I Walk the Line: How to Tell MDS From Other Bone Marrow Failure Conditions
- First Online:
- Cite this article as:
- Gondek, L.P. & DeZern, A.E. Curr Hematol Malig Rep (2014) 9: 389. doi:10.1007/s11899-014-0224-3
- 307 Views
Myelodysplastic syndromes (MDS) are clonal hematopoietic stem cell disorders characterized by peripheral cytopenias and ineffective hematopoiesis. MDS is an example of an age-related malignancy and its increasing prevalence and incidence can be attributed to a greater life expectancy in developed countries. Although frequently encountered in hematology/oncology clinics, MDS may constitute a diagnostic challenge especially with equivocal bone marrow morphology. Certain syndromes of bone marrow failure (BMF) may mimic MDS and formulating a correct diagnosis is vital for adequate prognostication as well as therapeutic approaches. Metaphase karyotyping (MK) is a very important diagnostic tool and marker of prognosis and can be an indicator of response to certain therapies. Unfortunately, chromosomal abnormalities may only be found in approximately 50 % of patients with MDS. In this review, we discuss the diagnostic approaches to patients with pancytopenia with a particular focus on the growing number of somatic mutations through new molecular testing.
KeywordsBone marrow failureMyelodysplastic syndromeAplastic anemiaLarge granular lymphocytosisParoxysmal nocturnal hemoglobinuriaMolecular markersSNP arraysSomatic mutations
Myelodysplastic syndromes (MDS) comprise a heterogeneous group of clonal hematopoietic stem cell disorders characterized by peripheral blood cytopenias and ineffective hematopoiesis . MDS is part of a wider spectrum of bone marrow failure (BMF) conditions that clinically and pathologically may share features. Despite distinct underlying etiology of the specific BMF, similar phenotypic characteristics may hamper the correct diagnosis, especially at early stages. Familiarity with various MDS mimics, their etiology, natural history, and prognosis are essential during diagnostic work-up, prognostication, and subsequent therapeutic decision-making [1–4].
In this review, we will focus on diagnostic strategies, which include standard hematopathological techniques, with a particular emphasis on the growing field of molecular diagnostics. While early diagnosis of BMF syndromes may be challenging, appropriate application of these newly available tests may facilitate timely diagnosis resulting in early introduction of therapies.
Diagnostic Approach to Marrow Failure
Diagnostic work-up for bone marrow failure
Onset of cytopenias
Inherited vs. acquired BMF
Inherited predisposition (e.g. GATA2 mutations)
Interstitial lung disease, idiopathic liver failure
Skeletal abnormalities (limbs)
Skin and nails abnormalities
Complete blood count with differential
BMF vs. RBC destruction
Vitamin levels, copper
FLAER assay for PNH
Chromosomal breakage test
Aspirate and biopsy
Iron stain/ringed sideroblasts
Sideroblastic anemia, nutritional deficiencies, RARS
Phenotypic abnormalities, CD34 count
Clonal abnormalities, prognosis
Hematopoietic cells defect vs. extrinsic etiology
Diagnostic Evaluation of MDS
MDS frequently presents as unilineage or multilineage cytopenias due to a hematopoietic stem cell (HSC) defect, leading to ineffective hematopoiesis and the propensity to malignant transformation in approximately 30 % of patients. A peripheral blood morphology is often helpful in assessing dysplasia in myeloid cells. Hypogranulation, hypo- or hypersegmentation, and the presence of pseudo Pelger-Huet cells are frequently observed in MDS and usually not appreciated in other disorders such as AA. A bone marrow biopsy with aspirate to assess marrow cellularity, dysplasia, number of blasts, and underlying fibrosis is a requisite. Additional flow cytometry and MK are also vital [6•]. In approximately 90 % of patients with MDS, the bone marrow examination is overtly hypercellular and dysplasia affecting erythroid or megakaryocytic lineages can be observed . A finding of excess blasts (≥5 %) is supportive of MDS or of an evolving acute leukemia. Bone marrow fibrosis may be an indicator of underlying MPN or of an overlap MDS/MPN. The confirmation of clonal hematopoiesis by MK is invaluable in cases with mild or very subtle dysplasia. Additionally, MK is the single most important predictor of survival. Some non-random chromosomal abnormalities are virtually diagnostic of a myeloid neoplasm [del(7/7q), del(5q), inv(3)] [7–9]. Others, like del(Y) may be seen in healthy elderly males with a significance that is not entirely clear [10, 11]. Given the fact that MDS is a very heterogeneous group of diseases, even in cases with established diagnosis, morphologic discrepancies still exist and correct classification of MDS subtypes may be problematic .
The diagnosis of MDS may be difficult in patients with a normal or non-informative karyotype, especially when frank dysplasia or characteristic morphologic markers like ring sideroblasts or increased blasts are not appreciated. As immediate therapy is not indicated at early stages of the disease, such patients can be closely monitored and repeat evaluation of the bone marrow can be performed at the time of worsening cytopenias, or at approximately 6 months from initial examination .
Increased numbers of CD34+ myeloid blasts may aid the diagnosis of MDS, especially in cases with a hypocellular marrow . Blasts can be assessed either by multicolor flow cytometry or immunohistochemistry . The latter test would allow for more precise quantification of CD34+ cells as hemodilution artifact is not a major problem in core biopsy evaluation but may result in spuriously lower CD34+ cells during aspirate analysis.
Finally, novel diagnostic tools including array-based karyotyping and massive parallel sequencing may prove useful in equivocal cases and are discussed in detail below [15, 16••, 17]. In most cases, these techniques can provide objective evidence of clonal hematopoiesis, prognostic stratification, and serve as markers of disease monitoring.
Distinguishing MDS from Other BMF Syndromes (MDS mimics)
Evaluation of Cytopenias in Younger Patients (Acquired vs. Inherited BMFs)
When evaluating the etiology of cytopenias in younger patients (usually <30 years old), one should strongly consider the possibility of an inherited condition. Often, acquired hypocellular conditions like AA may be challenging to distinguish from IBMFS [2, 3]. A detailed medical history, including prior complete blood counts and family history, can be particularly illustrative. Thorough physical examination may identify skeletal, skin, or nail anomalies indicating the presence of an inherited syndrome. In some instances, however, when phenotypic distinction is problematic, genotyping may be the only available alternative. Most IBMFS are caused by well-defined germline mutations: DNA repair (Fanconi anemia [FA]), mutations involving telomeres maintenance (dyskeratosis congenita [DKC]), ribosomopathies (Shwachman-Bodian-Diamond syndrome [SDBS], Diamond-Blackfan anemia [DBA]), mutations in thrombopoietin receptor (congenita amegakaryocytic thrombocytopenia), or other rare causes [18, 19]. The confirmation of underlying mutations is required to make a diagnosis of an IBMFS and referral to a specialized center is usually recommended.
DKC is another inherited BMF characterized by germline defect in telomere maintenance leading to premature telomere shortening . DKC, however, is not the only condition associated with short telomeres . The definition of clinically significant short telomere length varies in the literature due to methodology, but it is accepted that length below the 1st percentile is significant and can confirm the inherited form of marrow aplasia like DKC . Given the significant variation of critical telomere length in BMF literature, its use as a precise diagnostic test remains a challenge. Widely accepted methods to determine telomere length (TL) in peripheral blood leukocytes include flow cytometry-based fluorescence in situ hybridization (Flow-FISH) and polymerase chain reaction assay (PCR), with the former being the only test certified for clinical use [23–26]. Confirmation of short telomeres, although helpful, cannot definitively distinguish inherited from acquired forms of BMF . Decreased TL has been described in a subset of patients with acquired AA [27, 21, 24, 28]. The exact role and consequences of short telomeres in acquired AA, though, are not entirely clear. The prognostic value of TL and its use to guide therapeutic options is limited at present, though active investigations are ongoing . Telomeres have also been studied in MDS, with TL found to be only mildly reduced as compared to age-matched controls [29, 30]. In summary, TL is a useful test in patients in whom the diagnosis of DKC is suspected. In such cases, the TL is dramatically reduced reaching the level below 1st percentile and confirmation of presence of telomerase mutations is imperative.
