Le but de ce travail était de rechercher le rôle du microenvironnement dans la progression tumorale du cholangiocarcinome intrahépatique (CCIH) par une approche translationnelle, associant recherche fondamentale et clinique. Une étude transcriptomique du stroma tumoral a permis de mettre en évidence une signature spécifique de celui-ci, dont l’analyse non supervisée montrait un enrichissement dans les gènes de la matrice extracellulaire, du cycle cellulaire, de la voie TGFβ et des marqueurs de cellules souches. Ces résultats ont été validés au niveau protéique par immunohistochimie sur tissue microarrays à partir d’une cohorte indépendante. La corrélation de ces résultats avec les données cliniques a permis de démontrer que le niveau d’expression de l’Osteopontin dans le stroma était un facteur de risque indépendant de récidive et de survie. Par ailleurs, nous avons démontré que le taux sérique d’Osteopontin préopératoire des patients porteurs d’un CCIH était significativement supérieur à celui de sujets sains. Avec un seuil déterminé à 57,8 ng/ml, la sensibilité et spécificité de ce biomarqueur diagnostique était respectivement de 80 et 100%. De plus, nous avons apporté des arguments supplémentaires concernant le rôle des cellules souches cancéreuses dans la progression du CCIH, en mettant en évidence une corrélation entre le niveau d’expression de marqueurs souches tels qu’EpCAM et CD44 dans le stroma tumoral ainsi que dans le tissu fibreux du foie « sain » péri-lésionnel et le risque de récidive. Les résultats de notre étude ont confirmé le rôle central du microenvironnement dans la progression du CCIH, permis de mettre en évidence 2 nouveaux biomarqueurs pronostiques, et ouvert de nouvelles voies de recherche thérapeutiques. / The aim of this study was to specifically determine through a translational approach combining basic and clinical research, the role of the microenvironment in the tumor progression of intrahepatic cholangiocarcinoma (ICC). By gene expression profiling, we identified a signature that significantly discriminate the tumor stroma from non-tumor fibrous tissue, and the functional analysis of differentially expressed genes showed an enrichment in genes of the extracellular matrix , the cell cycle, the TGFb pathway and stem cell markers. Tissue microarray analysis using an independent cohort of ICC patients validated at a protein level the increased expression of selected candidate genes. Statistical analysis between basic and clinical data demonstrated that the stromal expression of Osteopontin was an independent prognostic marker for overall and disease-free survival. We also demonstrated that the preoperative serum level of Osteopontin was significantly higher in ICC patients than in healthy subjects. Our results identified the best diagnostic threshold to 57,8 ng/ml, associated with a sensitivity and specificity reaching to 80 and 100%, respectively. Moreover, we showed that level expression of stem cell markers such as EpCAM and CD44 in tumor stroma as well as in the fibrous non tumor liver tissue was correlated with recurrence, suggesting the pivotal role of cancer stem cells in ICC prognosis. In conclusion, our study confirmed the major involvement of the microenvironment in the progression of CCIH, allowed to identify two new prognostic tumor biomarkers, and highlighted new pathways for targeted therapeutics.

