Nephroblastoma is the most prevalent pediatric kidney tumor, which occurs primarily in younger children with the average age at diagnosis of 42,5 months for girls and 36,5 months for boys. Even though its treatment is currently very succesful and the overall survival rate reaches over 90 %, there are still more things to be discovered and improved. An important role for the right choice of treatment plays not only the histology of tumor, but also the chromosomal changes present at tumor. Some of them (for example 1q gain, simultaneous deletion of 1p and 16q, TP53 deletion) were confirmed as negative prognostic markers because they are associated with an increased risk of relapse or with anaplastic type of nephroblastoma that is included in a high risk group. These changes are therefore used together with the tumor histology for stratification of nephroblastomas. Some of these changes were found in a heterogeneous state (only in a part of the cells) in nephroblastoma, which also complicates the treatment of the patient and which cannot be solved when only one sample is taken from the tumor. In this work we concentrated on the detection of chromosomal changes present in nephroblastomas of 44 patients and their associations with clinical data. We have proved some of the known associations (22q...

Le carcinome à cellules claires du rein (ccRCC) est une tumeur très hétérogène. Le taux de métastase est de l’ordre de 50% et les métastases des ccRCC sont peu fréquemment opérées. L’objectif de la thèse est d’une part d’analyser les facteurs prédictifs de métastases dans la tumeur primitive, et d’autre part de comparer le phénotype des tumeurs primitives et de leurs métastases dans le ccRCC par différentes approches histopathologiques et génomiques. Parmi les facteurs prédictifs de métastases, la nouvelle classification OMS/ISUP remplace l’ancienne classification de Fuhrman. L’intérêt pronostique de la nécrose tumorale est également mis en évidence. Une tumeur avec un grade donné associée à la présence de nécrose a un pronostic qui se rapproche d’une tumeur de grade plus élevé sans nécrose. Le pourcentage des cellules de grade 4 pourrait également contribuer à la stratification pronostique des patients. Une différence de survie avec un pronostic défavorable est observée pour les tumeurs dont le pourcentage des cellules de grade 4 est plus élevé (>50% vs. <10%). Au niveau moléculaire le statut du gène VHL, gène suppresseur de tumeur inactivé par un « double hit », est impliqué dans le pronostic des patients. Les ccRCC sans aucune altération de VHL sont des tumeurs plus agressives, qu’il convient d’isoler. Il existe une similarité morphologique et immunohistochimique entre les métastases et la composante de plus haut grade des tumeurs primitives correspondantes. Le profil chromosomique des métastases n’est pas totalement superposable à celui des tumeurs primitives. Il existe anomalies cytogénétiques récurrentes dans des métastases de ccRCC à des sites différents (+2p, +3q, +5, +8q, +12, +20). L’hétérogénéité tumorale retrouvée au niveau des tumeurs primitives est également retrouvée au niveau des métastases sous forme d’hétérogénéité inter- et intra-métastatique. L’analyse combinée des profils génomiques et transcriptomiques de 14 échantillons prélevés au sein d’un ccRCC primitif et de ses métastases ont permis d’identifier trois classes de clones tumoraux distincts, ne suivant aucune logique géographique. Enfin il semble exister un phénomène de multi-colonisation, qui implique non pas un, mais plusieurs clones tumoraux qui pourraient agir conjointement dans le processus métastatique. / Clear cell renal cell carcinoma (ccRCC) is a heterogeneous tumor. The metastatic rate is 50% and metastases are only rarely surgically excised. The objective of this thesis was to analyze the predictive factors of metastasis in the primary tumor on one hand; and to compare primary and metastatic phenotypes on the other hand. We combined different histopathological and genomic approaches. Considering prognostic factors, the new WHO/ISUP classification replaces the previous Fuhrman grade. The interest of tumor necrosis is also highlighted. A tumor of a certain grade with necrosis has a prognosis that is close to a tumor of higher grade without necrosis. The percentage of grade 4 cells could additionally help in stratifying patients. A significant difference in survival is observed between tumors with more than 50% grade 4 cells and tumors with less than 10%. At a molecular level, the VHL gene status (tumor suppressor gene inactivated by a double hit) could be implicated in the prognosis of patients. ccRCCs with no alteration of the gene are more aggressive tumors that need to be identified. There exists a morphological and immunohistochemical similarity between metastases and the high-grade component in corresponding primary tumors. Moreover, the chromosomal profile of metastases differs from those of the corresponding primary tumors. Recurrent cytogenetic events are observed in different metastatic sites (+2p, +3q, +5, +8q, +12, +20). The tumor heterogeneity phenomenon in primary tumors is also observed in metastases with inter- and intra- metastatic heterogeneity. The combined analysis of transcriptomic and genomic analyses of 14 specimens extracted from a single ccRCC and its metastases divided samples into three classes of sub-clones with no spatial link. We observe a multi-colonization process that implies not only one but several tumor clones that could cooperate in the metastatic process.

