Identify latent chromosomal aberrations relevant to myelodysplastic syndromes

Myelodysplastic syndromes (MDS) are a group of heterogeneous hematologic malignancies. This study aims to identify latent chromosomal abnormalities relevant to MDS, which may optimize the current diagnosis of MDS. Affymetrix CytoScan 750 K microarray platform was utilized to perform a genome-wide detection of chromosomal aberrations in the bone marrow cells of the patients. The findings were compared with the results from traditional karyotypic analysis and FISH to reveal latent chromosomal aberrations. Chromosomal gain, loss, and UPD, and complex karyotypes were identified in those samples. In addition to established cytogenetic aberrations detected by karyotypic analysis, CytoScan 750 K microarray also detected cryptic chromosomal lesions in MDS. Those latent defects underlying multiple gene mutations may construe the clinical variability of MDS. In Conclusion, Affymetrix CytoScan 750 K microarray is efficient in identifying latent chromosomal aberrations in MDS.


Discussion
Chromosomal abnormalities are frequent in MDS and have many clinical implications. However, a great amount of patients don't indicate cytogenetic abnormalities. MDS patients with the same chromosome lesions may have heterogeneous phenotypes, implying that some hidden genomic changes may exist among the patients 15    Gain and loss of gene copies may result in gene over-expression, absence of any functional transcript, or modest changes in gene expression 16 . We have unveiled chromosomal aberrations in regions defined as CNV of contiguous clones undetectable by classical karyotypic analysis. For instance, we identified cryptic gain (3q27.1-qter) and loss (6q23.2-qter) in case 1#. In this region, Loss(6q23.2-q23.3) has been reported to involve in sAML 17 . Furthermore, cryptic chromosomal aberrations detected by microarray have been useful for a comprehensive analysis of gene mutations in MDS. The copy number status of TET2 (on 4q24), IRF1 (5q31.1), NPM1 (5q35.1), LAMB4 (7q31.1), EZH2 (7q36.1), ETV6 (12p13.2), TP53 (17p13.1), NF1 (17q11.2), ASXL1 (20q11.21), RUNX1 (21q22.12), and STAG2 (Xq25) have been investigated in a large cohort of MDS patients 18 . In our cohort, we have identified a 1.48 Mb loss at 4q24 in case 10#. TET2 gene in this region is a tumor-suppressor gene 19 . The deletion or mutation of TET2 often predict inferior prognosis in patients with chronic myelomonocytic leukemia 20 . Although aberration in chromosome 5 has been considered as the most prevalent chromosomal lesion in MDS, some cryptic defects which affect additional key genes have not been clarified comprehensively. The major commonly deleted region (CDR) has been delineated at band 5q31.1. We have identified a 97.71 Mb loss(5q14.3qter) in case 24#. Deletion of 5q may engender haploinsufficiency of many critical genes, including ribosomal protein S14 (RPS14), casein kinase 1 α1 (CSNK1A1), adenomatous polyposis coli (APC), heat shock protein family A (HSP70) member 9 (HSPA9), early growth response 1 (EGR1), DEAD-box helicase 41 (DDX41), NPM1, TRAF-interacting protein with forkhead-associated domain B (TIFAB), Diaphanous-related formin 1 (DIAPH1), microRNA (miR)-145 and miR-146a 21 . Haploinsufficiency for the ribosomal gene RPS14 has been reported to impede erythroid differentiation in the 5q-syndrome 22 . Heterozygous deletion of CSNK1A1 may upregulate WNT signaling and stimulate stem cell expansion 23,24 . Loss(5q14.3qter) in case 24# has been validated by FISH. We have also detected loss(7q11.21) in case 10# and monosomy 7 in case 22#, 23#. Deletion of 7q and monosomy 7 are also prevalent in MDS and often portend unfavorable outcome 25 . These chromosomal alterations can also cause haploinsufficiency of some key genes implicated in MDS. These genes comprise EZH2, CUX1 and MLL3 [26][27][28] . MLL3 haploinsufficiency cooperates with RAS mutation and Trp53 to exacerbate leukemia 26 . In addition, we have found gain(20q13.2qter) in case 3#. MacKinnon et al. have investigated AML and MDS patients with 20q amplification. They identified a 250 kb common region which subsumed HCK, TM9SF4, PLAGL2, and POFUT1 gene. These patients often had a higher proportion of erythroblasts. The amplification of 20q portends the existence of oncogene 29 . Generally, these latent chromosomal aberrations contain a lot of key genes which are tightly associated with the pathogenesis of MDS. CNV identified by Affymetrix Cytoscan 750 K Microarray can contribute to the differential diagnosis of subtypes in MDS.
