In dieser Arbeit wurde die Bedeutung anatomischer Parameter der Orbita für die Ätiologie und Ausprägung der endokrinen Orbitopathie und des Exophthalmus bei Morbus Basedow untersucht. Zusätzlich wurden geschlechtsspezifische Unterschiede der Orbita analysiert. Hierzu erfolgte eine 3D-kephalometrische Untersuchung an 123 computertomographischen (CT) Datensätzen (52 mit endokriner Orbitopathie, 71 Kontrolldatensätze). Mittels 56 anatomischer Landmarken wurden 155 Strecken und 20 Winkel je Datensatz bestimmt und statistisch evaluiert. Es zeigte sich eine symmetrische Anatomie mit deutlichen Geschlechtsunterschieden bei 39 Strecken, zwei Winkeln sowie bei Orbitavolumen und –fläche. Hierbei lagen bei Männern signifikant größere Werte vor. Hinsichtlich der Gruppen mit und ohne endokrine Orbitopathie fanden sich bei Frauen Unterschiede bei 13 Strecken und zwei Winkeln, bei Männern hingegen nur bei einer Strecke. Bezüglich der Position des Bulbus oculi korrelierten größere Orbitadimensionen mit einem größeren Bulbusabstand zur Trichterspitze. Die sagittale Konfiguration des Orbitarandes zeigte einen signifikanten Einfluss auf die jeweiligen Hertel-Exophthalmometriewerte. Hinsichtlich der anatomischen Parameter der knöchernen Orbita konnte somit keine wesentliche Bedeutung für die Ätiologie der endokrinen Orbitopathie gefunden werden. Andererseits stellt die anatomische Konfiguration des Orbitarandes eine bedeutsame Einflussgröße für die klinische Exophthalmometrie dar. Die Ergebnisse dieser Untersuchung können zur Orbitaanalyse und Operationsplanung sowie zur Einschätzung postoperativer Ergebnisse genutzt werden.:Inhaltsverzeichnis 1 1 Bibliographische Beschreibung 2 2 Einführung 3 2.1 Morbus Basedow und endokrine Orbitopathie 3 2.2 Fragestellung 6 3 Material und Methoden 7 3.1 Patienten 7 3.2 3D-Kephalometrie und Volumenbestimmung der Orbita 7 3.3 Statistik 11 4 Publikation 12 5 Zusammenfassung 33 6 Literaturverzeichnis 34 7 Darstellung des eigenen Beitrags 46 8 Selbstständigkeitserklärung 47 9 Curriculum vitae 48 10 Publikationsliste 49 11 Danksagung 50

The majority of cancer research projects mainly focus on the epithelial cancer cell, while the role of the tumour stroma has been largely neglected. Conventional 2D techniques, such as well plates and other kinds of tissue culture plastic, and animal models are mainly used to broaden our understanding of how tumours arise, develop, and induce metastasis. However, there is accumulating evidence suggesting a tremendous impact of the non‐cancerous tumour stroma on carcinogenesis, while other publications illustrate the great importance of advanced 3D in vitro models for cancer research. The overall goal of this work was to investigate how cancer associated fibroblasts (CAFs; the most abundant component in the tumour stroma) and normal prostate fibroblasts (NPFs), isolated from patients diagnosed with aggressive forms of prostate cancer, contribute to angiogenesis, an important hallmark of cancer progression. For this purpose, a 3D in vitro angiogenesis co‐culture model was established. At first, two (semi‐) synthetic hydrogel platforms, gelatine methacrylate (GelMA) and star‐shaped (star)PEG‐heparin hydrogels were characterised and their physicochemical properties were compared with each other. Interestingly, GelMA gels shrank while starPEG‐heparin gels swelled in cell culture medium over the course of 24 hours. The cell concentration, in addition to the stiffness, was critical for the formation of endothelial networks, and the knowledge of swelling behaviour enabled the adjustment of initial cell density to ensure the density between both gel types was comparable. Moreover, preliminary tests with mesenchymal stem cells demonstrated that the hydrogel can be actively remodelled, as evaluated by stiffness parameters at day one and seven of incubation. Growth factors (GFs) affect cellular fate and behaviour, and storage, presentation and administration of such chemokines can be critical for certain cellular applications. Due to the high anionic charge density of heparin, starPEG‐heparin hydrogels are known to reversibly immobilise several GFs and thereby might mimic the GF reservoir of the extra cellular matrix. Thus, transport processes of GFs with low and high heparin affinity inside these hydrogels were analysed by fluorescence correlation spectroscopy and a bulk diffusion approach. Results indicated that diffusion constants were synergistically decreased with increasing size and heparin affinity of the diffusant. Next, the capability of endothelial cells (ECs) to self‐assemble and organise into 3D capillary networks was tested in GelMA, starPEG‐heparin and Matrigel hydrogels. Only starPEG‐heparin hydrogels allowed the formation of interconnected capillaries in macroscopic hydrogel samples. However, as it is widely used to test for pro‐ and anti‐angiogenic agents, the 2D Matrigel angiogenesis assay was included for subsequent co‐culture experiments of ECs and fibroblasts in order to investigate how the stromal cells influence the formation of endothelial networks. For a detailed characterisation of 3D structures, a conventionally applied 2D method (Maximum Intensity Projection for 3D reconstructed images, MIP) was compared to an optimised 3D analysing tool. As a result, it was discovered that MIP analysis did not allow for an accurate determination of 3D endothelial network parameters, and can result in misleading interpretations of the data set. Indirect co‐cultures of hydrogel‐embedded ECs with a 2D layer of fibroblasts showed that fibroblast‐derived soluble factors, including stromal cell‐derived factor 1 and interleukin 8, affected endothelial network properties. However, only co‐encapsulation of ECs and fibroblasts in starPEG‐heparin hydrogel discs revealed remarkable changes in endothelial network parameters between CAF and NPF samples. In detail, the total length and branching of the capillaries was increased. For two donor pairs, the diameter of capillaries was decreased in CAF samples compared to NPF samples, underlining the high physiological relevance of this model. In contrast, significant differences in 2D Matrigel assays were not detected between, CAF, NPF and control (ECs only) samples. In summary, a 3D angiogenesis co‐culture system was successfully developed and used to characterise stromal‐endothelial interactions in detail. The combination of advanced biomaterials (starPEG‐heparin) and 3D analysing techniques goes beyond conventional 2D in vitro cancer research, and opens new avenues for the development of more complex models to further improve the acquisition of more biologically relevant data.

