Genome-wide arrays in routine diagnostics of hematological malignancies

Authors

  • Annet Simons,

    1. Laboratory of Tumor Genetics, Department of Human Genetics, Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands
    Search for more papers by this author
  • Birgit Sikkema-Raddatz,

    1. Department of Genetics, University of Groningen, University Medical Center Groningen, Groningen, the Netherlands
    Search for more papers by this author
  • Nicole de Leeuw,

    1. Department of Human Genetics, Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands
    Search for more papers by this author
  • Nicole Claudia Konrad,

    1. Department of Medical Genetics, Cancer Cytogenetic Unit, Lausanne University Hospital, Lausanne, Switzerland
    Search for more papers by this author
  • Rosalind J. Hastings,

    1. Cytogenetic European Quality Assessment and United Kingdom Cytogenetic European Quality Assessment for Clinical Cytogenetics, John Radcliffe Hospital Women's Center, Oxford University Hospitals NHS Trust, Oxford, United Kingdom
    Search for more papers by this author
  • Jacqueline Schoumans

    Corresponding author
    1. Department of Medical Genetics, Cancer Cytogenetic Unit, Lausanne University Hospital, Lausanne, Switzerland
    • Department of Medical Genetics, Cancer Cytogenetic Unit, Lausanne University Hospital, 1011 Lausanne, Switzerland.
    Search for more papers by this author

  • For the Focus on CNV Detection with Diagnostic Arrays

Abstract

Over the last three decades, cytogenetic analysis of malignancies has become an integral part of disease evaluation and prediction of prognosis or responsiveness to therapy. In most diagnostic laboratories, conventional karyotyping, in conjunction with targeted fluorescence in situ hybridization analysis, is routinely performed to detect recurrent aberrations with prognostic implications. However, the genetic complexity of cancer cells requires a sensitive genome-wide analysis, enabling the detection of small genomic changes in a mixed cell population, as well as of regions of homozygosity. The advent of comprehensive high-resolution genomic tools, such as molecular karyotyping using comparative genomic hybridization or single-nucleotide polymorphism microarrays, has overcome many of the limitations of traditional cytogenetic techniques and has been used to study complex genomic lesions in, for example, leukemia. The clinical impact of the genomic copy-number and copy-neutral alterations identified by microarray technologies is growing rapidly and genome-wide array analysis is evolving into a diagnostic tool, to better identify high-risk patients and predict patients' outcomes from their genomic profiles. Here, we review the added clinical value of an array-based genome-wide screen in leukemia, and discuss the technical challenges and an interpretation workflow in applying arrays in the acquired cytogenetic diagnostic setting. Hum Mutat 33:941–948, 2012. © 2012 Wiley Periodicals, Inc.

Introduction

Cytogenetic analysis, typically comprising conventional karyotyping of 20 metaphases and/or targeted fluorescence in situ hybridization (FISH), has proven extremely valuable in the clinical management of hematological malignancies. The identification of translocations, inversions, duplications, deletions, and whole chromosome aneuploidies are crucial in establishing a diagnosis and evaluating the prognosis, but most importantly, in making therapeutic decisions. However, conventional karyotyping is relatively costly due to its laborious nature; it also has technical limitations, and many potentially clinical relevant submicroscopic chromosomal abnormalities remain undetected.

The genetic complexity of cancer cells requires a sensitive technology enabling the detection of small genomic changes in a mixed cell population, as well as the ability to detect segmental regions of homozygosity, also known as regions of copy-neutral loss of heterozygosity (CNLOH) or referred to as acquired uniparental disomy (aUPD). Segmental aUPD can lead to the homozygosity of a preexisting pathogenic mutation, providing a growth advantage to an, already mutant, clone. Microarray-based copy number and genotype analysis, using high-density whole-genome comparative genomic hybridization (CGH) or single-nucleotide polymorphism (SNP) arrays (referred further on in this article as molecular karyotyping), has the important advantage over conventional karyotyping in that arrays are not dependent on the attainment of mitotically dividing cells within the tissue of investigation, as genomic DNA from tumor cells is used instead of metaphases. The resolution is determined by the genomic distance between the probes, as well as their sizes, and the information molecular karyotyping provides is directly linked to the physical and genetic map of the human genome. Microarrays, therefore, allow the identification of very small copy-number aberrations (CNAs) at high accuracy. Although commercially available arrays have been shown to be an indispensable tool for diagnosing patients with intellectual disabilities and/or multiple congenital abnormalities [Vermeesch et al. (2012), this issue], it has been more challenging to implement the technology in the diagnostic hematological genetic setting. Nevertheless, array-CGH (aCGH) or SNP arrays have been shown to be a cost-effective alternative to multiple FISH testing to identify genomic imbalances. The added clinical value of arrays has recently been demonstrated for a number of hematological diseases, such as chronic lymphocytic leukemia (CLL), myelodysplastic syndrome (MDS), multiple myeloma (MM), acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), and chronic myelomonocytic leukemia (CMML) [Dougherty et al., 2011; Dunbar et al., 2008; Gunnarsson et al., 2011; Hagenkord et al., 2011; Heinrichs et al., 2009; Shao et al., 2010; Simons et al., 2011; Tiu et al., 2009, 2011a, 2011b]. Consequently, array-based molecular karyotyping is slowly but surely finding its way into the clinical laboratories for acquired cytogenetics.