Le myélome multiple (MM) est une néoplasie B caractérisée par l'accumulation d'un clone plasmocytaire dans la moelle osseuse. Cette pathologie demeure incurable d'où la nécessité d'identifier de nouvelles cibles thérapeutiques. L'utilisation des puces à ADN a permis d'identifier, de nombreux gènes dont l'expression par les cellules de MM est associée à un mauvais ou bon pronostic. La plupart des gènes pronostics identifiés dans le MM codent pour des protéines impliquées dans les processus de réplication, de réparation et de recombinaison de l'ADN. Nous avons voulu aller plus loin dans l'identification et la fonction biologique de ces gènes pronostics. D'une part, nous avons recherché les gènes présentant des pics d'expression très élevés ‘gènes spikés' chez une fraction des patients. Ces gènes sont généralement associés à des évènements oncogéniques. D'autre part, nous avons identifié des gènes pronostics non associés à la machinerie du cycle cellulaire et qui sont fortement exprimés dans des cellules souches pluripotentes ou adultes. L'identification de ces gènes nous a permis de construire un score pronostic très puissant, éliminant les scores pronostics publiés à présent. Un autre aspect majeur est l'élucidation des mécanismes biologiques impliquant ces gènes et qui sont responsables de la résistance aux traitements et/ou de la courte durée de survie des patients, afin de pouvoir les reverser. Nous avons donc évalué le rôle biologique de DEPDC1A un gène fortement exprimée dans les cellules de MM en association avec un mauvais pronostic. Le knockdown conditionnel de DEPDC1A par l'utilisation d'un shRNA, inhibe la croissance des lignées de myélome avec une accumulation des cellules en phase G2/M du cycle cellulaire. Cette accumulation est associée à la phosphosphorylation et à la stabilisation de P53, et à l'accumulation de P21/WAF1. Le knockdown de DEPDC1A résulte également en l'expression de marqueurs de cellules plasmocytaires matures dans les lignées de MM : CD31, CD38, CD138, IL6R, CXCR4, CD9, VLA6. DEPDC1A contrôle donc le cycle cellulaire des plasmocytes tumoraux en interférant avec la voie P53 et bloque leur différenciation. Ces travaux montrent que DEPDC1A pourrait jouer un rôle essentiel dans la croissance des cellules de myélome et pourrait être une cible moléculaire prometteuse pour de nouvelles drogues ou de peptides-vaccins dans le MM. / Multiple myeloma (MM) is a B neoplasia characterized by the accumulation of a plasma cell clone in the bone marrow. This disease remains incurable, hence the need to identify new therapeutic targets. The use of DNA microarrays has identified many genes whose expression in MM cells is associated with poor or good prognosis. Most of the genes identified in the MM predictions encode proteins involved in DNA replication, repair and recombinaison processes. We wanted to go further in the identification and biological function of these prognostic genes.First, we looked for genes that have a spike expression, i.e. they are highly expressed in MMCs of a fraction of patients. These genes are generally associated with oncogenic events.On the other hand, we have identified pluripotent and adult stem cell genes unrelated to cell cycle and aberrantly expressed by human multiple myeloma cells in association with poor prognosis. The identification of these genes has allowed us to build a powerful prognostic score, stonger than already published scores.Another major aspect is the elucidation of biological mechanisms involving these genes that are responsible for drug resistance and/or short-term survival of patients, to revert them. We evaluated the biological role of DEPDC1A gene which are highly expressed in MM cells in association with a poor prognosis. The conditional knockdown of DEPDC1A by using an shRNA, inhibits the growth of myeloma cell lines with an accumulation of cells in G2/M phase of cell cycle. This accumulation is associated with phosphosphorylation and stabilization of p53, and accumulation of P21/WAF1. The knockdown of DEPDC1A also results in the expression of mature plasma cell in MM cell lines: CD31, CD38, CD138, IL6R, CXCR4, CD9, VLA6. DEPDC1A therefore controls the cell cycle of plasma cells by interfering with the p53 pathway and blocks their differentiation. This work shows that DEPDC1A could play a role in the growth of myeloma cells and could be a promising molecular target for new drugs or vaccine peptides in MM.

