Neuroendokrine Neoplasmen der Bauchspeicheldrüse (PanNEN) sind eine seltene und vielfältige Form von Krebs. PanNENs umfassen hochgradige PanNECs und NETG3 PanNETs, sowie die öfter diagnostizierten NETG1 und NETG2 PanNETs. Hochgradige PanNENs weisen eine schlechte Prognose auf, und sind histologisch schwierig zu diagnostizieren. In dieser Studie wird eine auf Methylierung basierende Klassifizierung vorgestellt, die PanNETs von PanNECs unterscheidet und die zugrunde liegende Komplexität in Bezug auf molekularen Merkmalen und den Ursprungszellen der PanNENs aufzeigt. Zuerst wurden PanNENs auf Grundlage ihrer Methylierungsprofile gruppiert, wodurch sich die PanNETs und PanNECs in zwei Gruppen A und B aufteilten. Während Tumore der Gruppe B häufig Veränderungen in KRAS, TP53 und SMAD4 aufweisen, umfasst das Mutationsspektrum von Gruppe A Veränderungen in klassischen PanNEN-Genen wie MEN1, DAXX, ATRX und VHL. Darüber hinaus ist Gruppe A durch chromosomale Aberrationen geprägt, während Gruppe B eine signifikante fokale Deletion des RB1-Lokus aufweist. Anhand von Methylierungsprofilen von alpha-, beta-, duktalen und azinären Pankreaszellen sowie von Expressionsmustern normaler Zelltyp Markergene innerhalb der Tumore folgt, dass die PanNETs alpha-, beta- und intermediär-ähnliche Tumore endokrinen Ursprungs sind, während die PanNECs azinäre Tumore vermutlich exokrinen Ursprungs sind. Ausprägung des Azinuszellenprofils und Expression des SOX9-Proteins in Gruppe B sind vergleichbar mit denen des duktalen Adenokarzinoms der Pankreas (PDAC), einer Tumorentität mit nachgewiesen exokrinen Zellursprung. Insgesamt ergeben die neuen Erkenntnisse dieser Arbeit ein umfassendes Profil, das PanNET- und PanNEC-Tumore genetisch und epigenetisch eindeutig charakterisiert. Weiterhin liefert diese Arbeit starke Beweise für die aufkommende, aber unbewiesene Theorie des exokrinen Ursprungs von PanNECs und somit einen neuen Ansatz für die Behandlung dieser seltenen, aber oft tödlichen Krankheit. / Pancreatic Neuroendocrine Neoplasm (PanNEN) is a rare form of cancer comprising a heterogeneous set of high-grade tumors PanNECs and NETG3, in addition to the more commonly diagnosed NETG1 and NETG2 PanNETs. High-grade PanNENs display poor prognosis among patients and remains challenging to diagnose histologically. This study presents a methylation-based classification, which precisely distinguishes PanNETs and PanNECs, exposing the complexity evident in molecular and tumor cell-of-origin features of PanNENs. My work establishes distinct PanNEN subgroups based on methylation profiles. The PanNETs and PanNECs were separated into two groups: Group A and Group B, respectively. While Group B tumors are enriched for recurring alterations in KRAS, TP53 and SMAD4, the mutational spectrum of Group A encompasses alterations in classical PanNEN genes including MEN1, DAXX, ATRX and VHL. Recurring whole chromosomal aberrations are evident in Group A tumors, in contrast to Group B tumors, which reveal a significant focal deletion of the RB1 locus. Using methylation profiles of normal pancreatic cell types, in addition to expression patterns of normal cell type marker genes within tumors, the study concluded that the PanNET tumors are alpha-like, beta-like, and intermediate-like tumors of endocrine origin, while PanNECs of Group B are acinar-like tumors with a potential exocrine origin. The proportion of acinar-cell methylation profile and expression of SOX9 protein in Group B tumors are comparable to Pancreatic Ductal adenocarcinoma (PDAC), a tumor entity of exocrine cell-of-origin. Together, the novel findings of this thesis establish a comprehensive profile distinctly characterizing PanNET and PanNEC tumors at the genetic and epigenetic molecular level. Importantly, this work provides strong evidence for the emerging yet unproven theory of an exocrine origin of PanNECs, offering a new approach for treating patients with this rare but often fatal disease.

Gliomas are the most common primary tumors of the central nervous system believed to arise from glial cells. Invasive growth and inherent propensity for malignant progression make gliomas incurable despite extensive treatment. I have developed a life-like orthotopic glioma model and used this and other in vivo models to study basic mechanisms of glioma development and treatment. Previous studies had indicated that experimental gliomas could arise from glial stem cells and astrocytes. The present thesis describes the making and characterization of a novel mouse model, Ctv-a, where gliomas are induced from oligodendrocyte progenitor cells (OPCs). Our study shows that OPCs have the capacity to give rise to gliomas and suggests in light of previous data that the differentiation state of the cell of origin affects tumor malignancy. CDKN2A encodes p16INK4a and p14ARF (p19Arf in mouse) commonly inactivated in malignant glioma. Their roles in experimental glioma have been extensively studied and both proteins have tumor suppressor functions in glial stem cells and astrocytes. Here, we demonstrate that p19Arf only could suppress gliomagenesis in OPCs while p16Ink4a had no tumor suppressive effect. Functional DNA repair is pivotal for maintaining genome integrity, eliminating unsalvageable cells and inhibiting tumorigenesis. We have studied how RAD51, a central protein of homology-directed repair, affected experimental glioma development and have found that expression of RAD51 may protect against genomic instability and tumor development. Angiogenesis, the formation of new blood vessels from pre-existing ones, is a central feature of malignant progression in glioma. Antiangiogenic treatment by inhibition of vascular endothelial growth factor receptor signaling is used in the clinic for treatment of some cancers. We have investigated the effect of an alternative antiangiogenic protein, histidine-rich glycoprotein (HRG), on glioma development and found that HRG could inhibit the formation of malignant gliomas and completely prevent the formation of glioblastoma.