L’hétérogénéité tumorale (HT) est un trait caractéristique du cancer. Elle peut être rencontrée chez la majorité des tumeurs malignes solides, au travers de la clinique, la biologie, l’histologie, et la génétique. Au-delà de l’hétérogénéité inter-patient, on distingue l’hétérogénéité spatiale, qui regroupe l’hétérogénéité au sein de la tumeur primitive, entre tumeur primitive et métastases, entre métastases, au sein d’une métastase et l’hétérogénéité temporelle, qui fait référence à l’évolution de la tumeur au cours du temps de manière spontanée ou sous l’effet des traitements. L’HT explique la survenue inéluctable de la résistance aux thérapies anticancéreuses actuelles. L’objectif général de cette thèse était d’explorer l’hétérogénéité tumorale au plan clinique, histologique et génétique en utilisant le modèle d’hépatectomies pour métastases d’origine colo-rectales. Dans l’article 1, nous avons étudié la réponse histologique à la chimiothérapie chez des patients opérés de métastases hépatiques colo-rectales après chimiothérapie systémique ou intra-artérielle. Ainsi, nous avons montré que la réponse histologique complète était plus souvent observée après administration d’oxaliplatine par voie intra-artérielle et était associée à un meilleur pronostic. Cependant la réponse histologique complète n’est observée que chez une minorité de patients. Nous avons émis l’hypothèse qu’une réponse histologique incomplète représentent un groupe hétérogène de patients, ayant des pronostics différents. Ceci nous a conduit à proposer dans l’article 2 une méthode reproductible pour évaluer la réponse histologique, incorporant taille et nombre de nodules. La relecture des lames de pièces d’hépatectomie nous a conduit à observer une hétérogénéité de la réponse histologique au sein d’un même patient (hétérogénéité intermétastatique) et de l’aspect histologique (nécrose, fibrose). Nous avons ensuite exploré la valeur pronostique de cette hétérogénéité et chercher à identifier les facteurs prédictifs de cette hétérogénéité. Une réponse dissociée (une différence de réponse histologique > 50% entre deux nodules chez un même patient) a été observée chez 20% des patients et ne modifiait pas le pronostic. Chez les patients ayant une hétérogénéité pathologique, une discordance (mutation et absence de mutation pour les gènes KRAS, BRAF, NRAS et PI3K) entre les métastases a été mis en évidence dans 28% des patients analysés (article 3). Afin d’étudier l’hétérogénéité génétique selon le site tumoral et le type de prélèvements, nous avons utilisé les données disponibles de la plateforme de biologie moléculaire (article 4). Nous avons ainsi mis en évidence que les métastases hépatiques étaient moins souvent mutées pour KRAS, NRAS, BRAF que les tumeurs primitives ou les lésions pulmonaires. Nous avons ensuite recherché une hétérogénéité génétique intra-métastatique pour KRAS, NRAS, BRAF et PI3K au sein d’une métastase hépatique colo-rectale (article 5). Parmi les 54 patients ayant une tumeur unique analysable (2 prélèvements par tumeur), une discordance pour KRAS (N=2) et BRAF (N=1) a été mise en évidence chez 3 patients (5%). L’ensemble de ces travaux confirment que l’HT est observée à travers de nombreux points de vue. L’élaboration de nouvelles stratégies de prise en charge du cancer devra prendre compte cet aspect. / Tumor heterogeneity is a typical feature of cancer. It is observed in the majority of solid malignancies regardless of the point of view: clinical, biological, pathological, and genetic. Different levels of heterogeneity have been described: interpatient, spatial heterogeneity (within the primary tumor, between primary and distant lesion) and temporal heterogeneity (tumor evolution under the influence of treatment). Tumor heterogeneity widely explains the emergence of resistant clones to anticancer drugs. The objective of the current thesis was to explore tumor heterogeneity at a clinical, pathological and genetic level, by using the model of hepatectomy for colorectal liver metastases (CLM).In the article 1, we studied pathological response to chemotherapy in patients operated on