Introduction

The progressive accumulation of genetic aberrations such as copy number abnormalities, in the form of gains or losses of genetic material affecting certain regions of the genome, and particular gene mutations, provide the basis for cancer development [1,2]. However, recent studies have revealed the presence of an alternative mechanism, termed chromothripsis, in which massive chromosome rearrangements occur in a one-step catastrophic event, indicating that chromosome instability is a central aspect of cancer cell biology. A key feature of chromothripsis is the occurrence of tens to hundreds of clustered genomic rearrangements usually in one or, in some instances, several chromosomes. This complex abnormality can affect an entire chromosome, a chromosome arm, or be confined to a single region of a chromosome [2–9]. These rearrangements usually appear crisscrossing the involved regions [4], and chromosomes affected by chromothripsis show a characteristic pattern of copy number oscillations between two (or occasionally three) copy number states [4,8]. By far the simplest explanation of such rearrangements is that, at some stage during cancer development, distinct chromosomes or chromosomal regions are broken into many segments and then inaccurately stitched back together by DNA repair mechanisms [2,7,10]. Chromothripsis was initially observed in chronic lymphocytic leukemia (CLL), but it is present in a wide range of human cancers, including multiple myeloma (MM), acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), Hodgkin lymphoma, bone cancers, medulloblastoma, neuroblastoma, colorectal cancer and melanoma. Some congenital disorders also show chromothripsis [3–17].

Myelodysplastic syndromes (MDS) are a clinically heterogeneous group of clonal hematopoietic stem cell disorders characterized by morphological dysplasia, ineffective hematopoiesis and peripheral blood cytopenias [18]. MDS and chronic myelomonocytic leukemia (CMML), an entity sharing features of myelodysplastic syndromes and chronic myeloproliferative neoplasms (MPN) [19], have a highly variable clinical course [20]. The presence of chromosomal abnormalities is a recurrent hallmark of both MDS and CMML patients, with consequences for their diagnosis, risk stratification and prognosis [18,21,22]. In fact, these genetic changes are directly incorporated into the Revised International Prognostic Scoring System (IPSS-R) for MDS and the CMML-Specific Prognostic Scoring System (CPSS) [21,22]. In addition, gene mutations are also very frequent (80–90%) in MDS and related myeloid neoplasms [23,24]. These mutations affect transcription factors such as RUNX1 and BCOR, epigenetic modulators such as TET2, DNMT3A, IDH1/2, ASXL1 and EZH2, tumor suppressor genes such as TP53, several components of the RNA-splicing machinery such as SF3B1, SRSF2, U2AF1, and ZRSR2, genes involved in DNA replication such as SETBP1, and genes of the cohesin complex such as STAG2, RAD21, SMC1A, and SMC3. The list of genes carrying mutations involved in the pathogenesis of MDS is still growing [18,23–25].

Conventional metaphase cytogenetics (CC) is still the gold standard for karyotypic studies; however, diagnosis and prognostication may be difficult in the 10–15% of patients with non-informative cytogenetics, due to the absence of mitosis, or the 40–60% with a normal karyotype [26,27]. Additionally, the presence of complex karyotypes with three or more chromosomal abnormalities may hinder the identification of the chromosomes involved in these changes [28]. Therefore, these CC techniques are not sufficient for a thorough study of these myeloid malignancies. The use of molecular genome-wide scanning techniques allows the identification of cryptic abnormalities in patients with a normal karyotype and the better characterization of unbalanced genetic changes [23,24,26,27,29,30]. Thus, array-based karyotyping revealed MDS and related myeloid neoplasms with a normal karyotype to have one or more genomic abnormalities, including deletions of TET2 and RUNX1 genes (40%) [26,27,29]. These studies suggest the potential clinical utility of such genome-wide scanning techniques for the management of MDS and related diseases.

To gain insight into the characterization of the molecular changes in MDS and to explore novel genetic abnormalities occurring in these disorders, an integrative study combining array-based comparative genomic hybridization (aCGH) and next-generation sequencing (NGS) in a series of MDS and MDS/MPN patients was carried out. The results demonstrated the presence of infrequent but recurrent chromothripsis involving chromosome 13 in these diseases and also showed that aCGH could be used in a clinical setting as a complementary method to conventional cytogenetics for identifying copy number changes in MDS and CMML patients.

