The association of constitutional chromosome imbalance in patients with intellectual disability with or without related dysmorphism and malformations is well established. The resolution of conventional cytogenetic examination is limited to imbalances of 5-10Mb. Patients with characteristic phenotypes which allude to a specific microdeletion or duplication syndrome may be investigated using locus specific fluorescent in situ hybridisation (FISH). Subtelomere FISH, a recently new improvement for cytogenetics screening, detects subtelomeric rearrangements in around 6% of patients with idiopathic disability. However it is evident that for these patients, most do not have a recurrent pattern of dysmorphism or malformations suggesting imbalance in a particular chromosome region. Array CGH has the potential to detect chromosome imbalances beyond that of current technology allowing the whole genome can be screened in a single hybridisation at a resolution limited only by the genomic distance between the arrayed target clones. The aim of this study was to develop a custom whole genome array and utilize this array to screen a number of diverse patient groups. Rather than immediately begin with the development of a whole genome array a smaller pilot study was initiated, in so enabling the efficacy of the methodology to be tested. A small clinical/ subtelomere array was designed and constructed to screen for cytogenetic imbalances within the first 5Mb of each chromosome end (excluding acrocentric chromosomes) together with the number of known clinically significant regions. This clinical/subtelomere array (chapter 3), was ulilised to map the extent of deletion and/or duplication in patients with previously determined subtelomere abnormalities. This was followed with the screening of a small group of patients with idiopathic intellectual disability (chapter 4). Novel Copy number changes were identified together with a number of changes determined to be non-pathogenic variants. The methodology used in the utilization of this array could determine copy changes in patients, however it became clearly evident that the effective resolution was compromised when a number of clones were shown to map to other sites of the genome or cross hybridise to multiple sites. With this knowledge the next stage of the project, a whole genome array, primarily constructed from a FISH validated clone set, reduced the possibility of mapping discrepancies. These FISH mapped clones did not give a even genomic coverage. All clones from this set were mapped using Ensembl resources. Any identified gaps (greater than 1Mb) were covered using clones from a second clone set (32K), giving a resolution of ~0.2-1Mb. Three diverse groups of patients were screened using the whole genome array. Using a novel scoring system which evaluated degree of intellectual disability/developmental delay, dysmorphism, presence of malformations and the presence of an apparently balanced cytogenetic rearrangement, twenty five patients with idiopathic disability were assessed. Four novel copy number changes were determined of which three were determined likely to contribute to the phenotype of the patient. The fourth patient is under investigation. Chapter 6 saw the application of the whole genome array to a second patient group; medically terminated pregnancies, in utero fetal deaths or newborn infants with one or more facial dysmorphism or fetal malformation. One potentially pathogenic copy number change was ascertained from this study. The third patient group, those with retractable epilepsy (chapter 7) showed two copy number changes both of which will require further follow up and investigation, including screening with a higher density array platform. The screening of more than seventy varied patients has shown the efficacy of these two custom platforms, to detect previously undetermined copy number changes. / Thesis (Ph.D.) -- University of Adelaide, School of Paediatrics and Reproductive Health, 2010