Molecular Karyotyping and Its Technical Aspects

During the past decade, many array platforms have been developed and optimized, resulting in a large variety of affordable commercial arrays being available. The arrays contain either nonpolymorphic copy-number probes or polymorphic probes (SNPs), or a combination of both, and each platform has its advantages and limitations, which are discussed below (see also further discussion by Vermeesch et al. (2012), this issue).

Microarray-Based Comparative Genomic Hybridization

Array-CGH is highly flexible for customized designs, facilitating dense coverage of nonpolymorphic, that is, unique, sequence probes of selected genomic regions and avoiding highly polymorphic genomic loci present in the general population. The probe design on CGH arrays does not depend on the genomic location of SNPs, as is the case for SNP arrays. Genomic regions containing SNPs will be better covered on an SNP-based array, whereas regions of “SNP deserts” will be poorly covered. The flexible probe design allows reliable detection of very small genomic changes, down to a single exon, and can be used for accurate breakpoint determination [Lindstrand et al., 2010].

Array-CGH offers also a high sensitivity for detecting small CNAs in a small proportion of the cell population tested, due to a higher signal-to-noise ratio, resulting in a better dynamic range [Kresse et al., 2010; Zhang et al., 2008]. Array-CGH has been shown to reliably detect deletions < 1 Mb in 11% of the cells, even in a suboptimal DNA sample isolated from two unstained bone marrow smears (see Fig. 1A).

Figure 1.

Detection of type I and type II deletions at 13q14 in CLL. A: Array-CGH signal intensity log2 ratio plot from Cytosure software (Oxford Gene Technologies, UK) displaying an 800 kb type I deletion at the 13q14 locus in a CLL sample, confirmed by interphase FISH on a bone marrow smear in 11% of cells. The genomic DNA hybridized on the array was isolated from cells taken from two unstained bone marrow smears. B: SNP array ratio plot from Genome Studio software (Illumina), displaying the B-allele frequency plot and the signal intensity logR ratio from two consecutive deletions (4.2 and 2.9 MB, respectively), at chromosome locus 13q14–13q21, indicating a type II deletion in a CLL sample. The deletions were confirmed by interphase FISH on a bone marrow smear in 28% of cells. The deletion is visible by the B-allele frequency, while it is very subtle in the logR ratio plot.

Genome-Wide SNP Array

Single-nucleotide polymorphism arrays were primarily developed for genetic association studies, whereas, in the meantime, they have also proven to be extremely useful for detecting copy-number changes. They simultaneously provide genotyping information, which makes the detection of somatic CNLOH or aUPD possible. As a consequence, arrays containing polymorphic probes (SNP arrays) have been widely utilized for molecular karyotyping of tumors. In particular, somatic large stretches of homozygosity without loss of gene dosage have been found in variable frequencies (20–50%) in blood and bone marrow samples from patients with different types of leukemia. It is recognized that CNLOH can generate homozygosity for mutated tumor suppressor genes or oncogenes involved in tumor transformation. For this reason, the majority of studies on hematological neoplasms have been performed using SNP arrays [Dougherty et al., 2011; Dunbar et al., 2008; Hagenkord and Chang, 2009; Heinrichs et al., 2009; Tiu et al., 2009, 2011b]. Initially, the sensitivity for detecting genomic abnormalities in mixed cell population using SNP arrays was rather low due to software limitations, but the sensitivity increased dramatically when the B-allele frequency algorithm was added to the analysis software (Fig. 1B). Reliable detection of low-grade clonal mosaicism, at least in the range of 20–30% in leukemic samples, has now been reported [Dougherty et al., 2011; Heinrichs et al., 2010].