Materials and Methods

Patients

A total of 301 patients diagnosed with MDS (n = 240) or MDS/MPN (n = 61) at the time of diagnosis were studied. The main clinical characteristics of the patients are summarized in Table 1. The median age was 77 years (range, 11–93 years), and 191 patients (63.5%) were male. Diagnoses were established according to the 2008 World Health Organization criteria [31] (Table 1). This study was performed in accordance with the Declaration of Helsinki guidelines, and was approved by the Local Ethical Committees “Comité Ético de Investigación Clínica, Hospital Universitario de Salamanca”. All patients provided written informed consent.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Main characteristics of the whole series of patients included in the study.

https://doi.org/10.1371/journal.pone.0164370.t001

Results

Chromothripsis on chromosome 13 is a recurrent abnormality in high-risk MDS

The analysis of copy number profiles derived from aCGH data identified complex genomic rearrangements on the aCGH chromosome plots, showing multiple non-contiguous CNAs, with the hallmarks of chromothripsis. Based on previous studies, evidence of chromothripsis was defined as the presence of at least ten changes in segmental copy number between two or three copy number states on an individual chromosome [6]. Using these criteria, three MDS cases (1.2%) exhibited chromothripsis (Fig 1A and 1B), with more than 11 copy number changes involving exclusively chromosome 13. The copy number states rapidly alternate between one (deletion), two (normal) and three (gain) copies. The patterns of genomic alteration were different between the three high-risk MDS patients. However, it should be pointed out that involvement of a total of 91 genes mapping on chromosome 13 were common to the three patients. As examples XPO4, FLT3 and FLT1 were commonly amplified genes; BRCA2 and RB1 were commonly deleted genes (S6 Table). All these results were validated using an independent microarray (SurePrint G3 Human CGH Microarray, 8x60k, Agilent Technologies). The three patients with chromothripsis were diagnosed as high-risk MDS (RAEB 3/40; 7.5%), two of them were RAEB-1, with 6% and 8% of BM blasts, respectively, and the remaining patient was RAEB-2, with 12% of BM blasts. All three patients died within one year. All of them had a complex karyotype revealed by aCGH (3/17; 17.6%; patients #026, #027, #072; Fig 1A), with a median of 21 CNAs (range, 19–33) throughout the whole genome. The three patients showed genomic losses on 5q23.2-q35.3, two of them also carried losses on 7q22.3-q36.3 and 15q11.1–21.2 (S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Recurrent chromothripsis on chromosome 13 in high-risk MDS.

(A) Whole genome view ratio plots derived from aCGH data of MDS patients (#026, #027 and #072) showing chromothripsis on chromosome 13, indicated by the red-shaded box. The three patients had complex karyotypes: patient #026 had 33 aberrations and affecting seven chromosomes; patient #027 had 19 abnormalities affecting five chromosomes; patient #072 had 21 aberrations affecting six chromosomes. The Y-axis represents the log₂ ratio values of MDS:control signal intensities for each probe. The X-axis illustrates all the probes in the array sorted by chromosome and physical mapping position. Chromosome numbers are indicated below the X-axis. (B) Detailed view of the whole chromosome 13 in patients #026, #027 and #072 showing a complex pattern of alternating copy number gains and losses. The grey area represents the thresholds of signal values (log₂ ratio) to call CNAs, red lines indicate segmented copy number profiles, and boxes shaded in pale-red and pale-blue depict copy number losses and gains, respectively. Patient #026 had 18 alternating copy number changes, patient #027 had 11, and patient #072 had 16 changes along the whole chromosome 13. Copy number profiles differed between these patients. The Y-axis represents log₂ ratios and the X-axis shows all probes of chromosome 13 sorted by chromosome position. Genomic location (Mb) is indicated below the X-axis. (C) Distribution of TP53 mutations identified by amplicon-based deep sequencing in the three MDS patients with chromothripsis. All TP53 mutations were located in the sequence-specific DNA binding domain. One patient had two mutations in heterozygosis, while the other two patients had one mutation each in homozygosis. The variant allele frequencies (VAFs) are represented in brackets. Each circle represents a mutation found in one patient. Green and red circles depict missense and nonsense mutations, respectively. Each patient is illustrated by a different-colour triangle. The complete coding region of TP53 is illustrated and the respective exons and amino acid (AA) positions are indicated at the bottom. The following protein domains are shown: TAD1 and TAD2, amino-terminal transactivation domains 1 and 2; DBD, sequence-specific DNA-binding domain; NLS, nuclear localization signalling domain; 4D, carboxy-terminal tetramerization domain; Neg, negative regulation domain. Gene variants were represented using the R package “R453PlusToolbox” [53].

https://doi.org/10.1371/journal.pone.0164370.g001

All cases showing chromothripsis carried TP53 mutations as revealed by NGS (Fig 1C). Specifically, two missense mutations (p.Lys132Arg, p.Pro278Leu) and two nonsense mutations (p.Gln136*, p.Gln167*) were identified. The p.Gln167* and p.Lys132Arg mutations, located in exon 5, were observed in one patient each with a variant allele frequency (VAF) of 93.5% and 92.5%, respectively. However, the p.Gln136* and p.Pro278Leu mutations were observed in the same patient with VAFs of 46.5% and 43.5%, and were located on exon 5 and exon 8, respectively. Taken together, most of these mutations affected exon 5 of TP53, and all of them were located in the sequence-specific DNA binding domain, which plays a central role in transcriptional transactivation (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of mutations found in the analyzed genes.

https://doi.org/10.1371/journal.pone.0164370.t002

aCGH and NGS allow the identification of hidden recurrent genetic CNAs and gene mutations in MDS

A total of 285 abnormalities were identified (1–33 changes per patient) in 71 of the entire series of 301 patients (23.6%): 61 of 244 (25%) were MDS patients, while 10 of 58 (17.3%) had a diagnosis of CMML. Copy number losses (72.6%) were more frequent than gains (27.4%). The detected CNAs were distributed amongst all chromosomes except for chromosomes 14 and 16.

Among the global series, CNAs were present in 9.3% of the normal karyotype patients, in 86.7% of cases with abnormal cytogenetics, and in 30% of patients with unsuccessful cytogenetic analyses (Fig 2A). The most frequent large recurrent aberrations were del(5q) (35.2%), del(20q) (18.3%), -Y (14.1%), trisomy 8 (14.1%), del(7q) (12.7%), +1/+1q (7%), -18/del(18p) (5.6%), del(17p) (5.6%), del(11q) (5.6%), del(4q) (4.2%), del(15q) (4.2%), and del(12q) (4.2%) (Fig 2B). The most frequent aberrations in CMML were–Y (5.2%) and gains on 1q (3.4%), while in RCMD losses on 5q (6.1%), 20q (5.4%), and–Y (3.4%) predominated. By contrast, in RAEB patients, losses involving 5q (21.7%), 7q (13%) and 17p (6.5%), and trisomy 8 (10.9%) were frequently observed (Fig 2B). In addition, in 38 of 301 cases (12.6%), aCGH revealed the presence of 81 small aberrations (≤5 Mb), which were below the detection limit of CC. In 21 of these cases only one cryptic CNA was detected: 17 deletions and four gains. In the other 17 cases, two or more cryptic aberrations were observed, consisting of 49 deletions and 11 gains. Notably, these cryptic CNAs involved regions such as 2p23.3, 4q24, 5q33.1, 7q22.1, 21q22.12, 21q22.3 and Xp14, where genes implicated in the pathogenesis of MDS and MDS/MPN are located, including DNMT3A, TET2, SPARC, CUX1, RUNX1, U2AF1 and BCOR, respectively (Fig 2B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Summary of recurrent CNAs found in the global series.