Evaluation of Cytopenias in the Context of Hypocellular Marrow (AA and hMDS)
Most patients with MDS have normal or hypercellular marrow at the time of diagnosis. However, 5–20 % of patients may initially present with decreased marrow cellularity [31, 32]. The diagnosis in such patients may be challenging, as the presentation closely resembles AA. The diagnosis is even more difficult in cases without obvious bone marrow dysplasia, especially since mild erythroid dysplasia can be also seen in some cases of AA . The evidence of clonal hematopoiesis as determined by cytogenetic studies is considered diagnostic of MDS. Unfortunately, 40–60 % of MDS patients have normal or non-informative MK. Non-informative results may be particularly problematic in this group of patients given their deeply aplastic marrow and frequently limited amount of cells available for analysis. The limited amount of diagnostic material may also hamper the morphological evaluation of the bone marrow. Appropriate therapy is guided by morphological and cytogenetic evidence of clonal hematopoiesis and differs between AA and MDS patients. Although immunosuppressive therapy is effective in some patients with hMDS, an accurate diagnosis has major prognostic implications [33, 34].
Newer molecular techniques, discussed later in this review, may be an important adjunct in diagnostically challenging cases. In one study, single nucleotide polymorphism array (SNP-A) karyotyping applied along with MK in patient with AA and hMDS resulted in improved detection of chromosomal lesions allowing for better distinction between AA and hMDS. The fact that some clonal abnormalities including copy-neutral loss of heterozygosity (CN-LOH) were found in AA patients may raise the question whether the correct diagnosis was made. It is, however, possible that some degree of clonality exists in AA due to heavily depleted HSC pool, especially that some of the lesions may subside after immunosuppressive therapy . Whether the detected lesions are associated with an increased risk of malignant transformation is yet to be determined.
It is well accepted that an MDS clone originates from an immature HSC and the level of CD34+ progenitors correlates with clonal expansion and disease progression [36, 37]. Conversely, in AA, hematopoietic stem and progenitor cells are profoundly depleted due to immune attack [38, 39]. Thus, quantitative analyses of CD34+ hematopoietic progenitors may be helpful in distinguishing AA from hMDS. In one single center study, patients with a normal or increased CD34+ cells were found to have clonal cytogenetic abnormalities, suggesting the diagnosis of hMDS rather than AA. The opposite was true for patients with a low percentage of CD34+ cells who were ultimately diagnosed with AA and responded to immunosuppression .
Evaluation of Cytopenias in the Context of Normo- and Hypercellular Marrow (LGL, PRCA, and Inflammatory Diseases)
The majority of MDS cases are characterized by normo- or hypercellular bone marrow. Unlike hMDS, the differential diagnosis of a hypercellular marrow with no clonal abnormalities is relatively broad and includes PRCA, LGL, marrow suppression associated with systemic inflammation, and other immune-mediated cytopenias. Patients with clonal hematopoiesis due to, for instance, PNH may also present with cytopenias and a hypercellular marrow, but PNH is relative easy to rule-out using flow cytometric analysis (absence of membrane CD55 and CD59 in PNH cells). Distinguishing MDS from PRCA, LGL, and inflammatory diseases is more challenging .
Unlike MDS, in which a block in differentiation and ineffective hematopoiesis is largely attributed to cell-intrinsic defects, cytopenias associated with PRCA, LGL, and inflammatory diseases are usually driven by cell extrinsic immune attack. Functional in vitro colony-forming assays are helpful in establishing the correct diagnosis. As expected, colony growth in vitro is preserved in cases of immune-mediated attack (PRCA, inflammatory diseases). Additionally, robust colony formation is a predictor of response to immunosuppressive therapy [41–44]. Burst-forming units-erythroid (BFU-E) assays can also be used to exclude MDS in cases where dysplasia or markers of clonal hematopoiesis are equivocal. BFU-E growth above 40 colonies per 105 marrow mononuclear cells virtually eliminates MDS as an underlying etiology .
