, haploinsufficiency of CSNK1A1, which is located at 5q32 and encodes casein kinase 1-alpha, is responsible for the sensitivity to lenalidomide, which accelerates the ubiquitination and degradation of remaining casein kinase 1-alpha through a cereblon-dependent process. Importantly, the coordinate loss of RPS14, miR-145, and CSNK1A1 has a greater impact on the del(5q) impact than single loss of any one gene. TP53 mutations and 17p loss frequently develop in the setting of del(5q), often acting as a resistance mechanism following lenalidomide treatment.
You're Reading a Preview
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here