Chromosomal alterations of pediatric malignancy in a West Texas population

Background: Comparative genomic hybridization (CGH) microarrays are used for genome-wide evaluation of copy number variations (CNV) of known prognostic significance; however, unannotated variants of uncertain significance (VUS) are frequently present. To identify potentially actionable targets, we retrospectively analyzed VUS loci using CGH data from 192 cases of cancer or genetic disorders treated at TTUHSC. Methods: DNA was hybridized onto CytoSure Constitution V3 arrays, scanned with the Agilent microarray D scanner, and analyzed by CytoSure Interpret Software. Results: We found 794 distinct CNVs, the most frequent being 14q32.22 (112 rearrangements), 14q11.2 (100), 8p11.2 (98), 15q11.1-q11.2 (83), and 8p23.1 (77). In particular, 8p11.22 alterations were found in many pediatric tumors, with gain/loss ratio of 4.7. Linkage of TACC1, TM2D2, KAT6A and ADAM32 was indicated by a similar 5-year survival rate of 75.3% (n = 253), which was greater than in unaltered cases (62.2%, n = 15,809 cases) in The Cancer Genome Atlas database. Conclusion: Knockdown of genes occurring at variants of uncertain significance (VUS) loci may help identify new therapy targets.

by the botanists Hugo DeVries, Carl Correns, and Erich von Tschermak, 4 the discovery of the role of chromosomes as the microstructural basis of heredity can be attributed to Theodor Boveri, who used sea urchin cells to show that concerted interactions of chromosomes, centrosomes, and centrioles were necessary for transmission of genetic traits. 5 Progress in cytogenetics was hampered by the physical limits of light microscopy and the vagaries of cytological preparation, as exemplified by the fact that even the number of human chromosomes was not correctly determined until 1956 6 after over 7 decades of estimation. 7,8 Improvements in microscopy proved critical for the development of the science of cytogenetics, as microscopy provided crucial details on the number, size, shape, and staining patterns of chromosomes.

Rajan et al. Chromosomal Alterations of Pediatric Malignancy in a West Texas Population
Cytogenetic knowledge, combined with the work of numerous biochemists, biophysicists, and molecular biologists, provided evidence for the one-gene one-enzyme theory, 9 the function of nucleic acids as genetic biomolecules, 10 elucidation of the structure of DNA, 11 nucleic acid sequencing methods, 12 recombinant DNA technology, 13 microarrays, and massively parallel nucleotide sequencing methods 14 which comprise modern clinical human genetics, the central driver of the progress of modern medicine to cure diseases.
Present day clinical cytogenetics evolved through standardization of cytological preparation and image processing techniques yielding consistent identification of aberrations based on characteristic alternating light and dark colored stripes that appear when denatured chromosomes are stained with Giemsa stains (G-bands) 15 or fluorescent stripes when stained with quinacrine (Q-bands) denaturing processes. 16 Improved cytological methods allowed visualization of chromosomal aberrations associated with numerous hereditary syndromes 17 including large-scale chromosomal rearrangements, monosomies, and trisomies.