LGL Associated With Other BMFs
LGL can overlap with other acquired BMF such as PRCA, MDS, and AA. The pathophysiology of LGL cells in this setting is not entirely clear but may signal an immune system response to clonal cells. It is important to determine whether the LGL clone is an underlying cause of pancytopenia or rather a co-existing condition. In some instances, even though LGLs are seen on peripheral blood smear or flow cytometric analysis, they remain polyclonal and are likely reactive. Given its overlap with other BMF disorders, when LGLs are present, looking for other underlying bone marrow disorder like MDS is important. Using STAT3 mutational status to distinguish LGL is not helpful because these mutations are also detected in AA and MDS when associated with LGL, as well as in classical LGL phenotype (Felty’s syndrome) of isolated neutropenia and rheumatoid arthritis [45, 46]. STAT3 mutations have also been reported in PRCA . The presence of LGL cells concurrently with MDS correlated with poor response to immunosuppressive therapy (IST) as compared to pure LGL . Better response to IST was, however, observed in PRCA associated with an LGL clone [41, 42]. The identification of concurrent LGL cells, although important, may not warrant LGL-directed therapies.
Markers of Clonal Hematopoiesis in the Diagnosis and Prognosis of MDS
MK remains the diagnostic gold standard and is invaluable in cases with peripheral cytopenias, absence of increased bone marrow blasts, and discrete or no dysplasia. Unfortunately, the test is informative in only 40–60 % of MDS patients. In addition to its diagnostic value, MK has been the single most important predictor of survival and, in some cases, response to targeted therapies [49••]. For example, lenalidomide has been shown to be effective in patients with 5q deletion [50, 51]. Methodology of MK requires bone marrow aspirates in order to obtain adequate bone marrow material containing viable cells. Additionally, the test is non-informative in approximately 10 % of patients due to difficulties in cell culture. The detection rate of relatively small copy number changes is limited by the resolution of the microscopy and the lesions of <1 million base pairs are frequently missed. Finally, CN-LOH which is frequently seen in MDS, cannot be detected by MK.
Given these limitations, there has been a search for more sensitive and precise cytogenetic methods. Microarray-based assays have proved to be a useful diagnostic and prognostic tool. SNP-A is used concurrently with MK in a variety of hematological malignancies. The high resolution allows for detection of small, submicroscopic copy number changes in patients with unremarkable MK. Moreover, SNP-A allows for detection of CN-LOH, which is a frequent genetic aberration in cancer and remains undetectable by traditional MK. SNP-A is particularly useful in instances of inaspirable bone marrow. In such patients, diagnosis can be facilitated using SNP-A performed on DNA isolated from peripheral blood. Analogically, useful results can be obtained in patients with non-informative MK due to poor cell growth. In a study of 174 patients with MDS, MDS/MPN, and secondary leukemias, using SNP-A, clonal chromosomal aberrations were detected in 75 % of cases and CN-LOH was seen in 20–30 % of patients . SNP-A also allowed for detection of additional copy number change and CN-LOH in 119 patients with low-risk MDS . In addition to being a complementary diagnostic test, chromosomal abnormalities detected by SNP-A are an important prognostic marker [52, 17, 54].