O osteossarcoma (OS) é o tumor ósseo maligno mais freqüente da infância e adolescência com uma taxa de sobrevida livre de eventos de 50 70% após 3 anos.. O pico de incidência ocorre na segunda década da vida, característica que sugere uma relação entre o rápido crescimento ósseo da adolescência e o desenvolvimento da neoplasia. Ainda, o conhecimento das bases genéticas é insuficiente. Estudos de citogenética clássica têm demonstrado que o OS caracteriza-se por exibir alta heterogeneidade cariotípica, incluindo altos graus de aneuploidia e rearranjos estruturais complexos. A técnica de CGH constitui uma ferramenta valiosa na analise do perfil genômico de tumores sólidos, e tem confirmado a complexidade das alterações cariotípicas em OS, descrita pela citogenética convencional. Não obstante, os estudos existentes são divergentes, e poucos têm estudado as informações obtidas por CGH em relação com a progressão tumoral. O objetivo do presente estudo foi identificar a presença de desequilíbrios cromossômicos em amostras de OS por meio da técnica de CGH. Os experimentos de CGH foram realizados de acordo com o descrito por Kallioniemi et al (1994). Foram analisadas nove amostras (3 biópsias, 5 ixressecções após quimioterapia e 1 metástase). O CGH detectou desequilíbrios cromossômicos em todas as amostras. Os ganhos foram mais freqüentes que as perdas. Muitas alterações cromossômicas foram observadas, especialmente ganhos nos cromossomos 1q, 2, 3p, 4, 5p, 6, 7, 8, 11p, 14q, 16, 21q e X; e perdas nos cromossomos 1p, 2q, 3q, 5q, 9q, 11q e 17q. As regiões mínimas de sobreposição mais freqüentes foram ganhos de 2p13-p14, 2q36-q37, 4q21 e 8p22, e perdas de 1p34.2, 3q22-q23 e 3q24. Três pacientes apresentaram amostras pareadas, e as alterações cromossômicas detectadas foram muito variadas, refletindo a heterogeneidade cromossômica intratumoral em cada caso. A mais alta divergência clonal entre as amostras pareadas foi observada entre a amostra de ressecção e a amostra metastática correspondente, mostrando a complexidade cromossômica adquirida durante a progressão e metastatização no caso descrito. Ainda são necessárias investigações adicionais, que contribuíam para a caracterização completa dos genes localizados nessas regiões. / Osteosarcoma (OS) is the most frequent aggressive bone malignancy affecting children and young adults with an event-free survival of 50-70% after 3 years. The incidence peak occurs during the second decade of life, suggesting a relationship between rapid bone growth and the development of this tumor. The knowledge of the genetic basis behind tumor progression is still limited. Conventional cytogenetic studies have demonstrated that OS exhibits high cariotipic heterogeneity, with different degrees of aneuploidy and complex structural rearrangements. The CGH is an important tool for studying the genomic profiles of solid tumors, and has confirmed the complexity of cariotipic alterations in OS. However, previous studies have shown divergent results and few have correlated them with tumor progression. The objective of present study was to identify chromosome unbalances in nine samples of OS by CGH. 3 biopsies, 5 resections before quimioterapy and 1 metastasis were analyzed. The experiments were performed accordingly with Kallioniemi et al (1994). CGH detected chromosomal imbalances in all samples. Gains were more frequent than losses. Many chromosomal alterations were observed, especially gains at 1q, xi2, 3p, 4, 5p, 6, 7, 8, 11p, 14q, 16, 21q and X; and losses at 1p, 2q, 3q, 5q, 9q, 11q and 17q. The minimal regions of superposition were gains of 2p13-p14, 2q36-q37, 4q21 and 8p22, and losses of 1p43.2, 3q22-q23 and 3q24. Three patients had consecutive samples, and the chromosomal alterations varied, reflecting the chromosomal heterogeneity for each case. The highest clonal divergence among the consecutive samples was observed between resection and the corresponding metastatic sample, showing the chromosomal complexity acquired during the progression and dissemination in this case. Additional investigations for the characterization of genes at these regions are necessary.