Orientador: Yvens Barbosa Fernandes / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Ciências Médicas / Made available in DSpace on 2018-08-24T09:26:09Z (GMT). No. of bitstreams: 1 Zullo_JoaoFlavioDaniel_M.pdf: 4994617 bytes, checksum: 7f874ff0599b4cec270f92a8010dc856 (MD5) Previous issue date: 2014 / Resumo: As hemorragias intracerebrais supratentoriais espontâneas, correspondem a cerca de 10 a 20% de todos os acidentes vasculares cerebrais (AVC), acometendo mais de 4 milhões de pessoas por ano em todo o mundo(13). Apresenta as mais altas taxas de morbi-mortalidade dentre os diferentes tipos de AVC, sendo que a mortalidade desta patologia alcança índices ao redor de 30-40% dos doentes, além de grande número de pacientes com seqüelas graves, sendo que seu principal fator de risco é a hipertensão arterial sistêmica (HAS). Desde o trabalho de McKissock(1) at al em 1961, vem se tentando descobrir qual a melhor maneira para tratar pacientes que apresentam tais hemorragias. Com esse intuito, avaliamos 67 pacientes atendidos no HC/Unicamp e no Hospital Estadual de Sumaré/Unicamp (HES) entre os anos de 2007 e 2012, sendo 41 com hemorragias de núcleos da base e 26 com hematomas lobares. Estudamos a influência de vários fatores no prognóstico dos pacientes com hemorragias intracerebrais supratentoriais espontâneas, como idade, nível de consciência à admissão hospitalar (usando-se a escala de coma de Glasgow), profundidade do hematoma em relação à superfície cerebral, volume do hematoma, presença de hemorragia intraventricular, desvio de linha média, tempo entre ictus hemorrágico e cirurgia e tipo de tratamento (craniotomia ou conservador). O nível de consciência na chegada ao hospital é fator determinante com relação ao prognóstico funcional avaliado em 30 dias após o ictus, independente da modalidade de tratamento adotada, porém com diferentes nuances dependendo da localização da lesão (núcleos da base ou lobar). Algumas particularidades, como tamanho do hematoma, podem determinar o resultado prognóstico funcional, assim como o tratamento cirúrgico, hemoventrículo e desvio de linha média (DLM) tendem a influenciar o prognóstico / Abstract: Spontaneous supratentorial intracerebral hemorrhages correspond to 10 to 20% of all stroke cases, affecting more than 4.000.000 people/year worldwide(13). It shows the highest rates of morbidity and mortality among the different types of stroke. The mortality of this disease achieves around 30- 40% of all patients, besides a large number of pacientes with severe sequels. Its main risk factor is high blood pressure. Since McKissock(1) at al published his work in 1961, researchers are trying to figure out the best way to treat patients with this type of hemorrhage. Aiming this situation, we evaluated 67 patients treated at HC/Unicamp and State Hospital of Sumaré (HES)/Unicamp between 2007 and 2012, with 41 basal ganglia and 26 lobar hematomas. We studied the influence of several factors on the prognosis of patients with spontaneous supratentorial intracerebral hemorrhage, such as age, consciousness at hospital admision (using the Glasgow Coma Scale), hematoma depth, hematoma volume, intraventricular hemorrhage, midline shift, time between bleeding and surgery and treatment option (craniotomy or conservative). The consciousness on arrival at the hospital is a determinant factor relative to functional outcome in 30 days after bleeding, regardless of treatment modality adopted, but with different nuances depending on the lesion location (basal ganglia or lobar). Some special features, like the size of the hematoma, can determine the funcional outcome, as well as surgery, ventricular hemorrhage and midline shift have a trend to influence prognosis / Mestrado / Neurologia / Mestre em Ciências Médicas

INTRODUÇÃO: As doenças cardiovasculares estão entre as principais causas de óbito no Brasil e no mundo. Dentre as doenças cardiovasculares, a insuficiência cardíaca (IC) participa de maneira importante para morbi-mortalidade por ser via final de todas as entidades que acometem o coração. A internação hospitalar constitui momento crucial no tratamento e sobrevida dos pacientes com IC. Neste momento, em que o estado da doença atinge seu período mais crítico, é de grande importância o conhecimento dos pacientes com maior risco, que necessitam de cuidados mais intensos. No entanto, a apuração dos custos hospitalares é tarefa difícil, principalmente nas situações de alta complexidade, onde a utilização de recursos nos diversos setores do hospital, materiais e medicamentos, é muito heterogênea. Assim, a busca de variáveis clínicas capazes de ajudar a identificar os pacientes com maior