for CLM after either systemic or intra-arterial hepatic oxaliplatin-based chemotherapy. We showed that complete pathological response was more often observed after intra-arterial chemotherapy and yield far better outcomes. However, complete response was observed in a minority of patients. We hypothesized that uncomplete pathological response encompass a large group of patients with different oncological outcomes. This lead us to propose a reproducible method (article 2) to assess pathological response, including size and number of nodules. Pathological review found heterogeneity in the response among nodules within the same patients and in the type of pathological features (necrosis, fibrosis). We then explore the prognostic value of pathological heterogeneity and sought to identify predictors of heterogeneity. A dissociated response (difference in pathological response > 50% between two nodules) was observed in XX and did not impact long-term outcomes. IN patients with dissociated response, genetic heterogeneity (mutation and no mutation for KRAS, NRAS, BRAF and PI3K) between metastases was shown in 28% of patients (article 3). To study genetic heterogeneity according to tumor location and the tumor tissue analyzed (specimen, biopsy), we used data from the department of molecular biology (article 4). We found that liver metastases were more often wild type for KRAS, NRAS, BRAF compared to lung metastases or primary tumors. In the article 5, we then sought to evaluate intrametastatic heterogeneity (KRAS, BRAF, NRAS, PI3K) within a single liver metastasis (2 samples per tumor). Among the 54 patients with single lesion analyzed, a discrepancy for KRAS (n=2) and BRAF (n=1) were observed un 3 patients (5%). This work confirms that tumor heterogeneity can be observed at various levels. Elaboration of new therapeutical strategies will have to take this aspect into consideration.

Regulation of cancer cell phenotype by the tumor microenvironment has motivated further investigation into how microenvironmental factors could contribute to tumor initiation, development, and therapeutic resistance. Analyzing how the microenvironment drives tumor development and cancer cell heterogeneity is particularly important in cancers such as glioblastoma multiforme (GBM) that have no known risk factors and are characterized by a high degree of heterogeneity. GBM patients have a median survival of 15 months and therefore are in great need of more effective therapeutic options. The goal of this research is to generate in vitro models of the heterogeneous brain tumor microenvironment, with a focus on vascular dynamics, to probe the impact of microenvironmental cues on tumor progression and to integrate the tumor models with highly sensitive analytical tools to characterize the epigenome of discrete phenotypic subpopulations that contribute to intratumoral cellular heterogeneity. As GBM tumors are characterized by a dense vasculature, we delved into microenvironmental factors that may be promoting angiogenesis. The correlations emerging between inflammation and cancer led to analysis of the inflammatory molecule lipopolysaccharide (LPS). We utilized 3D micro-tissue models to simulate vascular exposure to ultra-low chronic inflammatory levels of LPS and observed an increase in vascular formation when brain endothelial cells were exposed to ultra-low doses of LPS. We also utilized our micro-tissue models to analyze histone methylation changes across the epigenome in response to microenvironmental cues, namely culture dimensionality and oxygen status. The H3K4me3 modification we analyzed is associated with increased gene transcription, therefore the alterations we observed in H3K4me3 binding across the genome could be a mechanism by which the tumor microenvironment is regulating cancer cell phenotype. Lastly, we developed a microfluidic platform in which vascular dynamics along with microenvironmental heterogeneities can be modeled in a more physiologically relevant context. We believe the studies presented in this dissertation provide insight into how vasculature primed by chronic inflammation and