Objetivos: Analisar por meio da técnica de sequenciamento direto a presença de alterações moleculares nos genes SHH, SIX3, ZIC2 e TGIF1 em indivíduos com diagnóstico clínico de HPE. Analisar por meio da técnica de CGH-array a presença de alterações moleculares em indivíduos com diagnóstico clínico de HPE previamente submetidos à análise por sequenciamento direto. Local: Laboratório de Genética e Citogenética Humana HRAC/USP, Bauru-SP. Casuística e metodologia: Foram selecionados 50 indivíduos, de ambos os sexos com idades entre 03 meses a 50 anos com diagnóstico clínico para HPE. Todos foram analisados por meio da técnica de sequenciamento direto para os genes SHH e TGIF1 completamente e para os genes ZIC2 e SIX3 parcialmente. Dentre os indivíduos que não apresentaram alterações na técnica de sequenciamento oito indivíduos com fenótipo mais grave foram selecionados para a análise por CGH-array. Resultados e discussão: Foram analisados 50 indivíduos por meio da técnica de sequenciamento direto dos gene SHH e TGIF1, foram encontradas duas variantes patogênicas na análise do gene SHH, no caso 1 a variante p.24G>P foi identificada, e no caso 2 foi identificada a variante c.1031del C. No gene TGIF1 foram encontrados cinco polimorfismos já descritos na literatura. Foi identificada uma nova variante silenciosa no éxon 1 do gene ZIC2 p.Q46Q (c. 431 G>A) e um polimorfismo já descrito na literatura em dois indivíduos no gene SIX3. A análise por CGH-array revelou a presença de uma microdeleção no caso 37, de 1,5Mb no cromossomo 17p12 entre as posições genômicas 14,052,279-15,102,307. A mesma deleção foi encontrada na mãe, sendo que esta região nunca foi associada a HPE. Conclusão: A técnica de sequenciamento direto é uma ferramenta muito importante no diagnóstico molecular da HPE, a padronização do sequenciamento direto para os genes ZIC2 e SIX3 poderá auxiliar em diagnósticos mais precisos em estudos futuros dentro do HRAC/USP. O emprego de novas técnicas como CGH-array pode indicar novas relações entre regiões cromossômicas e os múltiplos fatores envolvidos na formação da HPE. / Objective: Analyze through direct sequencing technique the presence of molecular changes on the genes SHH, SIX3, ZIC2 and TGIF1 on individuals with clinical diagnosis of HPE. Analyze through array-CGH technique the presence of molecular changes on individuals with clinical diagnosis of HPE previously submitted to the direct sequencing analyzes. Local: Genetics and Human Cytogenetics Laboratory, HRAC/USP, Bauru-SP. Methods: Were selected 50 individuals from both genders with ages between 03 months and 50 years clinically diagnosed with HPE. Everyone was analyzed through the direct sequencing technique for the genes SHH and TGIF1 completely and for the genes ZIC2 and SIX3 partially. From those individuals which did not have shown changes on the direct sequencing technique, eight individuals with more severe phenotype were selected to the analysis through array-CGH. Results an Discussion: Were analyzed 50 individuals through the technique of direct sequencing of the genes SHH and TGIF1, were found two pathogenic variants in the analysis of SHH gene, in the case 1, the variant p.G24P was identified, and in the case 2 was identified the variant c.1031delC. On the TGIF1 gene were found five polymorphisms already described on the literature. Was identified a new silent variant on the exon 1 of the ZIC2 gene p. Q46Q(c.431G>A) and a polymorphism already described in the literature in two individuals on the gene SIX3. The analysis through array-CGH revealed the presence of one microdeletion in the case 37, of 1,5 Mb on the region 17p12 between the genomic positions 14,052,279-15,102,307. The same deletion was detected in the mother, though this region was never associated to the HPE. Conclusion: The direct sequencing technique is a very important tool for the molecular diagnosis of the HPE, and the direct sequencing standardization for the genes ZIC2 and SIX3 might help in more precise diagnostics on HRAC/USP future studies. The employ of new techniques such as array-CGH may indicate new relations between chromosomal regions and the multiple hit involved in the development of HPE.

Objetivos: Analisar por meio da técnica de sequenciamento direto a presença de alterações moleculares nos genes SHH, SIX3, ZIC2 e TGIF1 em indivíduos com diagnóstico clínico de HPE. Analisar por meio da técnica de CGH-array a presença de alterações moleculares em indivíduos com diagnóstico clínico de HPE previamente submetidos à análise por sequenciamento direto. Local: Laboratório de Genética e Citogenética Humana HRAC/USP, Bauru-SP. Casuística e metodologia: Foram selecionados 50 indivíduos, de ambos os sexos com idades entre 03 meses a 50 anos com diagnóstico clínico para HPE. Todos foram analisados por meio da técnica de sequenciamento direto para os genes SHH e TGIF1 completamente e para os genes ZIC2 e SIX3 parcialmente. Dentre os indivíduos que não apresentaram alterações na técnica de sequenciamento oito indivíduos com fenótipo mais grave foram selecionados para a análise por CGH-array. Resultados e discussão: Foram analisados 50 indivíduos por meio da técnica de sequenciamento direto dos gene SHH e TGIF1, foram encontradas duas variantes patogênicas na análise do gene SHH, no caso 1 a variante p.24G>P foi identificada, e no caso 2 foi identificada a variante c.1031del C. No gene TGIF1 foram encontrados cinco polimorfismos já descritos na literatura. Foi identificada uma nova variante silenciosa no éxon 1 do gene ZIC2 p.Q46Q (c. 431 G>A) e um polimorfismo já descrito na literatura em dois indivíduos no gene SIX3. A análise por CGH-array revelou a presença de uma microdeleção no caso 37, de 1,5Mb no cromossomo 17p12 entre as posições genômicas 14,052,279-15,102,307. A mesma deleção foi encontrada na mãe, sendo que esta região nunca foi associada a HPE. Conclusão: A técnica de sequenciamento direto é uma ferramenta muito importante no diagnóstico molecular da HPE, a padronização do sequenciamento direto para os genes ZIC2 e SIX3 poderá auxiliar em diagnósticos mais precisos em estudos futuros dentro do HRAC/USP. O emprego de novas técnicas como CGH-array pode indicar novas relações entre regiões cromossômicas e os múltiplos fatores envolvidos na formação da HPE. / Objective: Analyze through direct sequencing technique the presence of molecular changes on the genes SHH, SIX3, ZIC2 and TGIF1 on individuals with clinical diagnosis of HPE. Analyze through array-CGH technique the presence of molecular changes on individuals with clinical diagnosis of HPE previously submitted to the direct sequencing analyzes. Local: Genetics and Human Cytogenetics Laboratory, HRAC/USP, Bauru-SP. Methods: Were selected 50 individuals from both genders with ages between 03 months and 50 years clinically diagnosed with HPE. Everyone was analyzed through the direct sequencing technique for the genes SHH and TGIF1 completely and for the genes ZIC2 and SIX3 partially. From those individuals which did not have shown changes on the direct sequencing technique, eight individuals with more severe phenotype were selected to the analysis through array-CGH. Results an Discussion: Were analyzed 50 individuals through the technique of direct sequencing of the genes SHH and TGIF1, were found two pathogenic variants in the analysis of SHH gene, in the case 1, the variant p.G24P was identified, and in the case 2 was identified the variant c.1031delC. On the TGIF1 gene were found five polymorphisms already described on the literature. Was identified a new silent variant on the exon 1 of the ZIC2 gene p. Q46Q(c.431G>A) and a polymorphism already described in the literature in two individuals on the gene SIX3. The analysis through array-CGH revealed the presence of one microdeletion in the case 37, of 1,5 Mb on the region 17p12 between the genomic positions 14,052,279-15,102,307. The same deletion was detected in the mother, though this region was never associated to the HPE. Conclusion: The direct sequencing technique is a very important tool for the molecular diagnosis of the HPE, and the direct sequencing standardization for the genes ZIC2 and SIX3 might help in more precise diagnostics on HRAC/USP future studies. The employ of new techniques such as array-CGH may indicate new relations between chromosomal regions and the multiple hit involved in the development of HPE.