High-resolution cytogenetic analyses have revealed less readily detectable but increasingly identifiable microdeletion and microamplification syndromes, some eponymous such as Zelweger syndrome 18 and Norrie disease, 19 and others referred to only by the abnormal chromosomal locus, i.e., microdeletion 8p21.2, microdeletion 8p12p21, microamplification 8p23.1, and over 100 others. 20 In addition to hereditary disorders recognizable by congenital malformations or biochemical abnormalities, as predicted by Boveri, 21 the science of cytogenetics has revealed the role of germline as well as somatic chromosomal aberrations in the pathogenesis of cancer. 22 Indeed, each of the above syndromes is associated with rare forms of pediatric and adult malignancies. The first oncogenic somatic chromosomal aberration identified was the Philadelphia chromosome in chronic myelogenous leukemia, leading to a revolutionary treatment that inhibits the kinase encoded at the t(9;22)(q34;q11) locus. 23 Since then, chromosomal aberrations have been established as driver events, prognosticators, or predictors of therapeutic responses for multiple types of malignancies, including many hematologic cancers, 24 sarcomas, 25 gliomas, 26 prostate cancer, 27 and many others. 28 Cytogenetic studies have also identified numerous chromosomal loci with putative tumor suppressor genes associated with hereditary cancers or syndromes, including 29-31 neurofibromatosis, retinoblastoma, ataxia telangiectasia, Cowden's syndrome, Li-Fraumeni syndrome, and others. 32 The advent of fluorescence in situ hybridization (FISH), a test that detects copy number variations (CNV) based on annealing of fluorescent nucleotide probes to chromosomes, has greatly increased the power of clinical cytogenetic testing. 33,34 For example, detection of ERBB2 gene amplification is the clinical standard that determines whether breast cancer can be treated with anti-ERBB2 antibodies trastuzumab and pertuzumab. 35 In addition, several multiprobe FISH assays are broadly used clinically to detect amplification or deletion of genes, guiding prognostic as well as predictive biomarkers in a wide spectrum of malignancies including myeloid and lymphoid leukemia, 36,37 lymphoma, 38,39 multiple myeloma, 40 melanoma, 41 sarcoma, 42 and lung cancer. 43 Despite its clinical utility, FISH does have several limitations including false positivity due to occasional problems in detection or in interpretation of signal patterns in interphase nuclei, 44 inability to detect some inter-chromosomal rearrangements, 45 and inability to detect rearrangements under approximately 300 kb. 46 It is very useful clinically for confirming suspected copy number variations for loci already recognized as pathogenic, or as classifiers, predictors, prognosticators, or biomarkers for classification, but it is not wellsuited for discovery of new potentially pathogenic genes.
Comparative genomic hybridization (CGH) microarrays allow simultaneous detection of thousands of CNVs across the entire genome. Fluorescencelabeled total genomic DNA of the test sample is hybridized to metaphase chromosomes or DNA microarrays, quantified using array scanners, and normalized to a reference DNA sample, i.e., non-cancerous tissue. The intensity ratio correlates with CNVs in regions as short as a few thousand or even a few hundred bases. 47,48 Technical complexities that can yield false positives or negatives have been largely overcome through stringent validation, multiple array elements, and computational methods such that clinical application is now quite feasible and widely used. Though the aim is to detect pathogenic CNVs with clinically significant implications, many CNVs have been observed that can only be classified as variants of uncertain significance (VUS), which pose significant challenges 49,50 but also the opportunity to discover novel pathogenic mechanisms in cancer and other hereditary diseases.
Toward this goal, we catalogued CNVs observed in our West Texas pediatric (as well as some adult) oncology patients, identified CNVs present at unexpectedly high frequencies, and evaluated potential oncogenic roles of amplified genes by assessing the effects of gene knockdown on the growth of human cancer cell lines. We found a very high frequency of 8p11.2 amplification, with 4 genes (TACC1, TM2D2, KAT6A and ADAM32) residing at the locus.

MAterIAls And Methods
The Maxwell 16 LEV blood DNA kit and Maxwell 16 instrument were used to extract the DNA from peripheral blood leukocytes, bone marrow aspirates, and tumor samples. The Maxwell 16 MDx (Promega, Madison, WI, USA) cartridge was used for DNA purification, and NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) was used for estimation of purity. Purified slides were hybridized onto the CytoSure Constitution V3 array (8x60K) for 24 hours, scanned with the Agilent microarray D scanner, and evaluated for CNVs, LOH and uniparental disomy with the included CytoSure Interpret Software. All methods meet CAP/CLIA requirements. Software was updated using the Deciphering Developmental Disorders (DDD) study and the latest updates from ClinGen, the Clinical Genome Resource. Where applicable, conventional cytogenetics and FISH assays were used to validate results. Data were de-identified and no identifying registry was created. The data were sorted by karyotype, microarray results of gain or loss, diagnosis, gender, and age.