risco, morbidade hospitalar (e conseqüente maior tempo de internação), e o custo destas internações foram o escopo deste estudo. OBJETIVO: primariamente, identificar variáveis clínicas capazes de predizer prognóstico de sobrevida e custos de internação numa população de pacientes internados por IC. Secundariamente, determinar custo mediano destas internações, correlacionando os as variáveis clínicas, de etiologia da cardiopatia de base, e com o perfil hemodinâmico na admissão hospitalar. Visamos ainda projetar os dados da Instituição no modelo de regressão por árvore de decisão proposto pelo estudo ADHERE. MÉTODOS: Realizamos um estudo retrospectivo na qual foram analisados dados consecutivos referentes a internações de pacientes que chegaram ao Pronto Socorro do InCor e permaneceram no Hospital por mais de 24 horas, sendo internados nos anos de 2006 e 2007. Foram avaliados dados clínicos na chegada ao pronto atendimento e evolutivos durante a internação. Foi realizada avaliação de custo da doença durante internação hospitalar através de modelo misto de análises de custos diretos contabilizados por absorção total e rateio dos setores de apoio. Análises estatísticas incluíram modelos de: regressão de proporcional de Cox para variáveis de morbidade-permanência hospitalar, regressão logística para variáveis de mortalidade hospitalar, e regressão através de árvores de decisão para definição de variáveis prioritárias. RESULTADOS: Foram avaliadas 577 internações de pacientes diferentes, sendo 60% do sexo masculino, e idade mediana de 69 anos (57-77). As principais variáveis clínicas preditoras de tempo de internação para nossa população foram: perfil hemodinâmico C, necessidade de dobutamina, ventilação mecânica, ou antibióticos. As principais variáveis clínicas preditoras de mortalidade foram: fração de ejeção, pressão arterial sistólica, clearence estimado de creatinina, ocorrência de infecção hospitalar, e a necessidade de dobutamina, noradrenalina, ou cateteres centrais. Todas estas variáveis compuseram os modelos de regressão. O custo mediano das internações foi de R$ 4.450 (1.353 - 13.432), sendo o fator independente na análise multivariada, o tempo de internação hospitalar, que teve mediana de 5 dias (2-13). A mortalidade hospitalar geral foi de 132 pacientes (23%). CONCLUSÃO: As variáveis clínicas preditoras de tempo de internação para nossa população foram: perfil hemodinâmico, necessidade de dobutamina, ventilação mecânica, ou antibióticos. As variáveis clínicas preditoras de mortalidade foram a fração de ejeção, a pressão arterial sistólica, o clearence estimado de creatinina, a ocorrência de infecção hospitalar, e a necessidade de dobutamina, noradrenalina, ou cateteres centrais. Estas variáveis foram diferentes daquelas apontadas por outros estudos. A etiologia chagásica se correlacionou à maior incidência de choque cardiogênico, caracterizando assim maiores taxas de mortalidade, tempo de permanência, e custos frente às outras etiologias. A presença de choque cardiogênico na entrada se correlacionou a altas taxas de mortalidade, internações mais prolongadas, e maiores de custos de internação. O modelo descrito pelo estudo ADHERE pôde ser aplicado em nossa população, porém, propusemos outro modelo de árvore de decisão composto pelas variáveis: presença de choque cardiogênico uréia sérica, e pressão arterial sistólica, que apresentou maior acurácia em relação ao desfecho mortalidade hospitalar. O custo das internações variou muito de acordo com a evolução clínica dos pacientes, e conseqüentemente, seu tempo de internação hospitalar. No caso de pacientes atendidos pelo SUS, menos de um terço das internações tiveram custos inferiores ao valor médio das AIHs pagas por internações de pacientes com IC. / BACKGROUND: Heart diseases are the main mortality cause in Brazil and the rest of the world. Among those diseases, heart failure (HF) is utmost importance because it is the final pathway for overall heart diseases. Acute decompensate HF is a crucial situation while treating this disease because of its severity. At this critical time, stratification of risk is imperative in order to determine care. Hospital costs determination, however, is difficult in high complexity situations that use resources in a heterogeneity manner. The look for the clinical variables that could identify patients at higher risk for morbidity (and length of stay), mortality, and costs were the main aims of this study. OBJECTIVES: primarily to identify clinical variable able to predict survive and costs in a population of patients admitted by HF. Secondarily, determine median costs for the admissions, correlating these values to clinical