O osteossarcoma (OS) é o tumor ósseo maligno mais freqüente da infância e adolescência com uma taxa de sobrevida livre de eventos de 50 70% após 3 anos.. O pico de incidência ocorre na segunda década da vida, característica que sugere uma relação entre o rápido crescimento ósseo da adolescência e o desenvolvimento da neoplasia. Ainda, o conhecimento das bases genéticas é insuficiente. Estudos de citogenética clássica têm demonstrado que o OS caracteriza-se por exibir alta heterogeneidade cariotípica, incluindo altos graus de aneuploidia e rearranjos estruturais complexos. A técnica de CGH constitui uma ferramenta valiosa na analise do perfil genômico de tumores sólidos, e tem confirmado a complexidade das alterações cariotípicas em OS, descrita pela citogenética convencional. Não obstante, os estudos existentes são divergentes, e poucos têm estudado as informações obtidas por CGH em relação com a progressão tumoral. O objetivo do presente estudo foi identificar a presença de desequilíbrios cromossômicos em amostras de OS por meio da técnica de CGH. Os experimentos de CGH foram realizados de acordo com o descrito por Kallioniemi et al (1994). Foram analisadas nove amostras (3 biópsias, 5 ixressecções após quimioterapia e 1 metástase). O CGH detectou desequilíbrios cromossômicos em todas as amostras. Os ganhos foram mais freqüentes que as perdas. Muitas alterações cromossômicas foram observadas, especialmente ganhos nos cromossomos 1q, 2, 3p, 4, 5p, 6, 7, 8, 11p, 14q, 16, 21q e X; e perdas nos cromossomos 1p, 2q, 3q, 5q, 9q, 11q e 17q. As regiões mínimas de sobreposição mais freqüentes foram ganhos de 2p13-p14, 2q36-q37, 4q21 e 8p22, e perdas de 1p34.2, 3q22-q23 e 3q24. Três pacientes apresentaram amostras pareadas, e as alterações cromossômicas detectadas foram muito variadas, refletindo a heterogeneidade cromossômica intratumoral em cada caso. A mais alta divergência clonal entre as amostras pareadas foi observada entre a amostra de ressecção e a amostra metastática correspondente, mostrando a complexidade cromossômica adquirida durante a progressão e metastatização no caso descrito. Ainda são necessárias investigações adicionais, que contribuíam para a caracterização completa dos genes localizados nessas regiões. / Osteosarcoma (OS) is the most frequent aggressive bone malignancy affecting children and young adults with an event-free survival of 50-70% after 3 years. The incidence peak occurs during the second decade of life, suggesting a relationship between rapid bone growth and the development of this tumor. The knowledge of the genetic basis behind tumor progression is still limited. Conventional cytogenetic studies have demonstrated that OS exhibits high cariotipic heterogeneity, with different degrees of aneuploidy and complex structural rearrangements. The CGH is an important tool for studying the genomic profiles of solid tumors, and has confirmed the complexity of cariotipic alterations in OS. However, previous studies have shown divergent results and few have correlated them with tumor progression. The objective of present study was to identify chromosome unbalances in nine samples of OS by CGH. 3 biopsies, 5 resections before quimioterapy and 1 metastasis were analyzed. The experiments were performed accordingly with Kallioniemi et al (1994). CGH detected chromosomal imbalances in all samples. Gains were more frequent than losses. Many chromosomal alterations were observed, especially gains at 1q, xi2, 3p, 4, 5p, 6, 7, 8, 11p, 14q, 16, 21q and X; and losses at 1p, 2q, 3q, 5q, 9q, 11q and 17q. The minimal regions of superposition were gains of 2p13-p14, 2q36-q37, 4q21 and 8p22, and losses of 1p43.2, 3q22-q23 and 3q24. Three patients had consecutive samples, and the chromosomal alterations varied, reflecting the chromosomal heterogeneity for each case. The highest clonal divergence among the consecutive samples was observed between resection and the corresponding metastatic sample, showing the chromosomal complexity acquired during the progression and dissemination in this case. Additional investigations for the characterization of genes at these regions are necessary.