results
Our aim was to determine the overall frequency of each chromosomal aberration and to discover novel disease associations. The sample population of the study consisted of 192 individuals, 166 of whom were pediatric patients. The ages ranged from 1 month to 70 years, with the median age being 10.1 and 20.7 years for the pediatric and adult patients, respectively. The male/female ratio was 1.1 and 1.2 for the pediatric and adult patients, respectively. Final histopathological diagnosis was unavailable for 28 cases, identified on test requisitions only by suspected diagnosis, clinical reason to perform the test, or site of biopsy. While all 192 patients were included in calculating the relative frequency of chromosomal alterations, cases without definitive pathological diagnosis were excluded from disease-specific analyses.

Frequently Altered chroMosoMAl locI
We found 794 unique chromosomal loci. The ratio of gains to losses was 1.04. At loci with gains, 11.2% involved mosaic gains and 5.0% had two copy gains; at loci with losses, mosaic loss was found in 10.0% and two-copy loss was present in 4.0%. Among the

MIcroduplIcAtIon syndroMes
In recent years, a number of microdeletion and microduplication syndromes have been associated with specific rearrangements at several chromosomal loci. [59][60][61][62][63] These are often associated with dysmorphic features or other phenotypes but have not been specifically linked to malignancies. Some of the most common loci in our study (14q11.2, 8p11.2, 8p23.1, 10q26.13, and 17q12) are involved in these syndromes. Clinical features commonly involve facial abnormalities, developmental delay, intellectual disability, and cardiac defects, although unique features include precocious puberty in 8p11.2; inverted duplication, cryptorchidism, and diaphragmatic hernia in 8p23.1 deletion syndrome, and maturity onset diabetes of the young in 17q12 deletion syndrome. A potential association of these syndromes with cancer has not been previously reported.

Rajan et al. Chromosomal Alterations of Pediatric Malignancy in a West Texas Population
This locus has been of interest in breast 65 and hepatocellular carcinoma, 64 and evidence suggests that specific genes at this locus are altered in colorectal, prostate, gastric and renal cell cancer. 68 There have been multiple oncogenes associated with this site, including FGFR1 at 8p11.23. 69 There have also been multiple cancer-related fusion proteins and translocations identified involving this site. 66 The high ratio of gains to losses (80 vs. 18) in our study would support the possibility of multiple oncogenes at this locus.
The three segments of 8p11.2 involved in our study (8p11.21, 8p11.22, 8p11.23) have all been reported to contain several genes that contribute to the development of cancer (Figure 1). These include KAT6A, TACC1, TM2D2, and ADAM32. 67,70 Notably, 84 out of 98 patients had 8p11.22 rearrangements The ADAM32 (Metallopeptidase Domain 32) gene encodes a disintegrin family membrane-anchored protein involved in brain development, fertilization, tumor development and inflammation. It is located in a cluster of other disintegrin and metallopeptidase family genes on chromosome 8. Alternative splicing results in multiple transcript variants encoding different isoforms. TACC1 (Transforming Acidic Coiled-Coil Containing Protein 1) is a candidate breast cancer gene located near FGFR1 at 8p11. 8 that is amplified in some breast cancers. It is possibly involved in promoting cell division prior tissue differentiation. KAT6A (Lysine Acetyltransferase 6A) is a member of the MOZ, YBFR2, SAS2, and TIP60 family of histone acetyltransferases. It is a part of a complex that acetylates lysine-9 residues in histone 3. Chromosomal translocations of this gene have been associated with acute myeloid leukemia. Alternative splicing results in multiple transcript variants. TM2D2 (Beta-Amyloid-Binding Protein-Like Protein 1) encodes a structural module related to the seven transmembrane domain G protein-coupled receptor superfamily. This protein may have regulatory roles in cell death or proliferation signal cascades. This gene has multiple alternatively spliced transcript variants which encode two different isoforms.
