Genome-wide screening methods in tumors of the central nervous system and cancer predisposition

Abstract

Since the development of screening methods that can be used on an exome- or genome-wide scale such as array-based comparative genomic hybridization (array-CGH) and next generation sequencing (NGS), these techniques have been employed to analyze large patient and tumor cohorts and are also frequently used for diagnostic purposes. In this study, array-CGH was used to analyze DNA from tumors of the central nervous system to identify somatic copy number changes. Approximately 300 tumor samples from the German Glioma Network and an intracranial malignant peripheral nerve sheath tumor were analyzed using array-CGH to identify specific patterns of copy number alterations in the different tumor entities. Furthermore, array-CGH and whole exome sequencing (WES) were performed on DNA from peripheral blood from a patient presenting with a complex phenotype including cancer predisposition to identify causative germline aberrations. <br /> The first project addressed molecular aberrations in gliomas classified as grade II and grade III by the World Health Organization (WHO) including astrocytomas, oligoastrocytomas and oligodendrogliomas as well as anaplastic astrocytomas and anaplastic oligoastrocytomas. Tumor samples of the different glioma entities were analyzed using array-CGH, in order to detect common genetic imbalances in the gliomas of WHO grade II and III. Together with the German Glioma Network the mutation status in the IDH1 (isocitrate-dehydrogenase 1) and IDH2 genes was determined. Most of the WHO grade II and grade III gliomas harbored an IDH1 or IDH2 mutation. It could be shown that WHO grade II gliomas displayed DNA copy number changes less frequently than WHO grade III gliomas. Interestingly, a small group of IDH1/2 wild-type WHO grade II astrocytomas were detected which displayed glioblastoma-like genomic imbalances. Patients with an IDH1/2 wild-type astrocytoma of WHO grade II displaying a glioblastoma-like genomic profile, i.e. gains on chromosomes 7, 19 and 20 as well as losses of chromosomes 9 and 10, possibly would benefit from a more intensive therapy strategy. Therefore these analyses have future implications. Furthermore, frequent deletions of chromosomal arms 1p and 19q were found in oligodendroglial tumors or in mixed astrocytic tumors displaying also an oligodendroglial component. Both alterations were significantly less frequent in astrocytic tumors. <br /> The molecular analysis of primary WHO grade IV glioblastomas was subdivided into two parts. As the prognosis of primary glioblastoma is still poor, long-term survival of more than three years after diagnosis is rare in these patients. Thus, the first part of the analysis (project 2 of this work) focused on tumors from patients who exhibited long-term survival. Genomic profiles of glioblastomas from long-term survivors were compared to those from short-term and intermediate-term survivors. The IDH1/2 mutation and the MGMT promoter methylation status were also determined in these tumors. This analysis showed that patients with long-term survival were younger and corresponding tumors more often had an IDH1/2 mutation and a MGMT promoter methylation. Genomic imbalances were prominently different between IDH1/2 mutant and IDH1/2 wild-type tumors, but not between survival groups of IDH1/2 wild-type glioblastoma patients, suggesting that long-term survival is due to other, e.g. host-related factors. <br /> The second part of the analysis (project 3 of this work) focused on tumor recurrence of primary WHO grade IV glioblastomas. Glioblastomas have a tendency to recur despite combined surgical resection, radiotherapy and temozolomide chemotherapy. When the tumor recurs the WHO grade remains the same, therefore, the recurrent tumor is treated similar to the primary tumor. Genomic profiles of 27 primary and recurrent IDH1/2 wild-type glioblastoma from the same patient were compared to determine genetic patterns of glioblastoma progression. After comparing the array-CGH profiles of the primary and recurrent tumors, taking the tumor cell content into account, a difference profile for each tumor pair was generated. Subsequently, three molecular relapse groups were defined (Equal, Sequential and Discrepant). Seven of the 27 (26%) tumor pairs were identified to be Equal pairs, showing no DNA copy number differences between primary and recurrent tumor, suggesting a monoclonal cell composition of both tumors. In nine of 27 (33%) tumor pairs, the same and additional chromosomal imbalances were found in the recurrent tumor as compared to the primary tumor (Sequential pairs). These findings suggest a sequential acquisition or selection for aberrations during tumor progression. In eleven of 27 (41%) pairs, the difference profiles of primary and recurrent tumors were divergent, i.e. the recurrent tumors contained additional chromosomal aberrations but had also lost others (Discrepant pairs). These findings suggest a polyclonal composition of the primary tumors and considerable clonal evolution. Interestingly, losses on chromosomal band 9p21.3, harboring the CDKN2A/B locus, were significantly more common in primary tumors of the Sequential and Discrepant tumor pairs, also called non-Equal pairs. Analyzing regions of chromosomal differences between primary and recurrent tumors 46 candidate genes associated with tumor recurrence were identified. Frequently, the identified genes for apoptosis regulators, possibly