No / In the past two decades comparative genomic hybridization (CGH) and array CGH have become crucial and indispensable tools in clinical diagnostics. Initially developed for the genome-wide screening of chromosomal imbalances in tumor cells, CGH as well as array CGH have also been employed in genotoxicology and most recently in toxicogenomics. The latter methodology allows a multi-endpoint analysis of how genes and proteins react to toxic agents revealing molecular mechanisms of toxicology. This chapter provides a background on the use of CGH and array CGH in the context of genotoxicology as well as a protocol for conventional CGH to understand the basic principles of CGH. Array CGH is still cost intensive and requires suitable analytical algorithms but might become the dominating assay in the future when more companies provide a large variety of different commercial DNA arrays/chips leading to lower costs for array CGH equipment as well as consumables such as DNA chips. As the amount of data generated with microarrays exponentially grows, the demand for powerful adaptive algorithms for analysis, competent databases, as well as a sound regulatory framework will also increase. Nevertheless, chromosomal and array CGH are being demonstrated to be effective tools for investigating copy number changes/variations in the whole genome, DNA expression patterns, as well as loss of heterozygosity after a genotoxic impact. This will lead to new insights into affected genes and the underlying structures of regulatory and signaling pathways in genotoxicology and could conclusively identify yet unknown harmful toxicants.

This study evaluated comparative genomic hybridisation (CGH) as a tool to detect candidate regions of the genome associated with chemoresistance. Using a variation on conventional CGH, DNA from three cell lines that were resistant to thymidylate synthase (TS) inhibitors (tomudex [TDX] or 5-fluorouracil [5-FU]) and their sensitive parent cells were evaluated. In MCF-7 and H630, cells that were resistant to TDX, a specific TS inhibitor with no other known cytotoxic potential, only a single region of change (18p gain) was apparent. The third cell line H630R10, which was resistant to 5-FU, had changes in several genomic regions following the acquisition of resistance, including 18p. Gain in the chromosomal region containing the TS gene (18pll.32) was detected by CGH in all three resistant cell lines. However, additional novel regions of interest were identified in the cells that were resistant to 5-FU, a cytotoxic agent known to have several other modes of cytotoxicity besides TS inhibition. These results suggested that CGH is of potential use in the detection of regions of the genome involved in chemoresistance. Having shown the potential of CGH as a tool for assessing chemoresistance at the genomic level, steps toward clinical application of this technique were evaluated. A prerequisite for study in archival pathology samples was successful DNA extraction and universal amplification of tumour DNA from paraffin-embedded tumour sections for CGH analysis. Degenerate oligonucleotide primed - polymerase chain reaction (DOP-PCR) was performed on minute quantities (50ngs) of fresh cell line DNA (H630R10) and tumour DNA (osteosarcoma), as well as paraffin-embedded DNA from the same case. The results of these DOP-PCR CGH reactions were compared with conventional CGH using l|0.g quantities of fresh DNA from both H630R10 cell line and osteosarcoma. The CGH profiles of the conventional CGH and DOP-PCR CGH did not show a high level of concordance, only 55% of the gains and 83.3% of losses detected by conventional CGH were detected by DOP-PCR CGH The use of universal amplification by DOP-PCR in paraffin-embedded sections was not taken forward into clinical evaluation. A study of colorectal cancer (CRC) was initiated which involved the microdissection of 29 Dukes' C CRC tumours from fresh frozen material for CGH analysis. This conventional CGH analysis of CRC tumours involved assessing each tumour twice by reversal of fluorochromes. Only genomic regions that were detected as changed in both forward and reverse profiles were accepted. This approach detected several regions of genome as changed across the 29 tumours. In all, 108 gains (a mean number of 3.7 aberrations per tumour, range 1-12) and 85 losses (a mean number of 2.9 aberrations per tumour, range 0-11) were detected in the 29 tumours. CGH analyses identified certain chromosomal regions as more likely to be changed than others. The most frequent aberrations detected across the 29 tumours was a loss of chromosomal arm 18q, seen in 31% of the tumours assessed. Gain was also common at some sites throughout the genome, for example, gain of chromosomal arms, 13q and 20q was seen in 27.6% of cases. Mann-Whitney U tests investigating the association between specific chromosomal aberrations such as gain of 20q or loss of 18q and known markers of CRC tumourigenesis (p53, p27, p21, Rb, cyclin Dl, PCNA, P-catenin, e-cadherin, c-erbB-2, bcl2, EGFR and c-erbB-2) assessed by immunohistochemistry (IHC) in 29 tumours found no association. Testing of the total number of genomic aberrations detected (loss + gain = genetic grade) rather than the frequency of aberration at specific chromosomal loci also found no association with the CRC tumour markers. Finally, the association between the chromosomal aberrations detected by CGH was investigated in relation to patient survival. This thesis has demonstrated the value of a global approach to the study of chemoresistance and tumourigenesis through the application of powerful technology such as CGH.