variables, etiologies of HF, and hemodynamic profile at entrance. We aimed also to run our data in the tree regression model previously proposed by the ADHERE registry. METHODS: we reviewed consecutively 577 admissions records of different patients admitted by acute decompensated heart failure that stayed for more than 24 hours at the hospital during 2006 and 2007. Clinical data at the admissions and in-hospital follow-up data were analyzed. Costs analysis was performed through a mix model of microcosting (for direct resources) and average costing (for indirect resources). Statistical analysis included regression models as follows: Cox proportional for length of stay variables, logistic for hospital mortality, and classification and regression tree for defining priority variables. RESULTS: among the 577 patients, 60% were men; median age was 69 years (57- 77). The main predictor variables for length of stay were as follows: C hemodynamic profile, need for dobutamine, mechanic ventilation, or antibiotics. The main predictor variables for mortality were as follows: ejection fraction, systolic blood pressure, estimated creatinine clearance, occurrence of hospital infections, and need for dobutamine, norepinephrine, or central catheters. All these variables composed the regression models. Median admission cost was R$ 4.450 (1.353 13.432). Length of stay was an independent factor for predicting costs, with median of 5 days (2-13). Inhospital mortality rate was 23% (132 patients). CONCLUSION: The main predictor variables for length of stay were as follows: hemodynamic profile, need for dobutamine, mechanic ventilation, or antibiotics. The main predictor variables for mortality were as follows: ejection fraction, systolic blood pressure, estimated creatinine clearance, occurrence of hospital infections, and need for dobutamine, norepinephrine, or central catheters. These variables differ from other studies that evaluated similar outcomes. Chagas heart disease etiology was correlated to higher rates of cardiogenic shock, mortality rates, length of stay, and costs. The model used in the ADHERE registry could be used in our population; however, we proposed another variables integrating the regression and classification tree (systolic blood pressure, blood urea nitrogen, and hemodynamic profile C). This model presented greater accuracy for hospital mortality in our population. The cost of admissions ranged according to clinical evolution of the patients, and as consequence of length of stay. Less than a third of the admissions reimbursed by the government had their costs below the mean estimated value for reimbursement

L’objectif du travail a été d’explorer la valeur pronostique pour les tumeurs de la vessie des polymorphismes de gènes associés à l’inflammation et du microenvironnement tumoral lymphocytaire CD8+. Pour les marqueurs constitutionnels, deux approches ont été conduites concomitamment, l’une explorant de façon globale les gènes associés à l’inflammation, l’autre ciblant un gène inflammatoire d’intérêt, PDL1, impliqué dans des points de contrôles immunologiques. A l’échelle du génome, en utilisant des méthodes statistiques soit classiques soit innovantes (dites multi marqueurs), nous avons démontré que les variants (SNP) dans les gènes TNIP1, CD5 et JAK3 étaient associés au risque de récidive des tumeurs de vessie non invasives du muscle alors que les variants dans les gènes MASP1, AIRE et CD3 étaient associés au risque de progression. Dans un deuxième temps, l’association entre variants dans le gène de PDL1 et pronostic des tumeurs de vessie a été explorée en appliquant une méthode classique « SNP par SNP » et une approche à l’échelle du gène. Nous avons identifié une forte association entre des variants de PDL1 et le pronostic de tumeurs de vessie envahissant le muscle dans une large cohorte prospective, mais sans pouvoir répliquer ce résultat dans une série issue du consortium TCGA.Dans les tumeurs n’envahissant pas le muscle, nous avons développé et évalué une méthode d’évaluation standardisée de l’infiltrat lymphocytaire CD8+, cellules T-cytotoxiques impliquées dans la mort des cellules tumorales. L’analyse morphométrique après double immuno-marquage des cellules tumorales et des lymphocytes CD8+ et numérisation a permis d’estimer de façon séparée le compte des cellules inflammatoires dans la tumeur et le stroma, et d’estimer l’hétérogénéité spatiale intra-tumorale. Nous avons montré que cette hétérogénéité limite les estimations de l’infiltrat CD8+ sur les puces tissulaires (Tissue Micro Array) qui échantillonnent les tumeurs de façon restrictive. Sous cette réserve, nous avons identifié dans les tumeurs