explaining why these cells escape therapy induced apoptosis. Taken together about 75% of IDH1/2 wild-type recurrent glioblastomas acquire additional genomic alterations during progression. This process is possibly facilitated by the loss of genetic material from chromosomal band 9p21.3 in the primary glioblastomas. These tumor recurrence-associated chromosomal changes may contribute to therapy resistance, e.g. by copy number alterations of apoptosis regulatory genes. The analysis of the genomic differences between primary and recurrent glioblastomas may identify newly acquired genetic properties targetable by salvage therapies for a more effective treatment of patients with recurrent glioblastoma. <br /> In the fourth project of this work, two intracranial tumor samples from a 47-year old female patient were retrospectively analyzed using array-CGH in order to determine their clonal relationship. The histological diagnosis of the first tumor was an unusual pituitary adenoma, but the second tumor that had developed 3 months later was diagnosed as a rare malignant peripheral nerve sheath tumor. Though the two tumor samples were different in their histopathology, the question arose if the first and the second tumor contained the same copy number changes and thus most likely developed from the same origin. Array-CGH of the first tumor revealed a complex pattern of chromosomal imbalances affecting all chromosomes but one (chromosome 16). Array-CGH of the second tumor revealed a similarly complex profile. About 80% of the 29 copy number changes detected in the second tumor were already present in the first tumor. These findings provide strong evidence for a clonal relationship between the two tumor samples and suggest that the second tumor was a recurrent tumor of the first lesion. It also could be shown that the genomic profiles of both tumors were highly similar to those of already published MPNST cases, indicating that the analyzed tumor indeed is a MPNST. Taken together, it could be shown that array-CGH can be successfully used to identify the clonal relationship between two histologically distinct tumors.<br /> The fifth project of this work aimed at identifying the germline aberrations underlying a complex phenotype including cancer predisposition. Symptoms of the patient included cognitive impairment, two neoplastic diseases (a both-sided mixed malignant germ cell tumor of the ovaries and an acute pre-B-lymphoblastic leukemia) prior to the age of 20 years, anomalies of skin pigmentation and short stature. Using array-CGH on DNA from peripheral blood from the patient and her mother, two maternally inherited microduplications in the chromosomal bands 6q27 and 22q11.21 were detected. The microduplication with the size of 0.26 Mb in 6q27 encompassed parts of the MLLT4 gene, a known fusion partner of MLL in leukemic cells. The microduplication in chromosomal band 22q11.21 had a size of 2.5 Mb, harboring approximately 50 genes. This region has been reported to be susceptible to chromosomal rearrangements and to cause the 22q11.21 microduplication syndrome when duplicated and is associated with a high variability, explaining in part the patient’s intellectual disability but not its severity. Particularly the two malignancies of the patient were not explained by the detected microduplications. Therefore, DNA from peripheral blood from the patient as well as from her patents was screened by WES in order to find further causative germline aberrations. A trio-based de novo analysis, subtracting the parental variants from variants detected in the patient, revealed a de novo CHEK2 variant (CHEK2,c.1427C>T;p.Thr476Met). This rare missense variant is a “HGMD disease mutation” contributing to breast cancer susceptibility. Using a different filter strategy for the WES data set of the patient, two known BLM founder mutations (BLM,c.1642C>T;p.Gln548X; BLM,c.2695C>T;p.Arg899X) were also detected. Sanger sequencing revealed that the patient was compound heterozygous for these two stop mutations in the BLM gene, i.e. that one of the mutations was inherited from the mother and the other from the father. Bloom syndrome, caused by mutations in the BLM gene is a rare autosomal recessive disorder associated with most of the symptoms found in the patient, including short stature, mild craniofacial dysmorphia, hypo- and hyper-pigmented skin lesions and cancer predisposition. Taken together, two genome-wide screening methods have been used to unravel the highly complex phenotype of the patient. Employing array-CGH and WES, two inherited microduplications as well as two inherited recessive founder mutations (BLM) and one de novo missense variant (CHEK2) were identified. The combination of Bloom syndrome with the 22q11.2 microduplication syndrome and the CHEK2 associated multi-cancer susceptibility syndrome, are presumed to cause the severe intellectual disability, and explain the cancer predisposition in the patient. This case demonstrated that complex phenotypes may be caused by more than one genetic alteration, rather a combination of copy number variants and point mutations may be the cause.<br /> In summary, array-CGH was used to detect somatic tumor aberrations in WHO grade II to WHO grade IV glioma entities as well as a case of a MPNST. Further, employing two genome-wide screening methods, array-CGH and WES, the complex genotype of a patient with syndromic cancer predisposition was unraveled, indicating that complex phenotypes may be caused by a number of different genetic alterations.