DNA copy number alterations (CNAs) are genetic changes that can produce adverse effects in numerous human diseases, including cancer. CNAs are segments of DNA that have been deleted or amplified and can range in size from one kilobases to whole chromosome arms. Development of array comparative genomic hybridization (aCGH) technology enables CNAs to be measured at sub-megabase resolution using tens of thousands of probes. However, aCGH data are noisy and result in continuous valued measurements of the discrete CNAs. Consequently, the data must be processed through algorithmic and statistical techniques in order to derive meaningful biological insights. We introduce model-based approaches to analysis of aCGH data and develop state-of-the-art solutions to three distinct analytical problems. In the simplest scenario, the task is to infer CNAs from a single aCGH experiment. We apply a hidden Markov model (HMM) to accurately identify CNAs from aCGH data. We show that borrowing statistical strength across chromosomes and explicitly modeling outliers in the data, improves on baseline models. In the second scenario, we wish to identify recurrent CNAs in a set of aCGH data derived from a patient cohort. These are locations in the genome altered in many patients, providing evidence for CNAs that may be playing important molecular roles in the disease. We develop a novel hierarchical HMM profiling method that explicitly models both statistical and biological noise in the data and is capable of producing a representative profile for a set of aCGH experiments. We demonstrate that our method is more accurate than simpler baselines on synthetic data, and show our model produces output that is more interpretable than other methods. Finally, we develop a model based clustering framework to stratify a patient cohort, expected to be composed of a fixed set of molecular subtypes. We introduce a model that jointly infers CNAs, assigns patients to subgroups and infers the profiles that represent each subgroup. We show our model to be more accurate on synthetic data, and show in two patient cohorts how the model discovers putative novel subtypes and clinically relevant subgroups.