Recent investigations have indicated that UPD can be responsible for homozygosity of mutations of critical genes within chromosomal regions 30 . Reduction to homozygosity as a result of UPD was preliminarily considered to be a mechanism for the inactivation of tumor suppressor genes 30 . We have demonstrated UPD in 10 cases with normal karyotypes. For example, we detected UPD(13q11-qter) in case 12#. FLT3 gene in 13q12 encodes class III receptor tyrosine kinase that regulates hematopoiesis 31 . FLT3-ITD internal tandem duplications have been observed during disease progression and confers an unfavorable prognosis 32 . We also identified UPD(11p11.2-pter) in case 13#. WT1 gene mutation has been reported in UPD 11p which is related to the pathogenesis of AML 33 . Furthermore, we have also found UPD (11q13.1-qter) in case 15#. The c-CBL gene is located in 11q23.3 7 . Clonal selection of UPD 11q and CBL gene mutation often reflected the progression of MDS to AML 34 . It is noteworthy that in RCMD case 16#, the UPD of region 17q22-qter harbored the ETV4 gene, which encodes an ETS transcription factor indispensable for hematopoiesis 35 . Additionally, we have found UPD(9p21.1-pter) in case 17#. UPD 9p is tightly associated with a homozygous activating JAK2(V617F) gene mutation, implying serious prognosis 7 . Consequently, nonrandom segmental UPDs identified in this cohort may contribute to the investigation of the pathogenesis of MDS underlying large deletions.
MDS with complex chromosomal aberrations often herald short survival and an increased risk of evolution to AML 36 . Complex karyotypes with multiple chromosomal changes are found in about 20% of newly diagnosed MDS patients and are relevant to a poor prognosis 3 . We exhibited 5 cases with complex chromosomal lesions, which have been listed in our supplementary file. A combination of traditional karyotypic analysis with Affymetrix Cytoscan 750 K Microarray may well provide a more comprehensive detection of complex chromosomal aberrations in MDS.
Given the recent discovery of many recurrent gene mutations in MDS, it's still urgent to validate prior mutational correlative data. The temporal order of mutation acquisition has reflected the importance of subclonal genetic events in MDS. For instance, mutations impacting RNA splicing and DNA methylation occur early in disease progression, while kinase activating mutations (such as KIT and NRAS) occur even later in disease progression 37 . Early detection of subclonal mutations may reflect significant prognostic variables in MDS 38 . In our cohort, we have found many discrepancies between MC and Cytoscan 750 K Microarray. Microarray analysis has a variable ability to detect mosaicism that FISH and karyotyping may not accurately detect the level of mosaicism 13 . The most likely reason for these discrepancies is probably that some aberrations are subclonal.
On the other hand, one criticism of microarray for detecting chromosomal aberrations in MDS is the possibility of "false positive" results or findings of unclear clinical significance. A proportion of alterations identified in the patients may reflect normal age-related chromosomal changes. For instance, we have found loss(Yq11.222-qter) in case 2#. Loss of the Y-chromosome (LOY) is described as both a normal age-related event and a marker of a neoplastic clone in hematologic diseases 39 . Paired normal DNA from the same MDS patient may reduce the number of false positives generated by microarray 40 .
Additionally, some patients with clonal cytopenia of undetermined significance (CCUS) that do not meet the criteria for MDS may also benefit from SNP-A. A recent study has applied combined comparative genomic hybridization and SNP-A to detect cryptic chromosomal lesions in both MDS and cytopenias of undetermined significance. Based on the combined array findings, 42% of patients with indeterminate morphologic findings were categorized as CCUS. Cryptic array findings among those patients comprised large-scale UPD (up to 118 Mb) and genomic deletion of loci implicated in MDS pathogenesis (eg, TET2 (4q22) and NUP98 (11p15)). The latent chromosomal lesions revealed by SNP-A helped to indicate clonal hematopoiesis and prompted classification as CCUS 41 . Hence microarray analysis significantly improves the detection rate of clinically significant findings.