Somatic mutations in MDS
20–30 % MDS
20–43 % MDS/MPN
Unfavorable in CMML
More favorable in AML
IDH1 associated with worse prognosis
17 % MDS
50 % MDS/MPN
Frequent in RARS and RCMD-RS
Frequent in RARS-T
Frequent in CMML
Associated with ET and MF
Frequent in aCML and CMML
Ten-eleven translocation 2 (TET2) gene is a putative tumor suppressor gene localized on chromosome 4q24. Loss-of-function mutations have been described in both myeloid and lymphoid malignancies as well as in healthy elderly individuals with clonal hematopoiesis. TET2 mutations are present in approximately 20–30 % of patients with MDS, 20–43 % of patients with MDS/MPN (including chronic myelomonocytic leukemia (CMML)), and 25–40 % of patients with secondary leukemia. Moreover, TET2 mutations are frequent in patients with normal MK, making them convenient indicators of clonal hematopoiesis in equivocal cases [16••]. Data regarding the prognostic significance of TET2 mutations remain divisive. In one study by Kosmider et al., mutated TET2 was a favorable prognostic marker with a 5-year survival of 76.9 % in TET2-mutated group vs. 18.3 % in the TET2 wild-type group . Several studies, however, showed no difference in survival based on TET2 mutation status [56, 57]. Although TET2 may seem to be a promising molecular marker facilitating diagnosis, mutations in this gene can be observed in up to 5 % of elderly individuals with clonal hematopoiesis as determined by X chromosome inactivation studies and no apparent hematological abnormalities . The exact risk of transformation in such patients is unknown, and it is unclear whether TET2 alone or additional genetic events are necessary for development of overt hematologic malignancy. Consequently, mutations in TET2 appear to be an early genetic event likely predisposing to clonal evolution and eventually malignant transformation. As such, it may persist in antecedent clone even after successful therapy, which precludes its use as a marker of response or residual disease.
Another example of a gene frequently mutated in hematologic malignancies and involved in epigenetic regulation via DNA methylation is DNA methyltransferase 3A (DNMT3A). Although more frequent in de novo acute myeloid leukemia (AML), mutations in this gene have been identified in about 3–8 % of MDS patients [59, 60]. Single institution experience suggests that mutated DNMT3A is a prognostic factor for both worse overall survival and higher rates of progression to leukemia . Recent data suggest the DNMT3A is an initial genetic event occurring in pre-leukemic stem cells and may be present in <0.1 % of otherwise healthy population. HSC clones harboring DNMT3A mutations can be also detected in post-remission samples . Whether the presence of DNMT3A mutations in complete remission samples is a predictor of disease relapse remains to be seen.
IDH1 and IDH2
Two genes encoding isocitrate dehydrogenase (IDH1 and IDH2) were found to be frequently mutated in AML and to a lesser degree in MDS and MDS/MPN with a frequency of <5 % [62–64]. IDH1/2 mutations are mutually exclusive with TET2 mutations indicating their functional redundancy and involvement in the same pathologic pathway. Also, similar to TET2, IDH1 mutations were more frequently encountered in patients with normal karyotype. Unlike in AML, where IDH mutations are favorable prognostic markers, IDH1 but not IDH2 mutations seem to be an indicator of inferior outcome in MDS [65, 66].
Genes Involved in Chromatin Modification
Loss-of-function mutations in additional sex comb-like 1 (ASXL1) gene involved in polycomb-group repressive complex 1 (PRC1) have been described in a variety of human malignancies including MDS. Approximately 17 % of MDS and up to 49 % of MDS/MPN patients harbor ASXL1 mutations [16••, 67, 68, 69•]. Frameshift and nonsense mutations are also an independent marker of adverse overall survival and shorter time to AML progression both in MDS and MDS/MPN [16••, 68, 70]. Although ASXL1 seems to be a marker or inferior outcome, it does not contribute to leukemic transformation and progression . Also, no association has been found between ASXL1 mutational status and response to lenalidomide or hypomethylating agents [71, 72]. Mutations in ASXL1 may also serve as an individualized marker of minimal residual disease (MRD) in a patient undergoing allogeneic HSCT, although the presence of mutation did not predict survival in the setting of BMT .
Enhancer of zeste homolog 2 (EZH2) located on the long arm of chromosome 7 encodes a histone methyltransferase subunit of PRC2 complex. Loss-of-function mutations have been observed in approximately 6 % or MDS patients [74, 75]. The presence of EZH2 mutations was mostly seen in lower-risk patients and was strongly associated with decreased overall survival in multivariate model [16••, 76••].