Durant molts anys la citogenètica ha estat l'únic mètode disponible per a analitzar els canvis genètics en els tumors. Malgrat això, els estudis cariotípics estan sovint obstaculitzats per problemes tècnics i metodològics. L'any 1992 es va descriure una nova estratègia, que es coneix com hibridació genòmica comparada (CGH), i que identifica en un sol experiment aquelles regions que s'hagin amplificat (indicadores d'oncogens), així com regions delecionades (indicadores de gens supressors de tumors). En aquesta tesi s'han estudiat 37 mostres de tumors de la família Ewing (ET), vuit xenografts de tumors pancreàtics i set metàstasis de tumors pancreàtics originades en els ratolins. En totes les mostres s'ha aplicat la CGH. En els ET s'han detectat guanys recurrents del braç llarg del cromosoma 1 (mínima regió comuna 1q21-q22), i guanys dels cromosomes 8 i 12, així com guany de 7q i 6p2,1-pter i pèrdua de 16q. Aquestes regions podrien contenir gens importants en el desenvolupament i/o progressió dels ET. Mitjançant Southern blot s'ha detectat l'amplificació de dos gens, FLG i SPRR 3, localitzats a 1q21. A més, mitjançant una anàlisi estadística s'ha observat que els guanys de 6p estaven associats a un pitjor pronòstic. També s'ha desenvolupat una nova estratègia, la CGH amb mescles de DNAs de diferents mostres del mateix tipus tumoral. Aquesta tècnica permet detectar en un únic experiment les alteracions presents a la majoria de les mostres. En el present estudi s'ha demostrat l'eficàcia d'aquest mètode amb nombroses mostres (14-28) de sis tipus tumorals diferents, que prèviament s'havien estudiat individualment. En els tumors pancreàtics s'han detectat guanys en els cromosomes 8 (8q24),15 (15q25-q26),16, 20q i 19q, i pèrdues en els cromosomes 18 (18q21), 6 (6q21 i 6q24-qter), 13 (13q21) i 10 (10q 14-pter), en ordre decreixent de freqüència. Els estudis de pèrdues al·lèliques a 10p14-pter no han permès una delimitació més exacta de la regió afectada, ja que tots els casos amb pèrdua per CGH han mostrat pèrdues al·lèliques per a tots els loci informatius. Les regions cromosòmiques 8q24 i 15q25-qter s'han estudiat amb més detall per hibridació in situ fluorescent i Southern blot. L'estudi dels oncogens MYC a 8q24 i FES i IGF1R a 15q25-qter ha mostrat, en general, un baix nivell d'amplificació. Podria ésser que solament unes poques còpies extra d'aquests gens fossin suficients per a tenir efectes en la tumorogènesi pancreàtica o bé que fossin altres gens localitzats en aquestes regions els que realment estarien amplificats i els quals jugarien un paper important en la carcinogènesi pancreàtica. D'altra banda, s'ha observat una forta relació clonal entre les metàstasis estudiades i els tumors originaris implantats en els ratolins. Totes les metàstasis tenien les mateixes alteracions que els implants i, a més, noves alteracions que podrien contenir gens relacionats amb la progressió metastàtica del carcinoma de pàncreas. La majoria d'aquestes alteracions addicionals ja estaven presents en els subclons de l'implant, però no en proporció suficient com per a ser detectades per CGH.

DNA copy number alterations (CNAs) are genetic changes that can produce adverse effects in numerous human diseases, including cancer. CNAs are segments of DNA that have been deleted or amplified and can range in size from one kilobases to whole chromosome arms. Development of array comparative genomic hybridization (aCGH) technology enables CNAs to be measured at sub-megabase resolution using tens of thousands of probes. However, aCGH data are noisy and result in continuous valued measurements of the discrete CNAs. Consequently, the data must be processed through algorithmic and statistical techniques in order to derive meaningful biological insights. We introduce model-based approaches to analysis of aCGH data and develop state-of-the-art solutions to three distinct analytical problems. In the simplest scenario, the task is to infer CNAs from a single aCGH experiment. We apply a hidden Markov model (HMM) to accurately identify CNAs from aCGH data. We show that borrowing statistical strength across chromosomes and explicitly modeling outliers in the data, improves on baseline models. In the second scenario, we wish to identify recurrent CNAs in a set of aCGH data derived from a patient cohort. These are locations in the genome altered in many patients, providing evidence for CNAs that may be playing important molecular roles in the disease. We develop a novel hierarchical HMM profiling method that explicitly models both statistical and biological noise in the data and is capable of producing a representative profile for a set of aCGH experiments. We demonstrate that our method is more accurate than simpler baselines on synthetic data, and show our model produces output that is more interpretable than other methods. Finally, we develop a model based clustering framework to stratify a patient cohort, expected to be composed of a fixed set of molecular subtypes. We introduce a model that jointly infers CNAs, assigns patients to subgroups and infers the profiles that represent each subgroup. We show our model to be more accurate on synthetic data, and show in two patient cohorts how the model discovers putative novel subtypes and clinically relevant subgroups.