in a segment containing ADAM32 and pseudogene ADAM3A. 30 ADAM32 has previously been found to be elevated in breast, colorectal, gastric, ovarian, prostate, renal, lung, and pediatric medulloblastoma 68 as well as hepatocellular carcinoma. 64 We found ADAM32 was gained in acute lymphoblastic leukemia, neuroblastoma, fibromatosis, Wilm's tumor, osteosarcoma, rhabdomyosarcoma, lymphoma, testicular cancer, and adenomatous rectal polyps, malignancies in which this association has not been previously reported. We reasoned that if this locus contained oncogenes, survival of patients with copy number gains would be worsened, and if tumor suppressors were present, survival could be improved. We queried TCGA (The Cancer Genome Atlas) using cBioportal to determine the overall effect of alterations of the four genes in all malignancies (n = 15, 809). 71 We found that the percentage of survivors at 5 years was greater in all cases in which any gene was altered as compared with cases without alterations ( Figure 2). Interestingly, the percent of survivors (75.3 + 1.7%) was nearly the same regardless of which gene was queried, suggesting co-occurrence of alterations, as would transpire if the amplified locus spanned all four genes in the majority of altered cases. Overall, gains significantly outnumbered losses across all cancers, except for the fibrolamellar variant of hepatocellular carcinoma where all altered cases were losses ( Figure 3). The similarity of histograms of alterations across cancers for all four genes also supported strong linkage (Figure 3). Though relatively little is known of the function of TACC1, it appears to participate in the mitotic spindle complex and is possibly regulated by Aurora kinases (Figure 4). 72,73 dIscussIon Our study is the first retrospective data analysis on microarray data for cancer patients in the West Texas region, and the results showed a number of findings of interest that have not been reported previously. In our primarily pediatric patient population with the main diagnoses of ALL, lymphoma, sarcoma, Wilm's tumor and neuroblastoma, the five most common rearrangements were at loci 14q11.2, 14q32.33, 15q11.1-15q11.2, 8p23.1, and 8p11.2. The two rearrangements with the most discernible patterns of expression were 8p23.1 and 8p11.2. The 8p11.2 locus has been previously reported to play a role in cancer and had narrowest range of chromosomal coordinates, allowing us to identify affected genes with greater precision. Though survival data from TCGA suggested that the four genes at this locus -ADAM32, TACC1, TM2D2 and KAT6Ashould be tumor suppressors, we found that knockdown of individual genes actually inhibited growth; the exception was growth stimulation only in melanoma following ADAM32 knockdown, a finding possibly related to complex effects on cancer cell survival with regard to multiple gene amplifications in humans. Knockdown studies confer value by allowing individual genes to be validated as possible cancer therapy targets. Ours is the first report focused specifically on the patient population of West Texas, an ethnically diverse cohort with exposure to significant environmental factors including extremely iron-rich soil contaminated with numerous herbicides such as carcinogenic glyphosate, pesticides such as endosulfan, and hydrocarbon pollutants from extensive oil and natural gas drilling. 74 Furthermore, the region may be impacted by the founder effect, as the population has remained relatively isolated.
This novel information has the potential to expand knowledge on the effects of microdeletion and microduplication syndromes as the most common loci in our study (14q11.2, 8p11.2, 8p23.1, 10q26.13, and 17q12) are involved in many syndromes, the clinical features of which primarily center around developmental anomalies. 14q11.2 is involved in both duplication and deletion syndromes, both of which involve facial anomalies such as hypertelorism. Duplication of 14q11.2 also involves ataxia, stereotypical hand movements, hypotonia, and cleft palate. 59 8p11.2 is involved in both inverted duplication and deletion syndromes, the former of which is more likely to be present in our patient population. 60 Clinical features of inverted duplication include developmental delays in speech and language, prominent forehead with inverted nostrils, short neck, precocious puberty, heart defects, abnormally increased height, initial hypotonia with developing hypertonia, spastic quadriplegia, and contracted joints in adulthood. Duplication and deletion syndromes have also been described for 8p23.1, the deletion syndrome, which matches the data in our study, involves low birth weight, cardiac defects, diaphragmatic hernias, and cryptorchidism. 61 17q12 is more commonly involved in deletion syndromes, although duplication syndromes have also been described and are marked by kidney failure, abnormal behavioral development, and mental retardation. 62 10q26.13 is involved in the deletion syndrome distal monosomy 10q. 63 Clinical features of this syndrome include abnormal facial features, cardiac defects, and ambiguous genitalia, though cancer susceptibility has not yet been described.
The 8p11.2 locus has been recognized to have a broad importance in cancer, and thus our study focuses on four genes of interest positioned at this locus: TACC1, KAT6A TM2D2 and ADAM32 [3]. Deletions and alternative splicing of TACC1 have been associated with an increased risk of gastric cancer, ovarian cancer, breast cancer, and glioblastoma. 72,75,76 In particular, a TACC1-FGFR fusion protein has been associated with the pathogenesis of glioblastoma multiforme in a subset of patients. 72 Abnormalities in KAT6A are associated with AML and breast cancer (105, 106). KAT6A-TIF2 and KAT6A-CBP fusions are cytogenetic markers in AML. 77 Analysis of a breast cancer cell line also showed elevated levels of KAT6A. 78 ADAM32 is a member of the ADAM family of genes that encodes cell surface proteins involved in inflammation, tumor formation, fertilization, and brain development. 64,79 ADAM32 encodes a protein possessing three functional domains: a disintegrin domain, a metalloproteinase domain, and an epidermal growth factor-like domain. The best-known role of ADAM32 is in the sperm-egg interaction and sperm development, where it is expressed during meiotic prophase in pachytene spermatocytes. 80 While there have been limited studies on the involvement of ADAM32 in cancer, three prior studies, including a published pediatric pan-cancer genome analysis, have yielded significant data supporting its role in malignancy. One study detects the presence of ADAM32 fusions in ERBB2-amplified breast cancer. 65 A second showed that ADAM32 was upregulated in hepatocellular cancer. 64 Genomic rearrangements in ADAM32 were also demonstrated in eight different tumor types including breast, colorectal, gastric, ovarian, prostrate, renal, and lung cancer, as well as pediatric medulloblastoma. 68 The 8p23.1 locus has a multitude of known oncogenes, such as those encoding the α-and β-defensin proteins, as well as zinc finger domain proteins and ubiquitin signal peptidase family genes. [81][82][83][84][85][86][87] The α-defensins are known to inhibit growth of CML and lung adenocarcinoma cells, though at lower levels they may increase growth of renal cell carcinoma. [81][82][83] The β-defensins suppress tumor invasion and migration as exemplified by beta defensin-1 (DEFB1) effects on oral squamous carcinoma. 82 SNPs in the β-defensin genes are associated with increased risk of developing glioblastoma, 84 while DEFB1 has been shown to have a protective role against cancer and is downregulated in HCV-infected hepatocellular cancer. 86 Several genes in the 8p23.1 locus encode for USP17-like proteins, which have been found to suppress tumorigenesis in breast cancer. 87 In our study, 8p23.1 showed deletions in ALL, Wilm's tumor, neuroblastoma, lymphoma and fibromatosis, which has not been reported previously.
Our study is limited by the fact that a DNA microarray can only detect CNVs and is not sensitive to translocations and inversions. Our study is also limited by sample size in the case of several types of cancer including hepatocellular cancer, medulloblastoma, rhabdomyosarcoma, renal cell carcinoma, ovarian tumor, melanoma and peripheral neural epithelioma, in which fewer than 3 cases each were recorded. Finally, it remains possible that many of the chromosomal aberrations described are secondary events without a driver function in cancer, though further evidence is needed to comprehensively elucidate the significance of these rearrangements.