epigenetic alterations in tumor cells could both contribute to enhanced tumor development. Modeling these biological processes in our advanced microfluidic platform further enables us to better understand microenvironmental regulation of tumor progression, uncovering new potential therapeutic targets. / PHD / Regulation of cancer cell behavior by the tumor microenvironment, which includes the surrounding extracellular matrix, native healthy cells, and signaling molecules, has motivated further investigation into how microenvironmental factors could contribute to tumor initiation, development, and therapeutic resistance. Analyzing how the microenvironment drives tumor development and heterogeneity in cancer cell behavior is particularly important in cancers such as glioblastoma multiforme (GBM) that have no known risk factors and are characterized by a high degree of heterogeneity. GBM patients have a median survival of 15 months and therefore are in great need of more effective therapeutic options. The goal of this research is to generate models of the heterogeneous brain tumor microenvironment with a focus on how microenvironmental cues impact blood vessel development, which facilitates tumor progression. We will also use these tumor models, along with sensitive analytical tools, to characterize epigenetic modifications that potentially contribute to tumor cell heterogeneity. As GBM tumors are characterized by a dense vasculature, we delved into microenvironmental factors that may promote blood vessel growth. The correlations emerging between inflammation and cancer led to analysis of the inflammatory molecule lipopolysaccharide (LPS). We utilized 3D tumor models to simulate blood vessel exposure to ultra-low chronic inflammatory levels of LPS and observed an increase in blood vessel formation when brain endothelial cells were exposed to ultra-low doses of LPS. We also utilized our tissue models to analyze histone methylation changes across the epigenome in response to microenvironmental cues, namely culture dimensionality and oxygen status. The histone methylation changes we observed across the genome could be a mechanism by which the tumor microenvironment is regulating cancer cell v behavior. Lastly, we developed a microfluidic platform in which blood vessel development along with microenvironmental heterogeneities can be modeled in a more physiologically relevant context. We believe the studies presented in this dissertation provide insight into how blood vessel exposure to chronic inflammatory factors and epigenetic alterations in tumor cells could both contribute to enhanced tumor development. Modeling these biological processes in our advanced microfluidic platform further enables us to better understand microenvironmental regulation of tumor progression, uncovering new potential therapeutic targets.

A heterogeneidade entre as células tumorais e o suporte a elas proporcionado pelos componentes do microambiente tumoral (TME) são os dois principais responsáveis pela progressão do câncer e por tornar essas doenças essencialmente incuráveis. Assim, identificar as principais dependências das células malignas, sejam elas internas ou advindas do meio extracelular, é fundamental para entender seu comportamento e propor terapias mais eficientes. Nesta tese, abordamos aspectos destas duas questões separadamente. Em um primeiro trabalho, investigamos as interações de células tumorais com células-tronco mesenquimais (MSCs), um dos principais componentes do TME. MSCs participam ativamente do nicho tumoral, especialmente por serem capazes de liberar uma vasta gama de moléculas que, via sinalização parácrina, podem modular as células ao seu redor. No entanto, os principais mediadores e respectivos efeitos do secretoma dessas células nos tumores ainda precisam ser melhor elucidados. Ao investigar esses efeitos em glioblastomas (GBM), um dos tumores primários mais agressivos em adultos, mostramos que o secretoma de células-tronco mesenquimais derivadas de tecido adiposo humano (hADSCs) foi capaz de bloquear a autofagia das células malignas. Nossos dados revelaram que o