In conclusion, Affymetrix Cytoscan 750 K Microarray have identified many cryptic chromosomal abnormalities relevant to MDS, which may interpret the clinical variability and enhance our understanding of the pathogenesis of MDS.

Material and Methods
Patients and Specimen. The cohort of this study comprises patients whose bone marrow aspirates were recruited in Kingstar Global company for pathologic diagnosis of MDS from December 2014 to July 2015. All specimens were acquired with patients' approval, under the protocols permitted by Institutional Ethics Committee of Wuhan university, in comply with Helsinki Declaration. Informed consent was signed for each patient. And any publication of identifying information was also approved by the participants.
Cytogenetic analysis. Traditional G-banding Karyotypic analysis was initiated on bone marrow aspirates by trypsin and Giemsa dye. Short-term cell cultures were carried out in medium supplemented with GM-CSF or conditioned medium III. Then the cells were harvested and metaphase preparations were performed according to standard procedures. Karyotypes were depicted in the light of International System for Human Cytogenetic Nomenclature 2016 42 . At least 20 metaphases per sample should be analyzed whenever possible.
FISH Fluorescence in situ hybridization (FISH) was performed according to the manufacturer's protocols, in order to validate chromosomal aberrations detected by Affymetrix Cytoscan 750 K Microarray. A total of 400 interphase nuclei were evaluated by two independent pathologists under fluorescent microscope. The locus-specific probes were displayed in Table 4. DNA preparation. DNA was extracted from bone marrow of individual patients using the QIAamp DNA Blood Mini Kit according to the manufacturer's instructions. The concentration and quality of DNA samples were evaluated by Nanodrop 2000 spectrophotometer (Thermo Scientific). DNA integrity was assessed by 1% agarose gel electrophoresis. The quality controls (QC) of Affymetrix CytoScan 750 K microarray required that DNA concentration should be no less than 50 ng/µL, OD260/280 is about 1.9, OD260/230 is about 2.0.
Cytoscan 750 K Microarray Assay. Affymetrix Cytoscan 750 K Microarray provides a genome-wide coverage with focus on cytogenetic relevant regions, including 550,000 markers for detecting copy number variation and 200,000 high performing SNP probes with genotype accuracy >99%. All probes are empirically selected for exceptional performance. The Affymetrix ® CytoScan ™ Assay protocol is optimized for processing 8 to 24 samples at a time to obtain whole genome copy number and SNP information. The Workflow of CytoScanTM Assay can be briefed as follows (Fig. 5): (1) Digestion of gDNA with Nsp I restriction endonuclease.
(3) Polymerase chain reaction (PCR) to amplify Ligated Samples and PCR Product Check.  All cases have followed the protocols and QC guidelines provided by the manufacturer.
Data analysis. The data of Cytoscan 750 K Microarray were analyzed using Chromosome Analysis Suite Version 2.0 (Affymetrix). The QC thresholds were: SNPQC ≥ 15.0; MAPD ≤ 0.25; Waviness SD ≤ 0.12. These QC metrics can evaluate the overall quality of SNP array data. Median Absolute Pairwise Difference (MAPD) represents the typical distance between marker pairs with respect to log2 ratios. SNPQC measures the degree of separation between genotype clusters aggregated across multiple markers. Waviness-SD gauges the differences between probe sets. The microarray data were interpreted according to the annotations of genome version GRCh37 (hg19). Only the samples which complied with QC criteria and identified CNV with over 100 Kb and at least 10 aberrant probes were chosen for further analysis. Identified CNVs were contrasted with the Database of Genomic Variants (http://dgv.tcag.ca/dgv/app/home) to exclude the polymorphic variations in healthy population. As for UPD, we used an algorithm that regards both location and size of >5 Mb aberrations in order to preclude nonclonal regions. To reckon the size of the affected genome in each patient, we recognized the total size of alterations in chromosomes, including CNV and UPD.