n’envahissant pas le muscle une association entre l’infiltrat lymphocytaire CD8+ et le stade tumoral Ta/T1, ainsi qu’avec le risque de récidive des tumeurs T1. A l’avenir, variations génétiques constitutionnelles dans les gènes de l’inflammation et évaluation de l’infiltrat tumoral inflammatoire pourraient être intégrées en vue d’améliorer la prédiction du pronostic des tumeurs vésicales. / The aim of this study was to explore prognostic value for bladder cancer of germline polymorphisms in inflammatory genes and tumor CD8+ lymphocytic microenvironment. For constitutional markers, two approaches were conducted jointly: one genome-based using specific GWAS statistical methods, the other gene-based focusing on PDL1, an inflammatory gene implicated in immune checkpoints. At the genome level, using both standard and innovative statistical methods (multi marker methods Bayesian Lasso and Bayes A) we demonstrated that variants (SNPs) in TNIP1, CD5 and JAK3 were associated with the risk of recurrence of non-muscle invasive bladder cancer (NMIBC) while SNPs in MASP1, AIRE and CD3 were associated with risk of progression. Meanwhile, association between PDL1 and prognosis of NMIBC and muscle invasive BC (MIBC) was explored using classical SNPS by SNP investigations and a gene based approach. We identified a very strong association between PDL1 variants and MIBC prognosis in a large prospective cohort but failed replicating those results in the TCGA consortium series.In non-muscle invasive bladder cancer, we developed and evaluated a standardized counting approach of CD8+ cells, T-cytotoxic lymphocytes implicated in tumor cells death. Morphometric analysis after double immuno-staining of tumor cells and digitalization allowed separate estimation of CD8+ cells in the tumor and stroma compartment and estimation of spatial intra tumoral heterogeneity. We demonstrated that this heterogeneity compromised CD8+ estimation on Tissue Micro Arrays, which sample the tumors in a restrictive manner. Keeping those limitations in mind, we identified an association between CD8+ inflammatory infiltrate and both NMIBC stage and T1 tumours risk of recurrence. In the future, germline variation in inflammatory genes and evaluation of tumor inflammatory infiltrate could be integrated for better prediction of bladder cancer prognosis.

Background: The overall good prognosis for Human Papillomavirus (HPV) driven base of the tongue cancer has prompted an increasing interest in whether this group could benefit from a less aggressive treatment regime. Different studies have drawn different conclusions as to which laboratory test should be used to identify these patients, using the surrogate marker p16, analyzing for HPV presence or both. Aim: The main purpose of this study was to investigate the presence of HPV-DNA and p16 in base of tongue cancer patients and their respective prognostic value, both used individually as well as combined. Material and methods: This was a retrospective cohort study consisting of 40 patients diagnosed with base of tongue cancer. The follow-up period was 5 years. Survival analysis was performed both depending on the combined results from the p16 immunohistochemistry analysis and the HPV DNA PCR, as well as separately. Results: Five-year survival rates were 73.9% for p16(+) and 17,6% for p16(-) subjects (p<0.001), 60.7% for HPV-DNA(+) and 25.0% for HPV-DNA(-) subjects (p=0.025). Five-year survival rates when combining p16 and HPV-DNA were 73.9% for p16(+)/HPV DNA(+), 25.0% for p16(-)/HPV DNA(-) and 0.0% for p16(-)/HPV DNA(+) (p<0.001). Conclusion: Our results add to previous research that p16 is a strong predictor of prognosis for base of tongue cancer patients, and could have the clinical implication of serving as a reliable tool for clinicians when determining prognosis and identifying patients who could benefit from treatment de-escalation in the future.

京都大学 / 0048 / 新制・論文博士 / 博士(医学) / 乙第13324号 / 論医博第2192号 / 新制||医||1043(附属図書館) / (主査)教授 井上 治久, 教授 高橋 淳, 教授 福原 俊一 / 学位規則第4条第2項該当 / Doctor of Medical Science / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(社会健康医学) / 甲第22835号 / 社医博第111号 / 新制||社医||11(附属図書館) / 京都大学大学院医学研究科社会健康医学系専攻 / (主査)教授 佐藤 俊哉, 教授 川上 浩司, 教授 木村 剛 / 学位規則第4条第1項該当 / Doctor of Public Health / Kyoto University / DFAM

京都大学 / 新制・課程博士 / 博士(医学) / 甲第23101号 / 医博第4728号 / 新制||医||1050(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 髙折 晃史, 教授 松田 文彦, 教授 藤田 恭之 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

京都大学 / 新制・課程博士 / 博士(医学) / 甲第23465号 / 医博第4772号 / 新制||医||1053(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 近藤 尚己, 教授 山本 洋介, 教授 松村 由美 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM