SNP Array Karyotyping Allows for the Detection of Uniparental Disomy and Cryptic Chromosomal Abnormalities in MDS/MPD-U and MPD

We applied single nucleotide polymorphism arrays (SNP-A) to study karyotypic abnormalities in patients with atypical myeloproliferative syndromes (MPD), including myeloproliferative/myelodysplastic syndrome overlap both positive and negative for the JAK2 V617F mutation and secondary acute myeloid leukemia (AML). In typical MPD cases (N = 8), which served as a control group, those with a homozygous V617F mutation showed clear uniparental disomy (UPD) of 9p using SNP-A. Consistent with possible genomic instability, in 19/30 MDS/MPD-U patients, we found additional lesions not identified by metaphase cytogenetics. In addition to UPD9p, we also have detected UPD affecting other chromosomes, including 1 (2/30), 11 (4/30), 12 (1/30) and 22 (1/30). Transformation to AML was observed in 8/30 patients. In 5 V617F+ patients who progressed to AML, we show that SNP-A can allow for the detection of two modes of transformation: leukemic blasts evolving from either a wild-type jak2 precursor carrying other acquired chromosomal defects, or from a V617F+ mutant progenitor characterized by UPD9p. SNP-A-based detection of cryptic lesions in MDS/MPD-U may help explain the clinical heterogeneity of this disorder.


INTRODUCTION
Acquired loss of heterozygosity (LOH) can occur either as a result of deletions or mitotic recombination (uniparental disomy [UPD]). LOH has been described in many malignant hematologic conditions including acute myelogenous leukemia (AML), as well as solid cancers [1][2][3][4][5]. Using 50K single nucleotide polymorphism arrays (SNP-A), we recently identified a high frequency of UPD in myelodysplastic syndromes (MDS) as well, occurring in approximately 30% of patients [6]. A similar result was subsequently found in low-risk MDS subtypes [7]. Moreover, other previously cryptic aberrations not detected by metaphase cytogenetics (MC) were identified in all sub-types of MDS, including secondary leukemias [6]. We have also extensively validated this technology with regard to its sensitivity and resolution for detecting acquired loss of heterozygosity, gains, and micro-deletions [8].
LOH can lead to decreased gene expression or hemi/ homozygosity of germ line variants or somatic mutations affecting the remaining allele. One such example is JAK2 (*147796) V617F, a gain-of-function mutation often associated with myeloproliferative syndromes (MPD), particularly polycythemia vera [9][10][11][12][13]. Studies using microsatellite markers, as opposed to other methods such as MC or fluorescence in situ hybridization (FISH), which cannot detect LOH due to UPD, have shown that UPD9p leads to homozygosity of the V617F mutation [14]. The V617F mutation has been rarely found in de novo AML [15][16][17][18], but has recently been associated with MDS/MPD overlap syndromes and in some cases, AML evolving from MPD [19,20].
The WHO classification of myeloid malignancies distinguishes MDS/MPD as a separate entity and includes chronic myelomonocytic leukemia (CMML), MDS/MPD-Unclassifiable (MDS/ MPD-U), and refractory anemia with ringed sideroblasts and thrombocytosis (RARS-t) [21]. Analysis of a large cohort of these patients has revealed that the V617F mutation is present in only small proportion of these patients [20]. However, we have shown recently that patients with RARS-t have a higher incidence of V617F [22] suggesting that V617F genotyping and evaluation would be useful in classification and clinical evaluation.
We hypothesize that if cytogenetic methods with a higher resolution are used, additional defects, including UPD of chromosomal regions other than 9p, will be detected. Furthermore, we stipulate that SNP-A can allow for a convenient method of identifying chromosomal changes in patients who transform to AML.
To investigate these hypotheses, we have applied 250K SNParrays (SNP-A) to examine genomic composition and identify previously cryptic chromosomal defects and molecular abnormalities in a group of patients with MDS/MPD-U and secondary AML developed from MDS/MPD-U both positive and negative for the V617F mutation. We hypothesize that while UPD9p leads to homozygosity of the JAK2 mutation in MPD, other occurrences of UPD on additional chromosomes in MDS/MPD-U patients may contribute to the phenotypic features that give rise to this ambiguous disease.

RESULTS
Detection of UPD9p in patients with MPD as a demonstration of SNP-A to identify copy-neutral LOH First, we performed SNP-A karyotyping in 58 control marrows. Normal copy number polymorphisms were easily detected and excluded from the patient analysis along with any others previously reported in the literature or on available internet databases [27]. LOH was detected in 4/58 healthy controls and tended to be smaller in size. Consequently, for the analysis of any patients, these changes were not deemed significant and similar to other lesions, were excluded from the analysis. Repeated analysis of samples (N = 6) showed high congruency (data not shown) and analysis of SNP calls spanning chromosome X in males revealed remarkable fidelity with an accuracy rate of .99% [8].
To determine whether or not SNP-A can be used as an accurate and effective method to identify acquired UPD, we studied patients with typical MPD homozygous for the V617F mutation. In 4 control patients homozygous for V617F (Pts. #34, #36, #37 and #38, Tables 1 & 2), UPD9p was easily detected by SNP-A; in addition, those positive control patients heterozygous for the V617F mutation showed normal diploid copy number by SNP-A (except for Pt. #31 with trisomy 9), and therefore lacked any UPD9p ( Figure 1). However, it is possible that an abnormal clone harboring UPD9p exists in heterozygous V617F patients but is too small to be detected using current methods. Regardless, similar to results recently published [8,28], we have shown that SNP-A provides a simple and effective tool for detecting acquired UPD.

Detection of cryptic chromosomal abnormalities in MDS/MPD-U patients
Using SNP-A, clonal lesions, including segmental UPD, were found in 23/30 (77%) patients as compared to 18/30 (60%) by conventional metaphase cytogenetics ( Table 2). All unbalanced copy number changes found by MC were confirmed by SNP-A, and in most instances, allowed for further refinement to more isolated regions. UPD, including UPD9p, appears to be a common chromosomal defect, occurring on chromosomes other than 9 in 8/30 patients (i.e. on chromosomes 1, 11, and 22) ( Table 3). In addition, other copy number changes were detected as well, including segmental micro-deletions of chromosomes 1, 5, 9 and 12, among others ( Table 2). Any shared/overlapping lesions found in 3 or more patients were isolated and are indicated in Table 4. Not surprisingly, the most common region of overlap was that of 9p spanning the region of the jak2 gene. Likewise, other common lesions often associated with these diseases were identified, including del5q and trisomy 8. However, three previously cryptic overlapping regions shared in 3 or more patients were identified ( Table 4). These regions consisted of three small segmental deletions of chromosome 7 (7q22.1, 7q34, and 7q36.1), a small cytoband (q14.1) of chromosome 11 (as defined by patients #9, #11 and #13), and small sub-sections of the q arm of chromosome 20 lying within the area of cytobands q11.23 to q12 (as defined by patients #4, #11, #20, and #21).
Analysis of patients with MDS/MPD-U who progressed to AML compared to those with a stable course of the disease showed, as expected, a greater number of lesions detected in the first group (8/8 vs. 15/22). Within cohorts studied, Kaplan-Meier analysis of survival shows no difference between patients with or without previously cryptic defects uncovered by SNP arrays (data not shown).
Comparison of chromosomal abnormalities in V617F+ and V617F-MDS/MPD-U We have identified patients (Tables 1 & 2) with a history of MDS/MPD-U both positive (N = 12) and negative (N = 18) for the JAK2 V617F mutation and have analyzed them using 250K SNP-A. Conventional MC revealed chromosomal aberrations in 9/12 (75%) of those positive for the mutation, including common abnormalities such as +8 and del5q. A majority of the V617F+ MDS/MPD-U patients (7/9) with abnormal MC showed previously undetected additional lesions, including UPD on chromosomes other than 9 in only 1/12 patients ( Table 2). Examples of deleted regions in patients positive for the mutation include segmental losses within chromosomes 2, 4, 5 and 20 and UPD on chromosomes 1p and 9p. However, these micro-deletions and instances of UPD appear to be highly variable from patient to patient with no substantial overlap occurring between more than two patients.
For those MDS/MPD-U patients negative for the V617F mutation, previously undetected lesions were also identified by SNP-A. Of these 18 patients, 9 showed abnormal MC. However, when SNP-A was applied, 5 additional patients with normal MC were found to have lesions, and 7 of the 9 patients with abnormal MC had lesions in addition to those detected by MC. Examples of these lesions include micro-deletions of chromosomes 6 and 20. UPD was more common in V617Fpatients, occurring in 7/18 patients, predominantly on chromosome 11 (in 4 of the 7 patients).
Detection of cryptic chromosomal abnormalities in patients with secondary AML evolving from JAK2 V617F+ disease SNP-A can also be used to identify lesions acquired during AML evolution. We investigated cytogenetic implications of AML transformation in 5 patients. In 2 patients with classical MPD (#34 and #38), SNP-A showed two distinct modes of transformation to AML. In patient #38, UPD9p was present at diagnosis, consistent with a homozygous JAK2 V617F mutation. However, upon transformation to AML, repeated SNP-A analysis now showed the presence of a new, V617Fleukemic clone with a normal chromosome 9 and previously absent micro-deletions on both chromosomes 4 and 19 ( Figure 2). In the second patient, transformation was accompanied by the presence of UPD9p (consistent with homozygous Jak2 mutation) and UPD7q. Both lesions were present in blasts as well as in mature granulocytes suggesting an evolution from within the JAK2 mutated clone ( Figure 3A, B, C).
We performed similar analysis of 3 MDS/MPD-U patients who transformed to AML (Tables 1 & 2; patients #26, #28 and #29). Similar to patient #38 above and in accordance with recent published findings [29,30], patient #26 showed two distinct clones upon transformation; one contributing to mature cell production characterized by the V617F mutation and a second with wild-type jak2 (data not shown). However, in 2 of 3 MDS/MPD-U patients (Pts. #28 and #29), we identified the presence of a clone possessing both the JAK2 V617F mutation and additional chromosomal lesion(s) associated with AML transformation. This result was obtained by comparison of SNP-A data from both granulocytes and sorted CD34+ blasts ( Figure 3D, E, F). More interestingly, patient #29 seems to harbor 2 clones: one clone with UPD9p that resulted in a homozygous JAK mutation and an AML clone which possesses 1 mutated and 1 wild type jak2 allele. Granulocyte idiograms demonstrated the presence of UPD9p (confirming the homozygous V617F clone) and an otherwise normal chromosome set. In CD34+ blasts, however, a normal chromosome 9 was present but a new del6p was detected, likely contributing to leukemic transformation. ARMS-PCR ( Figure 3F, right panel) showed a mutated jak2 (T) and wild-type allele (G) in the blast fraction suggesting that the AML clone arose from a cell heterozygous for the jak2 mutation.

DISCUSSION
Some cases of MPD show a pathophysiologic overlap with MDS [21,22,31], and in some instances, the JAK2 mutation is present. A significant clinical heterogeneity, ranging from common chronic course to less frequent evolution to AML [13,[16][17][18], exists among classical MPD patients with the JAK2 V617F mutation, and is even more diverse in patients with MDS/MPD overlap. It is possible that some of this variability in clinical phenotype may be related to the diverse spectrum of cytogenetic defects found in these patients. The presence of these cytogenetic abnormalities, as detected by SNP-A, supports the theory that MPD patients have an underlying propensity to chromosomal breaks and subclonal evolution. This feature may be particularly accentuated in patients with MDS/MPD-U overlap, although only a minority of these patients, perhaps with exception of RARS-t, harbor the V617F mutation [20,22].
We have compared the results of MC and SNP-A karyotyping in patients with MDS/MPD-U with and without the JAK2 V617F mutation. SNP-A allowed for the detection of previously cryptic chromosomal abnormalities in 19/30 patients with MDS/MPD-U, including those who showed normal MC. In addition to gains and deletions, SNP-A allowed for efficient detection of UPD. The most illustrative example of this type is UPD9p itself, which was found in all patients homozygous for the V617F mutation.   Recently, the utility of SNP-A for identification of UPD9p was demonstrated in both MPD [28] and MDS [7,8] and our results show that UPD due to mitotic recombination can affect not only the p-arm of chromosome 9 but also other regions of the genome. While there was an abundance of UPD found in these patients, no significant overlap occurred, perhaps due to the size of the cohort studied. However, one region of overlap, located on chromosome 11q14.1, harbors genes of particular interest, including rab30 (*605693), a proto-oncogenic member of the RAS family, and perhaps most notably, gab2 (*606203), which encodes an adaptor protein that mediates the interaction between STAT5 and phosphatidylinositol-3 kinase (PI3k) in activation of the PI3k/Akt pathway [32,33]. Copy-neutral loss of heterozygosity has been described in several malignancies [1][2][3][4][5] but due to the inability of MC to identify UPD, it has remained undetected in many patients. Previously, we have demonstrated that UPD is quite prevalent in patients with MDS and can be found in approximately 30% of cases studied by either 50K or 250K SNP-A [8].
In general, there was no difference in the types of lesions present in patients with and without the V617F mutation; however, UPD was present more often in patients who were negative for the mutation. As expected, UPD9p was not encountered in patients either wild-type or heterozygous for the JAK2 mutation.
In the course of our study, we have observed a number of patients who progressed from MDS/MPD-U and MPD to AML. SNP-A helped to delineate the origin of the AML clone. The pathogenesis of AML in patient #26 and #38 is consistent with previous reports which show that in a majority of cases, transformation to AML in patients with classical MPD occurs in a stem cell with a wild-type jak2 gene [29,30]. However, we have found that in some instances (pts. #28, #29, and #34), AML does originate from a stem cell with a mutant jak2 gene. The transformation process is often associated with the acquisition of additional defects (e.g. del6p as demonstrated in patient #29, or micro-deletions of chromosomes 4 and 19 as demonstrated in patient #38).
Our ability to detect cryptic lesions is related to the technical advantages of SNP-A-based karyotyping, which allows for the    identification of smaller lesions and copy number neutral changes (UPD); however, one drawback to SNP-A karyotyping is its inability to recognize balanced translocations [34]. In order to assess the significance of clonal chromosomal lesions present in patients with MDS/MPD-U, study of normal control specimens is essential. We have found a large number of known copy number polymorphisms detected in controls, but UPD was only found in 4/58 patients and tended to be much smaller than that seen in patients. For the purpose of our study, all copy number changes seen in patients were compared to those described both in public databases and those present in our own control group. Subsequently, all copy number variations found in normal controls were eliminated from our patient samples and analysis. In previous studies, we have extensively validated the results obtained by 50K and 250K arrays using FISH, MC, microsatellite PCR, and TaqMan PCR for copy number determination [6,8]. However, the sensitivity of SNP-A for the number of clonal cells in the sample is limited: as tumor content decreases, LOH detection rate steeply declines, and with ,20-30% tumor cells, no LOH can be detected, even when complete genotype information for both tumor and paired constitutive DNA is obtained [28].
In summary, our results demonstrate the applicability of SNP-A-based karyotyping for detecting clonal cytogenetic abnormalities in MDS/MPD-U. This new technology allows for precise definition of chromosomal aberrations, including copy-neutral LOH, and complements MC in detecting chromosomal lesions in MDS/MPD-U. Our results demonstrate that UPD is a common form of LOH in both JAK2 V617F+ and V617Fdisorders and is not restricted only to chromosome 9p but can affect other regions which may potentially point towards causative genes.

Patients
Bone marrow aspirates were collected from patients with MDS/ MPD-U (N = 30) and MPD (N = 8) (mean age 69 years; range 41-81) seen between 2002-2007. Patients were grouped according to the World Health Organization (WHO) classification system [35] ( Table 1). For the purpose of the study, we have focused our analysis on patients with MDS/MPD. This includes MDS/MPD-U, RARS-t, CMML and atypical CML. However, CMML and atypical CML represent a phenotypically distinct subtype of MDS/MPD and have not been included in this study. In 5 patients who transformed to AML we studied both original clones as well as leukemic blasts. Informed consent for sample collection was obtained according to protocols approved by the Cleveland Clinic IRB. Aspirates obtained from 58 healthy individuals (mean age 43 years; range 27-61) were used as controls.

Cytogenetic analysis
Cytogenetic analysis was performed on marrow aspirates according to standard methods. Chromosome preparations were Gbanded using trypsin and Giemsa (GTG), and karyotypes were described according to ISCN [36].

Cell separation
In some experiments, granulocytes were used as a source of DNA for analysis and were isolated during density centrifugation separation from pelleted cells. For separation of CD34+ cells and CD3+ lymphocytes, blood or marrow mononuclear cells were isolated using Ficoll density gradient centrifugation and then separated using magnetic beads (StemCell Technologies, Vancouver, Canda) on the RoboSep Instrument (StemCell Technologies) according to the protocol provided by the manufacturer. The purity of isolated cells was .95% as measured by flow cytometry (Coulter Elite, Hielah, FL).

DNA extraction
DNA was extracted from whole bone marrow with the ArchivePure DNA Blood Kit (5Prime, Gaithersburg, MD, USA). Red blood cell lysis solution was added to whole bone marrow at a 3:1 ratio and incubated with shaking for 10 minutes. The cells were pelleted and the DNA extracted as per the manufacturer's instructions. The concentration of the DNA was obtained using a ND-1000 spectrophotometer (NanoDrop, Wilmington, DE, USA) and the quality determined by gel electrophoresis.

SNP-A analysis
The Gene Chip Mapping 250K Assay Kit (Affymetrix, Santa Clara, CA) was used for SNP-A analysis. Following Nsp I digestion, fragmented DNA was ligated to adaptor followed by PCR amplification. The PCR product was hybridized to the GeneChip Mapping 250K Nsp Array, processed with the Fluidic Station and the Gene Chip Scanner 3000 (Affymetrix). Only DNA samples with a call rate of .90% were used. To determine the minimal clonal size that can be detected by SNP-A, dilution studies of trisomy 21 DNA with normal diploid DNA were performed; clones spanning more than 25-50% of the total cell population can be detected. Unlike comparative genomic hybridization, SNP-A does not require reference DNA as it obtains copy number and loss of heterozygosity calls through analysis of hybridization frequencies using probes designed to detect individual SNP alleles [34].

Microsatellite and gene copy number (CN) analysis
In general, SNP-A as a karyotyping platform was validated in our previous studies with regard to sensitivity, resolution, and ability to detect clonal vs. germ line chromosomal defects [8,34]. When appropriate, regions of LOH for this study were confirmed by microsatellite (MS) polymorphism analysis. Primer sequences were obtained from the NCBI database (http://www.ncbi.nlm.nih.gov). Forward primers were modified at the 59 end with FAM fluorescent dye. DNA was amplified and amplicons analyzed using ABI Prism 310 Genetic Analyzer (Applied Biosystems, Foster City, CA). CN analysis was performed by microsatellite analysis using a Real-Time TaqMan chemistry protocol [38]. The probe for detection of CA repeats was designed as a 21-bp oligomer containing GT repeats with FAM and Black Hole Quencher modifications on 59 and 39 ends, respectively. All reactions were performed in triplicate using the D12S1699 amplicon as an endogenous control.

Biostatistical evaluation
Signal intensity was analyzed and SNP calls determined using Gene Chip Genotyping Analysis Software Version 4.0 (GTYPE). CN was investigated using a Hidden Markov Model and CN Analyzer for Affymetrix GeneChip Mapping 250K arrays (CNAG 2.0) [39]. Segmental LOH was identified by a statistical assessment of the likelihood that consecutive SNP loci would exhibit heterozygosity given the corresponding allelic frequency of particular SNP in the normal population (CNAG). The two-sided Fisher's Exact test was used to analyze the difference between the distribution of dichotomized variables among the groups.