The majority of cancer research projects mainly focus on the epithelial cancer cell, while the role of the tumour stroma has been largely neglected. Conventional 2D techniques, such as well plates and other kinds of tissue culture plastic, and animal models are mainly used to broaden our understanding of how tumours arise, develop, and induce metastasis. However, there is accumulating evidence suggesting a tremendous impact of the non‐cancerous tumour stroma on carcinogenesis, while other publications illustrate the great importance of advanced 3D in vitro models for cancer research. The overall goal of this work was to investigate how cancer associated fibroblasts (CAFs; the most abundant component in the tumour stroma) and normal prostate fibroblasts (NPFs), isolated from patients diagnosed with aggressive forms of prostate cancer, contribute to angiogenesis, an important hallmark of cancer progression. For this purpose, a 3D in vitro angiogenesis co‐culture model was established. At first, two (semi‐) synthetic hydrogel platforms, gelatine methacrylate (GelMA) and star‐shaped (star)PEG‐heparin hydrogels were characterised and their physicochemical properties were compared with each other. Interestingly, GelMA gels shrank while starPEG‐heparin gels swelled in cell culture medium over the course of 24 hours. The cell concentration, in addition to the stiffness, was critical for the formation of endothelial networks, and the knowledge of swelling behaviour enabled the adjustment of initial cell density to ensure the density between both gel types was comparable. Moreover, preliminary tests with mesenchymal stem cells demonstrated that the hydrogel can be actively remodelled, as evaluated by stiffness parameters at day one and seven of incubation. Growth factors (GFs) affect cellular fate and behaviour, and storage, presentation and administration of such chemokines can be critical for certain cellular applications. Due to the high anionic charge density of heparin, starPEG‐heparin hydrogels are known to reversibly immobilise several GFs and thereby might mimic the GF reservoir of the extra cellular matrix. Thus, transport processes of GFs with low and high heparin affinity inside these hydrogels were analysed by fluorescence correlation spectroscopy and a bulk diffusion approach. Results indicated that diffusion constants were synergistically decreased with increasing size and heparin affinity of the diffusant. Next, the capability of endothelial cells (ECs) to self‐assemble and organise into 3D capillary networks was tested in GelMA, starPEG‐heparin and Matrigel hydrogels. Only starPEG‐heparin hydrogels allowed the formation of interconnected capillaries in macroscopic hydrogel samples. However, as it is widely used to test for pro‐ and anti‐angiogenic agents, the 2D Matrigel angiogenesis assay was included for subsequent co‐culture experiments of ECs and fibroblasts in order to investigate how the stromal cells influence the formation of endothelial networks. For a detailed characterisation of 3D structures, a conventionally applied 2D method (Maximum Intensity Projection for 3D reconstructed images, MIP) was compared to an optimised 3D analysing tool. As a result, it was discovered that MIP analysis did not allow for an accurate determination of 3D endothelial network parameters, and can result in misleading interpretations of the data set. Indirect co‐cultures of hydrogel‐embedded ECs with a 2D layer of fibroblasts showed that fibroblast‐derived soluble factors, including stromal cell‐derived factor 1 and interleukin 8, affected endothelial network properties. However, only co‐encapsulation of ECs and fibroblasts in starPEG‐heparin hydrogel discs revealed remarkable changes in endothelial network parameters between CAF and NPF samples. In detail, the total length and branching of the capillaries was increased. For two donor pairs, the diameter of capillaries was decreased in CAF samples compared to NPF samples, underlining the high physiological relevance of this model. In contrast, significant differences in 2D Matrigel assays were not detected between, CAF, NPF and control (ECs only) samples. In summary, a 3D angiogenesis co‐culture system was successfully developed and used to characterise stromal‐endothelial interactions in detail. The combination of advanced biomaterials (starPEG‐heparin) and 3D analysing techniques goes beyond conventional 2D in vitro cancer research, and opens new avenues for the development of more complex models to further improve the acquisition of more biologically relevant data.