Auteur / Autrice : | William Weens |
Direction : | Dirk Drasdo |
Type : | Thèse de doctorat |
Discipline(s) : | Mathématiques appliquées |
Date : | Soutenance en 2012 |
Etablissement(s) : | Paris 6 |
Mots clés
Mots clés contrôlés
Résumé
As recently demonstrated for liver regeneration after drug-induced damage, organizationand growth processes can be systematically analyzed by a process chain of experiments,image analysis and modeling [43]. The authors of [43] were able to quantitatively characterizethe architecture of liver lobules, the repetitive functional building blocks of liver,and turn this into a quantitative mathematical model capable to predict a previously unrecognizedorder mechanism. The model prediction could subsequently be experimentallyvalidated. Here, we extend this model to the multi-lobular scale, guided by experimentalfindings on carcinogenesis in liver [15]. We explore the possible scenarios leading to the different tumor phenotypes experimentally observed in mouse. Our model considersthe hepatocytes, the main cell type in liver, as individual units with a single cell basedmodel and the blood vessel system as a network of extensible objects. Model motion iscomputed based on explicit discretized Langevin equation and cell interactions are eitherHertz or JKR forces. The model is parameterized by measurable values on the cell andtissue scale and its results are directly compared to the experimental findings. In a fundamental first step we study if Wnt and Ras signaling pathways can explainthe observation of [15], that instantaneous proliferation in mutated mice can only be observedif around 70% of the hepatocytes become APC depleted. In a second step, weshow a sensitivity analysis of the model on the vessel stiffness and relate it to a tumorphenotype (experimentally observed) where the tumor cells are well differentiated. Weintegrate in a third step the destruction of vasculature by tumor cells to relate it to anotherexperimentallyobserved tumor phenotype characterized by the absence of blood vessels. Finally, in the last step we show that effects that are detectible for small tumor nodulesand reflect properties of the tumor cells, are not reflected in the tumor shape or phenotypeat tumor sizes exceeding half of the lobule size.