Gene-encoding proteins involved in mRNA splicing (SF3B1, ZRSR2, SRSF2, and U2AF1) were found to be the most frequently mutated group of genes in MDS [77••, 78–80]. Mutations in SF3B1 affected approximately 30 % of patients and were nearly pathognomonic for dysplasia associated with the presence of ringed sideroblasts [77••, 78, 81]. Interestingly, mutations in U2AF1, ZRSR2, and SRSF2 were seen in MDS cases without ring sideroblasts. SF3B1 mutations were found to be an independent predictor of better overall survival and lower risk of transformation to AML . Although affecting the same pathway, mutations in U2AF1 and SRSF2 were associated with worse overall survival and shorter progression to AML [79, 80]. It is now well accepted that mutations affecting mRNA splicing machinery are likely to be early/initiating events in disease evolution and are present in virtually all malignant cells. The presence of point mutations as detected by PCR may also serve as a molecular marker of MRD after chemotherapy or hematopoietic stem cell transplant [79, 80, 83].
Other Somatic Mutations
Mutations affecting transcription factors have been reported in both de novo AML and in nearly 10 % of MDS cases . RUNX1 mutations were also described as a germline variant in familial platelets disorder and myelodysplasia with a propensity to leukemic transformation .
Mutations in Casitas B cell lymphoma (c-CBL) were found by candidate screening in area of CN-LOH spanning long arm or chromosome 7 . CBL mutations were most prevalent in CMML (5 % of cases) and sAML (nearly 10 % of cases) but were relatively infrequent in MDS . Mutations in c-CBL tend to arise in a subclone of malignant cells during disease progression. As such, they are an indicator of poor outcome.
Mutations in Ras superfamily (NRAS, KRAS, NF1, PTPN11), DNA repair (TP53), cohesion complex (STAG2, SMC3), and SETBP1 have been observed in subsets of MDS patients and their frequencies and prognostic impact have been mentioned in Table 2 [88–95].
As above, a number of recurrently mutated genes have been identified in MDS. The total number of mutations present in a patient with MDS has important prognostic value, even independent of currently used clinical prognostic information [16••, 96•]. We have reviewed the most commonly mutated genes, some associated with worse outcome and decreased leukemia-free survival [76••]. Certain mutational combinations are also likely critical in MDS pathogenesis. Specifically, several groups have now examined the mutational allele frequency of each of the commonly mutated genes to estimate the temporal order of mutation acquisition in MDS and to understand the impact of subclonal mutations (defined as mutations not present in the entire proportion of malignant cells) on disease course [96•]. Studies have found that the detection of these mutations had a significant impact on outcome even when not present in the dominant clone. From a diagnostic standpoint, this suggests that molecular tests might simultaneously provide both mutational data and copy number changes capable of supplanting recurrent cytogenetic testing in MDS. However, there are always limitations and caveats to this type of testing. Mutations in isolation without appropriate clinical context must be interpreted with some caution. As an MDS community, we have yet to realize the full clinical utility of these innovative markers. It is only the patient with MDS and a clonal abnormality and a mutation in ETV6 or TP53 to whom consideration should be given for very aggressive therapy for their high-risk disease. SNP-A or genomic analyses are also useful in detecting genetic alterations that contribute to disease phenotype, but they are not in and of themselves disease-defining. Therefore, these should again be used with caution in isolation in diagnosing a bone marrow failure condition as we gain additional knowledge and experience in the field.
Establishing the correct diagnosis in patients with cytopenias can be challenging. Clinicians may encounter a diagnostic dilemma, particularly in patients with only mild cytopenias, no obvious dysplasia, and lack of cytogenetic abnormalities as determined by MK. A detailed work-up (Table 1) can direct the clinician toward the best diagnostic path and help distinguish acquired from inherited bone marrow failure. New molecular tools like SNP-A and somatic mutation analysis prove to be a helpful adjunct tool especially in equivocal cases. Early studies also suggest that certain somatic mutations and chromosomal changes detected by SNP-A may have an important prognostic value but prospective clinical studies will be required to prove their role as a risk stratification marker.
Compliance with Ethics Guidelines
Conflict of Interest
Dr. Lukasz P. Gondek and Dr. Amy E. DeZern each declare no potential conflicts of interest.
Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.