(A) Proportion of the whole series of patients with normal and abnormal aCGH profiles. Each aCGH category is then divided by the cytogenetic subgroups detected by CC studies: normal, abnormal and non-informative karyotype. Percentages represent the proportion of patients from the total number of patients within each cytogenetic subgroup. (B) Frequency of large recurrent genomic abnormalities and frequency of cryptic recurrent CNAs involving genes of known significance in MDS and MDS/MPN patients only seen by aCGH. All abnormalities are classified by MDS and MDS/MPN subtypes and color-coded as indicated on the right panel of the figure.

https://doi.org/10.1371/journal.pone.0164370.g002

Furthermore, an in-depth analysis by NGS of the regions with genomic deletions of >100 kb by aCGH, where recurrently mutated genes in MDS and CMML are located, was carried out. Thus, aCGH analysis identified two cases with a deletion in 2p23.3 (DNMT3A) (Fig 3A), seven cases with a deletion in 4q24 (TET2) (Fig 3E), six cases with a 17p13 deletion (TP53) (Fig 3G), five cases with a deletion in 21q22 (RUNX1) (Fig 3C) and two cases with an Xp11.4 deletion (BCOR) (Fig 3D). NGS studies detected that one patient with a DNMT3A deletion carried a missense mutation (p.Gly654Asp) in the other allele with a VAF of 73.5%. This mutation was located in the methyltransferase domain (Fig 3B). Three patients harboring a TET2 deletion carried one nonsense mutation (p.Arg550*), one missense mutation (p.Ser1284Phe) and one frameshift deletion (p.Leu1855Trpfs*32) each in the non-deleted allele. The variant allele frequencies were 96.5%, 3.5% and 89%, respectively. The latter two mutations affected the two evolutionarily conserved domains in the TET family proteins (Fig 3F). Of the six patients with a 17p13 deletion affecting the TP53 locus, four carried one missense mutation each (p.Cys275Tyr, p.Ile195Phe, p.Val274Asp, p.Tyr220Ser) with VAFs of 52%, 76.5%, 45% and 54%, respectively. These TP53 mutations were located in the sequence-specific DNA binding domain. None of the other studied genes showed mutations in those patients with losses in the regions of interest (Fig 3H). All mutations are summarized in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Combination of aCGH and NGS analysis for regions with frequently mutated genes in MDS and MDS/MPN.

(A, C, D, E, G) Detailed view of the whole chromosomes 2, 4, 17, 21 and X, where recurrent regions of deletion where found by aCGH, indicated by the red-shaded box. A magnified view of the minimal deleted regions with a schematic diagram showing the genes included within the deletion. For all figures, genomic locations are indicated in Mb, and the chromosome position (bp) and size (kb) of the minimal deleted regions are indicated in the upper part of each chromosome view ratio plots. The Y-axis represents the log₂ ratio values and all probes for each chromosome are sorted by genomic position along the X-axis. (A) A 533.4-kb deletion on 2p23.3 affecting the DNMT3A locus. (C) A 298.7-kb deletion on 21q22.12 affecting the RUNX1 locus. (D) A 697.1-kb deletion on Xp11.4 affecting the BCOR locus. (E) A 381.2-kb deletion on 4q24 affecting the TET2 locus. (G) An 11.14-Mb deletion on 17p13.3-p12 affecting the TP53 locus. Genes were represented using the R package “GenomeGraphs”. (B, F, H) Distribution of DNMT3A, TET2 and TP53 mutations identified by targeted amplicon-based deep sequencing. The variant allele frequencies (VAFs) are represented in brackets. The complete coding regions of DNMT3A, TET2 and TP53 are illustrated and the respective exons and amino acid (AA) positions are indicated below. Each circle represents a mutation found in a single patient. Green, red and blue circles depict missense, nonsense and frameshift mutations, respectively. (B) One patient carried a DNMT3A missense mutation located in the MTase domain. The following protein domains are shown: PWWP, proline-tryptophan-tryptophan-proline domain; ZNF, zinc finger domain; MTase, methyltransferase domain. (F) Three patients with a TET2 deletion harbored one nonsense, one missense and one frameshift mutation each. The two evolutionarily conserved domains, boxes 1 and 2, are shown. (H) Four patients carried one TP53 missense mutation each. The following protein domains are shown: TAD1 and TAD2, amino-terminal transactivation domains 1 and 2; DBD, sequence-specific DNA-binding domain; NLS, nuclear localization signalling domain; 4D, carboxy-terminal tetrame-rization domain; Neg, negative regulation domain. Gene variants were represented using the R package “R453PlusToolbox” [53].