secretoma de hADSCs ativou a via de sinalização de mTORC1 e reduziu a translocação nuclear de TFEB, um fator de transcrição chave que regula a autofagia e a a função lisossomal, nas células de GBM, impedindo que o fluxo autofágico fosse completado. Já em um segundo trabalho, no contexto da heterogeneidade celular em tumores, propusemos uma abordagem para análise de dados de céulas únicas focada em outliers. Minorias celulares com níveis anormalmente elevados, ou reduzidos, de expressão de determinados genes ou proteínas são em muitos casos responsáveis por resistir aos tratamentos e levar à recidiva da doença, ao mesmo tempo que, por serem outliers, são muitas vezes ignoradas ou excluídas das análises de dados. Assim, decidimos utilizar métodos estatísticos em dados de expressão de células únicas para detectar e analisar células outliers, comparando o seu comportamento com as demais células não-outliers. Denominamos essa abordagem de Single Cell OUTlier analysis (SCOUT) e a testamos em dados de células tumorais avaliadas por citometria de massas e por sequenciamento de RNA de células únicas (sc-RNA-seq). Como resultado, pudemos confirmar que, especialmente diante de determinados tratamentos, células outliers podem se comportar de maneira distinta de não-outliers, revelando informações potencialmente relevantes ao desenvolvimento de estretégias terapêuticas. Por fim, desenvolvemos uma ferramenta para automatizar a detecção e seleção de outliers em dados de célula única a fim de facilitar o estudo dessas células em diversos aspectos na pesquisa do câncer. / Intratumoral heterogeneity and the support provided by components of the tumor microenvironment (TME) to malignant cells are major contributors to cancer progression, and the two main factors that make this disease essentially incurable. Thus, identifying malignant cells dependencies, either in the intra- or extracellular environment, is fundamental to understand their behavior and propose more efficient therapies. In this thesis, we approached aspects of these two issues separately. In a first work, we investigated interactions between tumors and mesenchymal stem cells (MSCs), one of the main components in the TME. MSCs actively participate in the tumor niche, especially due to their capacity of releasing a wide range of molecules that can modulate cells in their surroundings. However, little is known about the effects of MSCs-derived molecules in tumor cells behavior. In investigating these effects on glioblastomas (GBM), one of the most aggressive primary tumors in adults, we found out that the secretome of human adipose-derived stromal cells (hADSCs) was able to block autophagy in malignant cells. Our data revealed that hADSCs secretome activated mTORC1 signaling pathway and reduced nuclear translocation of TFEB, a master transcription factor that regulates autophagy and lysosomal function, in GBM cells, preventing autophagic flux from being completed. In a second work, we addressed intratumoral heterogeneity by proposing an approach to analyze outliers in single cell data. Cellular minorities with abnormally high, or low, expression levels of certain genes or proteins are in many cases responsible for resisting treatments and lead to disease relapse, while for being outliers they are also frequently ignored or excluded from data analysis. Thus, we decided to apply statistical methods on single cell expression data to detect outliers and analyze them, comparing their behavior with the remaining non-outlier cells. We called this approach Single Cell OUTlier analysis (SCOUT) and tested it on tumor cell datasets obtained from mass cytometry and single cell RNA sequencing (scRNA-seq) experiments. Using SCOUT we were able to confirm that, especially upon specific treatments, outlier cells may behave differently from non-outliers, revealing potentially relevant information to aid in the development of novel therapeutic strategies. Finally, we developed a tool to automate detection and selection of outliers in single cell data with the aim to facilitate the study of these cells under different contexts in cancer research.