Cancer-Specific Arrays with Combined Nonpolymorphic and Polymorphic Probes

Specific cancer designs have already been used for array-CGH analysis of hematological malignancies, covering selected cancer-associated genes [Shao et al., 2010]. Furthermore, the Cancer Cytogenomics Microarray Consortium (http://www.urmc.rochester.edu/ccmc/) was formed to facilitate the sharing of data between centers across the world, developing guidelines for diagnostic array analysis, and making a robust cancer-specific custom probe design based on the gene list at the Wellcome Trust Sanger Institute (http://www.sanger.ac.uk/genetics/CGP/Census) and has been validated in multiple laboratories in order to set quality criteria. Currently, several commercial array platforms offer the combination of probe designs, covering hundreds to thousands of cancer-associated genes and allowing the detection of even single exon deletions or duplications of selected genes known to be important prognostic genetic markers (e.g., IKZF1, a gene frequently deleted in childhood ALL patients and, as such, associated with a high relapse risk [Mullighan et al., 2009]). These cancer-focused designs are combined with a high-resolution whole-genome backbone, simultaneously permitting the identification of novel lesions elsewhere in the genome. Furthermore, SNPs have been added to the CGH arrays to allow the detection of copy-neutral homozygous regions. However, the SNP coverage on the combined arrays is lower than on the traditional SNP arrays. Although SNP array analysis has the ability to identify very small CNLOH, the cutoff for a clinical hematology genetic setting is often set at 10–25 Mb [Maciejewski et al., 2009; Simons et al., 2011], as all commercially available array platforms can reliably detect CNLOH down to 10 Mb.

Arrays for Detection of Cancer-Specific Recurrent Balanced Translocations

A common limitation of SNP-array and CGH-array is the inability to identify balanced translocations and inversions. A modified array protocol, in which, prior to the hybridization step in the array procedure, a linear PCR amplification is performed across known recurrent translocation breakpoints in hematological neoplasms, enables detection of copy-number changes close to or at the breakpoints, as well as known recurrent translocations [Greisman et al., 2011]. This approach, also named translocation CGH (tCGH), not only allows accurate delineation of the breakpoints but, more importantly, also can identify the translocation partner (e.g., the MLL gene can fuse with about 70 different partners). tCGH arrays for a wide range of hematological recurrent translocation are commercially available (PerkinElmer, Spokane, WA). To combine tCGH and genome-wide copy-number analysis on one single array is only possible for targeted translocations. However, these types of arrays do not (yet) contain SNPs and, thus, are not able to identify regions of CNLOH.

Matched Normal Genomic DNA

A considerable quantity of copy-number variants (CNVs) have now been identified in the human genome. The detection of these constitutional variants cause confusion as to their clinical significance and databases collecting benign CNVs, such as the Database Genomic Variant (DGV; http://projects.tcag.ca/variation/), and dbVar (http://www.ncbi.nlm.nih.gov/dbvar) are indispensable tools when interpreting high-resolution, whole-genome array data [Church et al., 2010; De Leeuw et al. (2012), this issue].

Inheritance of two copies of the same germ line haplotype (CNLOH) has been found in up to 12% of normal controls when screening with SNP arrays [Tiu et al., 2011a]. It is, therefore, important to distinguish somatic disease-associated CNAs and CNLOH from germ line CNVs and copy-neutral aberrations, which requires a matched normal genomic DNA sample [Heinrichs et al., 2010]. In leukemia patients, whole blood cannot serve as a matched constitutional sample, as nonaffected tissue is needed. An optimal sample would be genomic DNA isolated from a skin biopsy, but it is less invasive and easier to obtain a buccal swab sample from the leukemia patient. Usually, a buccal swab sample can yield ∼2 µg DNA, which is sufficient for an array analysis. When using array-CGH technology, somatic genomic changes can be distinguished from constitutional CNVs in a single assay, if the matched DNA sample is used as reference DNA in the same hybridization. This will, in addition, reduce the cost per analysis. However, if SNP arrays are used, this would double the costs, as two separate array tests have to be run. In some countries, a genetic counseling session as well as a signed informed consent from the patient is required prior to obtaining constitutional DNA for testing and reporting unsolicited detected constitutional abnormalities. Genetic counseling is, however, not common practice when performing cytogenetic analysis searching for acquired genomic abnormalities in leukemia patients. When using a matched reference DNA in the same hybridization, the identification of constitutional aberrations can be avoided. The practical feasibility of the simultaneous collection of a normal matched sample (regardless of the tissue type) might prove to be difficult, as patients are referred from many different hospitals, where this extra step in routine care could be logistically complicated to implement.

Technical Challenges Relevant to Cancer Samples

Copy-number aberration and CNLOH analysis in cancer samples can be difficult, and obtaining reliable results will depend highly on the quantity and quality of the biological material tested. Figure 2 shows how different factors are gradually making array analysis more challenging (see also Vermeesch et al., 2012, this issue).

Figure 2.

Technical challenges using genomic arrays relevant to cancer samples. The gradual rise in difficulty level is displayed from light gray (easy, left side of the arrow) to black (difficult, right side of the arrow) depending on the size of the abnormality, number of abnormal copies, sample types, volume of material to investigate, mixed cell population, and detection of tumor heterogeneity. FFPE, formalin-fixed paraffin-embedded sample.

Clinical Implications of Molecular Karyotyping in Leukemia

Chronic Lymphocytic Leukemia

Chronic lymphocytic leukemia has a variable clinical course with a highly variable life expectancy [Swerdlow et al., 2008]. For risk classifications and clinical decision making in CLL, a standard panel of FISH probes or commercially available multiplex ligation-dependent probe amplification (MLPA) tests are usually used to identify deletions at 11q22 (ATM), 13q14 (DLEU1/2 and RB1) and 17p13 (TP53), duplications of 6q, and trisomy 12. Molecular karyotyping using arrays has proven an effective technique for detecting CNAs and aUPD at genomic regions with established prognostic significance in CLL, and provides an improved resolution compared with conventional karyotyping and FISH analyses [Hagenkord et al., 2011; O'Malley et al., 2011]. Furthermore, the clinically relevant genomic alterations in CLL involve deletions and duplications, whereas balanced translocations are relatively rare and are of unclear significance. Therefore, the multiple FISH or MLPA tests can easily be replaced by a single array test. Moreover, recent evidence has revealed heterogeneity within the 13q14 deletions, subdividing them into two distinct subtypes (type I deletions, < 2 Mb in size, not including the RB1 gene, and type II deletions, > 2 Mb in size, including RB1). The 13q14 type I and II deletions in CLL are biological and prognostic distinct entities [Ouillette et al., 2011; Parker et al., 2010]. Routine diagnostic analysis often comprises FISH analysis, utilizing locus-specific probes covering only the gene DLEU1/2, which permits the detection of deletions at the 13q14 locus, but not the accurate breakpoints or size of the deletion. For size mapping of this specific deletion, molecular karyotyping (Fig. 1) or MLPA is better suited and would allow a refined CLL patient risk stratification.

Myelodysplastic Syndrome

Myelodysplastic syndrome is a group of heterogeneous, clonal stem cell disorders, characterized by ineffective hematopoiesis, morphological dysplasia, and peripheral blood cytopenias [Swerdlow et al., 2008]. MDS is associated with significant morbidity and mortality, due to progressive bone marrow failure or evolution to AML, and the molecular pathogenic mechanisms underlying the transformation of MDS to AML are poorly understood. Currently, prognosis relies primarily on the World Health Organization (WHO) classification-based scoring system (WPSS) for MDS, which stratifies patients into risk groups based on the following three parameters: (1) number of cytopenias, (2) bone marrow blast percentage, and (3) cytogenetic risk, based on conventional metaphase karyotype and/or FISH analysis [Swerdlow et al., 2008]. In addition, it is important to identify patients with a deletion of the long arm of chromosome 5 because they respond well to the immune-modulating agent Lenalidomide. However, the 5q deletions are highly variable in size and include two different genomic regions, namely, deletions of chromosome band 5q31, encompassing the EGR1 gene, and deletions of bands 5q32 and 5q33 involving haploinsufficiency of RPS14. Molecular karyotyping enables accurate identification of both loci of the 5q deletions, which can be missed when applying only traditional metaphase karyotyping and FISH [MacKinnon et al., 2011].

Furthermore, normal karyotypes are found in 50–60% of the patients, for whom there are no molecular tests to distinguish MDS from benign bone marrow cell diseases making monitoring disease progression in these patients difficult. Emerging data demonstrate that MDS exhibits abundant CNAs and CNLOH, often in the setting of a normal metaphase karyotype and no previously identified clonal marker [Heinrichs et al., 2009; Thiel et al., 2011; Tiu et al., 2011b]. Tiu et al. (2011a) proposed a new prognostic risk score, integrating SNP array results in order to improve risk stratification for patients with MDS.

Chronic Myelomonocytic Leukemia

Chronic myelomonocytic leukemia is a clonal disorder of hematopoietic stem cells, often occurring in elderly patients, in which monocytosis is a major defining feature [Swerdlow et al., 2008]. It was originally classified as an MDS, based on the facts that dysplastic features were often present and progression toward acute leukemia was inevitable. However, the more recent WHO classifications system reclassifies CMML among myelodysplastic/myeloproliferative disease (MDS/MPN) [Swerdlow et al., 2008]. CMML is notoriously hard to treat and the prognosis is quite variable. No clear recurrent genomic alterations have been found by conventional karyotyping and array-CGH profiling has demonstrated a remarkable stable genome in the majority of the cases [Gelsi-Boyer et al., 2008]. Recently, CMML has been associated with somatic mutations in a growing number of genes. Somatic frameshift or point mutations are frequently found in CBL, TET2, EZH2, ASXL1, and the RAS pathway, while, in particular, alterations in TET2 and EZH2 have been associated with a poor prognosis [Grossmann et al., 2011]. SNP-based array analysis has identified large stretches of CNLOH regions as the sole genomic abnormality (Fig. 3A and B). These CNLOH regions may result in homozygous mutations with prognostic implications [Dunbar et al., 2008].

Figure 3.

CNLOH as sole abnormality in CMML and suggestions for nomenclature of array aberrations. A: LOH plot in Cytosure software (OGT) from a CMML sample, displaying a large region of CNLOH (red bar) on chromosome 4 and encompassing several cancer-associated genes with, in particular, TET2 at 4q24. The small black lines under the chromosome represent the heterozygous SNPs, whereas the homozygous SNPs are shown in the red lines. B: Copy-number plot of the same chromosome showing no copy-number change. C: Examples of different types of aberrations (arrows) detected in different leukemia samples by the 250 k SNP array platform (Affymetrix, Inc., Santa Clara, CA). In each example, the official description of the aberration according to ISCN 2009 [Shaffer et al., 2009] is given as well as the suggested translated molecular karyotype. Arrows indicate the different aberrations. 1: SNP array plot from chromosome 4 displaying a gain of the entire chromosome. 2: Interstitial loss of the EBF1 locus at 5q33.3. 3: CNLOH or aUPD of a large part of chromosome 1p, extending from 1p34.1 toward the telomere. 4: Gain of a large part of the 1q arm, from 1q23.3 to the end of the chromosome, a so-called telomeric duplication.

Acute Myeloid Leukemia

Acute myeloid leukemia is a heterogeneous group of myeloid neoplasias, for which, currently, apart from the mutational status of a few genes (e.g., FLT3, IDH1/2, CEBPA, or NPM1), cytogenetic analysis and age are the most important factors for risk assessment. The clinical management of AML patients mainly relies on detecting specific cytogenetic abnormalities linked to specific abnormalities that are subsequently correlated with specific treatment response and survival. Examples are t(8;21)(q22;q22)/RUNX1T1-RUNX1 and inv(16)(p13q22)/CBFB-MYH11, which are associated with a favorable outcome, and t(15;17)(q24;21)/PML-RARA, typical for acute promyelocytic leukemia, which can be specifically treated with an all-trans-retinoic acid [Swerdlow et al., 2008; reviewed by Grimwade et al., 2010]. However, almost half of AML cases presenting in the clinic display a normal karyotype. Patients with a normal karyotype and an NPM1 mutation but no FLT3 internal tandem duplication, or with a CEBPA mutation, have a favorable prognosis, whereas patients with a normal karyotype and a FLT3 ITD have a poor prognosis. An important predictor of therapy resistance and short survival is the monosomal karyotype, which is composed of at least two autosomal monosomies or one autosomal monosomy and one structural aberration [Breems et al., 2008; Estey, 2012]. A more detailed identification of somatic alterations (other than balanced translocations) may, therefore, provide crucial insights into the pathogenesis of AML and lead to a better tailored treatment. Recent array studies revealed acquired CNAs and regions of CNLOH that had an added independent prognostic impact on AML survival. In addition, regions of CNLOH were more often detected in patients with normal karyotypes than in aberrant karyotypes [Parkin et al., 2011; Tiu et al., 2009; Walter et al., 2009]. Furthermore, molecular karyotyping using genomic arrays has identified CNAs that are specifically detected in therapy-related AML, whereas other genomic alterations seem specific to primary AML [Itzhar et al., 2011].