https://doi.org/10.1371/journal.pone.0164370.g003

Discussion

The presence of specific chromosomal abnormalities and genetic changes is a hallmark of MDS [18,23–25,35,36]. In this study, we analyzed a large cohort of MDS patients by integrating two genetic methodologies: array-based comparative genomic hybridization (aCGH) and amplicon-based deep sequencing (NGS). Our results demonstrated the presence of chromothripsis, inferred from aCGH profiles, as an infrequent but recurrent genomic abnormality in high-risk MDS. In addition, our results showed that the combination of these conventional and genome-wide scanning approaches enables a better characterization of MDS and related neoplasms, and provides new information that could improve the current diagnostic and treatment of these patients.

Chromothripsis is a genetic abnormality in which tens to hundreds of clustered genomic rearrangements occur in a one-step catastrophic event. [2–9]. In the present study, chromothripsis was observed in three cases of high-risk MDS (two cases of RAEB-1 and one patient with RAEB-2), two entities known to progress gradually to a more aggressive disease. The occurrence of chromothripsis in myeloid malignancies has been demonstrated in AMLs by SNP array-based copy number profiling, in which 8% of AML patients carried massive and complex rearrangements consistent with chromothripsis [6,37]. The molecular basis of this genomic chaos has been long studied in the last years. Therefore several mechanisms such as ionizing radiation, premature chromosome compaction, DNA replication stress, telomere shortening, abortive apoptosis, TP53 mutations, micronuclei collapse, and hyperploidy formation have been suggested as implicated in this chromothripsis. [2,6–8,10,38–42]. In addition, approximately 50% of AML patients carrying TP53 mutations and approximately 40% of complex karyotype AML patients displayed chromothripsis, while only 1% of AML with wild-type TP53 and no cases with non-complex karyotype showed this aberration [6,37]. Moreover, it was also demonstrated that almost all medulloblastomas showing evidence of chromothripsis had TP53 mutations [6]. These findings reinforce the link between somatically acquired TP53 mutations and the presence of complex karyotypes with chromothripsis. It is of particular interest that the three MDS cases with chromothripsis reported in this study had complex karyotypes as revealed by aCGH and carried TP53 mutations, a previously described association [6]. Additionally, the outcome of the three cases was poor, a feature that is in accordance with high-risk MDS patients, the presence of a complex karyotype or TP53 mutations [18,21]. This complex genomic abnormality has not previously been comprehensively described in MDS patients, perhaps because the previous high-resolution copy number studies in MDS mainly concerned cases with a normal karyotype [23,26,27,29]. Other copy number studies in MDS with abnormal and complex karyotypes have been reported. These studies aimed to analyze the relationship of copy number to the CC data and their prognostic impact, and they did not show the presence of chromothripsis [28,30,43–45]. It should be noted that, in our study, chromothripsis was seen exclusively to affect the entire chromosome 13 in all three MDS patients. The presence of this genomic chaos restricted to a single chromosome has been described. Chromothripsis is typically found to affect different chromosomes at random, but was reported to involve chromosome 16p in 3/7 MM patients, and chromosome 21 in 5/9 iAMP21 ALL patients [5,17]. However, the presence of chromothripsis involving chromosome 13 has not been previously reported in myeloid malignancies [6,12,37,46]. In this study, chromothripsis affected only chromosome 13, with patterns of genomic alteration differing between the three high-risk MDS patients. FLT3 was commonly amplified, while BRCA2 and RB1 were commonly deleted in the cases with chromothripsis. FLT3 is an oncogene that regulates hematopoietic stem cell differentiation, proliferation and survival. FLT3-activating mutations were recurrently described in myeloid malignancies, mainly in AML, and are associated with poor prognosis [47]. RB1 and BRCA2 abnormalities play a role in the development of several cancers and are considered to be tumor suppressor genes [48,49]. In addition, two non-chromothripsis MDS cases (#074 and #126) showed CNAs affecting chromosome 13. One of these cases, diagnosed as RAEB-1, carried a deletion involving the RB1 locus. Our results suggest the involvement of the FLT3 gene and BRCA2 and RB1 inactivation in the pathogenesis of some cases of MDS.