Gene expression profiles are some of the most abundant sources of data about the cellular state of a collection of cells in an organism. Comparison of the expression profiles of multiple samples allows biologists to find associations between observations at the molecular level and the phenotype of the samples. A key challenge is to distinguish variation in expression due to biological factors of interest from variation due to confounding factors that can arise for unrelated technical or biological reasons. This thesis presents models that can explicitly adjust the comparison of expression profiles to account for specific types of confounding factors. One such confounding factor arises when comparing tissue-specific expression profiles across multiple organisms to identify differences in expression that are indicative of changes in gene function. When the organisms are separated by long evolutionary distances, tissue functions may be re-distributed and introduce expression changes unrelated to changes in gene function. We developed Brownian Factor Phylogenetic Analysis, a model that can account for such re-distribution of function, and demonstrate that removing this confounding factor improves tasks such as predicting gene function. Another confounding factor arises because current protocols for expression profiling require RNA extracts from multiple cells. Often biological samples are heterogeneous mixtures of multiple cell types, so the measured expression profile is an average of the RNA levels of the constituent cells. When the biological sample contains both cells of interest and nuisance cells, the confounding expression from the nuisance cells can mask the expression of the cells of interest. We developed ISOLATE and ISOpure, two models for addressing the heterogeneity of tumor samples. We demonstrated that modeling tumor heterogeneity leads to an improvement in two tasks: identifying the site of origin of metastatic tumors, and predicting the risk of death of lung cancer patients.

Gene expression profiles are some of the most abundant sources of data about the cellular state of a collection of cells in an organism. Comparison of the expression profiles of multiple samples allows biologists to find associations between observations at the molecular level and the phenotype of the samples. A key challenge is to distinguish variation in expression due to biological factors of interest from variation due to confounding factors that can arise for unrelated technical or biological reasons. This thesis presents models that can explicitly adjust the comparison of expression profiles to account for specific types of confounding factors. One such confounding factor arises when comparing tissue-specific expression profiles across multiple organisms to identify differences in expression that are indicative of changes in gene function. When the organisms are separated by long evolutionary distances, tissue functions may be re-distributed and introduce expression changes unrelated to changes in gene function. We developed Brownian Factor Phylogenetic Analysis, a model that can account for such re-distribution of function, and demonstrate that removing this confounding factor improves tasks such as predicting gene function. Another confounding factor arises because current protocols for expression profiling require RNA extracts from multiple cells. Often biological samples are heterogeneous mixtures of multiple cell types, so the measured expression profile is an average of the RNA levels of the constituent cells. When the biological sample contains both cells of interest and nuisance cells, the confounding expression from the nuisance cells can mask the expression of the cells of interest. We developed ISOLATE and ISOpure, two models for addressing the heterogeneity of tumor samples. We demonstrated that modeling tumor heterogeneity leads to an improvement in two tasks: identifying the site of origin of metastatic tumors, and predicting the risk of death of lung cancer patients.

Classification of lymphoid neoplasms is based mainly on histologic, immunologic, and (rarer) genetic features. It has been supplemented by gene expression profiling (GEP) in the last decade. Despite the considerable success, particularly in associating lymphoma subtypes with specific transcriptional programs and classifier signatures of up- or downregulated genes, competing molecular classifiers were often proposed in the literature by different groups for the same classification tasks to distinguish, e.g., BL versus DLBCL or different DLBCL subtypes. Moreover, rarer sub-entities such as MYC and BCL2 “double hit lymphomas” (DHL), IRF4-rearranged large cell lymphoma (IRF4-LCL), and Burkitt-like lymphomas with 11q aberration pattern (mnBLL-11q) attracted interest while their relatedness regarding the major classes is still unclear in many respects. We explored the transcriptional landscape of 873 lymphomas referring to a wide spectrum of subtypes by applying self-organizing maps (SOM) machine learning. The landscape reveals a continuum of transcriptional states activated in the different subtypes without clear-cut borderlines between them and preventing their unambiguous classification. These states show striking parallels with single cell gene expression of the active germinal center (GC), which is characterized by the cyclic progression of B-cells. The expression patterns along the GC trajectory are discriminative for distinguishing different lymphoma subtypes. We show that the rare subtypes take intermediate positions between BL, DLBCL, and FL as considered by the 5th edition of the WHO classification of haemato-lymphoid tumors in 2022. Classifier gene signatures extracted from these states as modules of coregulated genes are competitive with literature classifiers. They provide functional-defined classifiers with the option of consenting redundant classifiers from the literature. We discuss alternative classification schemes of different granularity and functional impact as possible avenues toward personalization and improved diagnostics of GC-derived lymphomas