The genomic CNAs identified in AML are not as common or recurrent as observed in ALL [Walter et al., 2009], but they do show, to some extent, overlap with anomalies found in MDS [Flach et al., 2011]. Interestingly, two recent studies have performed high-resolution array analysis on matched, paired normal, and leukemic samples of both MDS and AML patients, determining true somatic CNAs in genomic regions where frequently benign constitutional CNVs are identified in normal control individuals [Barresi et al., 2010; Starczynowski et al., 2011]. Because of the complex genomic architecture of these polymorphic regions, it is possible that copy-number changes, for example, are induced by stress on the hematopoietic system during disease progression of AML. These somatic CNAs are likely to contribute to disease evolution when they harbor genes involved in hematopoiesis or tumor development, but more data are needed to understand the underlying mechanisms [Starczynowski et al., 2011].

Acute Lymphoblastic Leukemia

Acute lymphoblastic leukemia often contains chromosome aneuploidies and recurrent chromosomal translocations, which are of prognostic importance and routinely used in clinical decision making. Hyperdiploidy (i.e., 47–54 chromosomes), for example, is associated with a relatively favorable prognosis, whereas hypodiploidy (35–45 chromosomes) is associated with a poor outcome. In addition, specific recurrent translocations in ALL, such as the Philadelphia translocation t(9;22)(q34;q11.2) and the t(12;21)(p13;q22) translocation resulting in BCR-ABL1 and ETV6-RUNX1 fusion genes, respectively, and MLL (11q23) rearrangements, have an impact on prognosis [Swerdlow et al., 2008]. Apart from these well-known aberrations, in recent years, many submicroscopic copy-number alterations with proven relevance to diagnosis, prognosis, and therapy response in ALL have been discovered by employing genomic array technologies [Collins-Underwood and Mullighan, 2010; Kuiper et al., 2007; Mullighan et al., 2007]. Examples of these aberrations include deletions in the genes CDKN2A/B, BTG1, IKZF, and EBF1. Besides the general inability of conventional karyotyping to detect such small CNAs, another large limitation of this technique in ALL is the low success rates due to low mitotic index and poor morphology and, therefore, poor chromosome banding quality. In a comparative study of conventional karyotyping versus molecular karyotyping using SNP arrays, the added value of arrays in routine diagnostics of ALL was demonstrated [Simons et al., 2011].

Multiple Myeloma

Multiple myeloma is a heterogeneous disease with a variable disease course, as well as response to therapy [Swerdlow et al., 2008]. MM is a plasma cell malignancy with cytogenetic instability and the molecular mechanisms underlying the development of MM are not yet fully understood. However, the identification of genomic rearrangements has demonstrated evidence of their importance in the classification of MM. Myeloma samples typically consist of a heterogeneous cell population, due to the low proliferation rates of the abnormal plasma cell populations. This hampers conventional karyotyping, and in most diagnostic laboratories, interphase FISH analysis is used as an adjunct, or has even replaced conventional karyotyping. A panel of probes to detect clinically relevant genetic aberrations is used (i.e., deletions at 13q14/DLEU1/DLEU2/RB1, 17p13/TP53 and 1p, duplications of 1q, hyperdiploidy [probes for chromosomes 5, 9, 11, 15 are frequently used], and rearrangements of 14q32/IGH). In addition, if plasma cells are purified prior to genetic testing, complex chromosome abnormalities are found in a significant proportion of the samples.

Multiple myeloma can roughly be divided into two genetic entities, the nonhyperdiploid tumors that often have a chromosomal translocations involving the immunoglobulin heavy chain (IGH) locus, and the hyperdiploid tumors that have a substantially lower prevalence of IGH translocations and, in general, have a better prognosis. In order to distinguish the two genetic groups and detect other relevant genetic lesions (see above), molecular karyotyping has the advantage of detecting CNAs across the whole genome, whereas FISH analysis enables only the interrogation of a limited number of loci within the same assay. In recent years, the use of genome-wide microarray-based genomic profiling has proven to be very valuable in the detection of clinically relevant CNAs (Avet-Loiseau et al., 2009; Agnelli et al., 2009; Walker et al., 2010) in MM. For characterization of IGH rearrangements, a translocation assay, such as the tCGH approach described above, could be employed to identify the presence of a translocation, as well as the partner gene, in a single hybridization.

Standardized Workflow and Array Data Interpretation for Leukemia Diagnostics

The use of molecular karyotyping using arrays, in a routine clinical setting for leukemia diagnostics, requires guidelines for standardized data interpretation and handling, and a uniform way of reporting the results. Recently, Simons et al. (2011) proposed an objective and standardized workflow for practical routine diagnostic use for ALL (see Fig. 4). This was based on a comparison of results obtained from 60 ALL patients in which conventional and molecular karyotyping was performed using a 250k SNP array platform in a diagnostic setting [Simons et al., 2011]. This approach is not limited to the management of ALL and a similar approach is easily applicable to other hematological malignancies in which CNAs are common. Of importance is the tumor load of the tested sample, which may differ considerably between different malignancies. For whole-bone marrow samples, where a low percentage of tumor cells would be expected, a tumor cell enrichment prior to DNA isolation is recommended. The general criteria for interpretation are given below, which were initially set up for the Affymetrix 250k SNP array platform, but may be adapted for other (higher resolution) platforms.

Figure 4.

Standardized workflow for array data interpretation and decision making in a hematooncogenetic diagnostic setting. A: Present in: Database of Genomic Variants (http://projects.tcag.ca/variation), dbVar (http://www.ncbi.nlm.nih.gov/dbvar), In-house databases of healthy persons. B: Containing genes or regions denoted as recurrent and annotated in cancer; present in: Atlas of Genetics and Cytogenetics in Oncology and Hematology (http://atlasgeneticsoncology.org), Mitelman Database of Chromosome aberrations (http://cgap.nci.nih.gov/Chromosomes/Mitelman), The cancer gene list of the Sanger institute (http://www.sanger.ac.uk/genetics/CGP/Census), Progenetix genome copy-number aberrations in cancer (http://www.progenetix.net/cgi-bin/pgHome.cgi), Other (in-house) defined lists of relevant genes, for example, specific cancer subtypes. C: TCR (T-cell receptor) or IG (immunoglobulin) genes: 2p12(IGK), 7p14.1(TCRG), 7q34(TCRB), 14q11.2(TCRA and TCRD), 14q32.3 (IGH), 22q11.2(IGL). D: The CNA is filed as a potential aberration and not reported in the final diagnostic result, unless it is confirmed by another test.* The limit of 5 Mb may be set differently.

All CNAs larger than 5 Mb called by the software algorithm used [see also de Leeuw et al., 2012, this issue] or detected by visual inspection, regardless of gene content, are denoted as true aberrations, with the exception of those known to be normal genomic variants (present in DGV [http://projects.tcag.ca/variation], dbVar [http://www.ncbi.nlm.nih.gov/dbvar], and/or found in in-house databases of healthy persons). The limit of 5 Mb has been arbitrarily chosen (initially determined because it matches the resolution of conventional karyotyping), but it may also be set to a higher resolution, depending on the type of array platform used or if matched tumor and reference DNA samples are used.

All CNAs called by the software algorithm and smaller than 5 Mb that are not excluded as normal variants are interpreted as aberrant only where they coincide with known (recurrent and annotated) aberrations previously reported in the literature. Several other databases can also be consulted such as the Atlas of Genetics and Cytogenetics in Oncology and Hematology (http://atlasgeneticsoncology.org/), Mitelman Database of chromosome aberrations (http://cgap.nci.nih.gov/Chromosomes/Mitelman), the cancer gene list of the Sanger institute (http://www.sanger.ac.uk/genetics/CGP/Census), and Progenetix genomic CNAs in cancer (http://www.progenetix.net/cgi-bin/pgHome.cgi). Of note, when using data from public databases for clinical interpretation, caution is warranted and one should verify the latest updates, if it contains curated data and what techniques (resolution) were used. For practical use, laboratories may build their own in-house database with defined genes and loci, which often exists as an option in the analysis software. However, there is a clear need for comprehensive publicly available and well-updated databases containing genomic lesions and their clinical implications in each specific hematological disease. This would facilitate global sharing of accumulating array data. Meanwhile, array analysis software packages are rapidly improving by adding “cancer relevant” tracks from growing experience on different types of leukemia, to assist interpretation. All other CNVs not fulfilling the above criteria are not included in the diagnostic report but filed as potential aberrations, unless they are confirmed by another test (e.g., FISH or MLPA).

Regions of homozygosity are only interpreted as aUPD, if they are > 10 Mb in size and if they extend toward the telomeres of the chromosomes involved [Heinrichs et al., 2010]. If nonmatched references are used, there will sporadically be a case with many regions of homozygosity larger than 10 Mb, which are mostly interstitial. This is highly likely constitutional LOH, resulting from consanguinity. In such cases, preferably matched normal DNA should be used in the array analysis, or regions of homozygosity should be left out of the analysis completely.

Focal deletions in antigen receptor genes (T-cell receptors and immunoglobulins) are considered to represent nonleukemic genomic rearrangements caused by (mono) clonality and are, for that reason, excluded from the final diagnostic report.

Diagnostic Report and Nomenclature of Detected Aberrations

As tumor samples often harbor many genetic and cytogenetic aberrations, the description of the genomic profiles obtained using the standardized International System for Human Cytogenetic Nomenclature (ISCN) [Shaffer et al., 2009] may result in complex reports, which are difficult for the referring medical doctor to interpret. We therefore recommend that, besides the reporting of all detected CNVs using the standardized ISCN nomenclature, the genomic profiles should also be converted into so-called microarray-deduced copy-number karyotypes and these should be added to the final diagnostic report. We suggest describing such a molecular karyotype (based on the ISCN nomenclature for conventional karyotyping) as illustrated in the following theoretical example:

“A genome-wide array analysis of a female bone marrow sample reveals gains of whole chromosomes X, 5 and 7, an acquired homozygous region on 8p (p24 to telomere), a loss of a part of 9p21 (interstitial), a gain (triplication) of 10q (q24 to telomere), and a gain (duplication) of 11pq.”

The official ISCN 2009 nomenclature would be:

equation image

where A is the starting base position and B is the end base position. The genomic browser type and version defining the base positions mentioned in the karyotype should also be included in the report.

The deduced copy-number karyotype would be:

equation image

A few examples of array plots of different types of aberrations and their descriptions according to ISCN 2009 nomenclature and subsequent translation to a molecular karyotype are shown in Figure 3C.

Finally, in the diagnostic report, the karyotype is further clarified and relevant genes located in the detected lesions and their clinical relevance (e.g., diagnostic and prognostic impact) should be mentioned with references [see also Vermeesch et al. (2012), this issue]. Regarding the reporting of technical information, the guidelines for molecular karyotyping for constitutional genetics diagnosis can be used [Vermeesch et al., 2007, 2012, this issue].

Conclusion

High-resolution molecular karyotyping is already an indispensable tool in constitutional genetics and it has started to become an integrated part in the clinical management of leukemia patients. It will, ultimately, be faster, more accurate, and more cost-efficient to elucidate somatic aberrations in hematological malignancies. Molecular karyotyping has already proven to be essential for identifying novel genomic abnormalities that escape detection with current diagnostic methodologies. The identification and accurate genomic mapping of genomic alterations in hematological disease have shown it is possible to refine the current risk stratification of patients and will eventually contribute to developing enhanced treatment modalities. Furthermore, global sharing of the genomic and pathological data that are now accumulating in publicly available databases will aid in better understanding the genetic mechanisms and driving pathogenic abnormalities.

In the future, with the introduction of massive whole-genome parallel sequencing, an even more complete map of the genomic changes present in malignant cells will be obtained. Although this technology can unmask known and novel cancer-associated genes, currently, it is only used for research purposes and is not yet applicable in the clinical diagnostic hematology cytogenetic setting due to cost, technical, and ethical reasons. However, it has already been demonstrated that high-resolution molecular karyotyping can be implemented to aid the diagnosis and prognosis of neoplasias.

Acknowledgements

We thank Jackie Senior for editorial advice and Marian Stevens-Kroef for critical comments on earlier drafts of this manuscript.

Disclosure Statement: The authors declare no conflict of interest.

Ancillary

Advertisement