The present study revealed the presence of cryptic abnormalities that were not targeted by FISH and that were below the threshold of resolution by conventional cytogenetics. In fact, 12.6% of MDS patients showed cryptic changes. Some of these submicroscopic CNAs involved regions with genes of known significance in MDS pathogenesis and were deletions in 2p23.3 (DNMT3A), 4q24 (TET2), 5q33.1 (SPARC), 7q22.1 (CUX1), 21q22.12 (RUNX1) and Xp11.4 (BCOR), and gains in 21q22.3 (U2AF1), that were detected in 19 patients [18,25,26,30]. These regions were equally likely to be involved in low- and high-risk MDS or CMML. In addition, the genes included in these recurrent cryptic deletions were further investigated by NGS to identify whether mutations occurred in the other allele. The sequencing results showed that only two of the seven cases with the TET2 deletion carried TET2 mutations, while only one of the two cases with losses in 2p23.3 showed DNMT3A mutations. Our results did not indicate any correlation between the presence of deletions and mutations in these MDS patients, which is in accordance with previously reported data [23]. Therefore, our results support the idea that conventional cytogenetic and aCGH studies could be complemented by the sequencing of multi-gene panels, which have already been described for MDS and related myeloid neoplasms, instead of single genes, in routine workflows for the study of these group of patients.

The presence of cytogenetic changes is a keystone in the prognosis of MDS and CMML patients. In fact, conventional cytogenetics is essential in most prognostic systems to stratify these hematological disorders [18,21,22]. However, in some subsets of MDS and CMML, conventional cytogenetic techniques fail to provide any results due to a lack of cell growth during culturing, and consequently some aberrations may be missed, making the diagnosis and prognostic stratification very difficult. Our study showed that 30% of karyotype failures carried genomic abnormalities revealed by aCGH. Four patients had complex karyotypes, two had trisomy 8 and one showed a del(5q), while the others showed deletions in 9p, 12q, 17p, 21q22, and -Y, and gains in 1q and 15q (S1 Table). Therefore, identifying these clinically relevant lesions is significant in patients with failed CC results. The clinical utility of SNP-A as a karyotyping tool in a series of MDS patients with unsuccessful cytogenetics has been previously demonstrated [50]. This method was also useful in those MDS patients with a normal karyotype when fewer than 20 good-quality metaphases are available for analysis. We demonstrated that in 10% of cases with 11–19 successful metaphases and in 21% of those with ≤10 harbored genomic aberrations aCGH will provide additional information that could redefine the prognostic risk of these patients, as previously suggested [51]. Thus, at least 20 metaphases need to be analyzed for a karyotype to be considered normal [52]. Therefore, the use of aCGH enabled the prognostic stratification according to the IPSS-R that could change the clinical management of this group of patients. Additionally, the presence of an abnormal aCGH profile was associated with high-risk patients.

In summary, the present report describes the presence of a high incidence of genomic changes in MDS and CMML patients by the integrative analysis of several molecular genetic methodologies. In addition to well-known copy number defects, the presence of chromothripsis involving chromosome 13 was a novel recurrent change in high-risk MDS patients. Moreover, aCGH analysis revealed the presence of cryptic abnormalities in genomic regions where MDS-related genes, such as TET2, DNMT3A, RUNX1 and BCOR, are located. Thus, the integrative analysis of conventional cytogenetics, aCGH and NGS in MDS will provide a better understanding of the molecular abnormalities occurring in these patients, and could improve the clinical management of MDS. The potential diagnostic and prognostic value of these new genomic abnormalities should be studied in prospective studies.

Supporting Information

Acknowledgments

The authors would like to thank Irene Rodríguez, Sara González, Teresa Prieto, Mª Ángeles Ramos, Almudena Martín, Ana Díaz, Ana Simón, María del Pozo, Vanesa Gutiérrez and Sandra Pujante from the Centro de Investigación del Cáncer, Salamanca, Spain, for their technical assistance.