Graduation Date

Fall 12-19-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Programs

Interdisciplinary Graduate Program in Biomedical Sciences

First Advisor

Michael A. Hollingsworth

Abstract

Despite advances in treatment and detection, the prognosis for pancreatic ductal adenocarcinoma (PDAC) remains poor. This is driven, in part, by a dense, collagen-rich tumor microenvironment that restricts drug delivery and drives invasive and metastatic behavior. This dissertation characterizes organ-specific collagen structure and organization and its alteration by chemotherapy. We also describe the development of a novel cross-xenograft model to query the tumor niche effects in preclinical modeling of pancreatic ductal adenocarcinoma (PDAC) and extend this into a minimally invasive ultrasound guided injection (USGI) model.

With unique Rapid Autopsy Program (RAP) and Normal Organ Recovery (NOR) pancreas and liver samples, we quantified five fibril collagen features (width, length, straightness, alignment, and density) using Second Harmonic Generation (SHG) microscopy and fiber analysis (CT-FIRE, CurveAlign). Fibrillar collagen density and alignment are known prognostic factors in PDAC, but neither the effects of common chemotherapeutics on these, nor collagen of the liver metastatic niche, have been investigated previously. We found liver metastases have an organ-specific collagen organization that is uniquely enhanced by FOLFIRINOX (FOL) treatment. Although our samples are from end stage disease and not surgical resection samples, collagen at the pancreas border was more aligned than at the tumor core. We also found collagen metrics correlated with survival length, further supporting the progression of collagen remodeling.

Modeling these unique collagen changes in vivo became a full-time job of mine for years. We implanted RAP-derived pancreas and liver tissue into mouse pancreas or liver, or subcutaneously (subQ) onto the shoulder flank, then conducted transcriptomic and SHG analyses to interrogate how well these models recapitulate the human disease. Orthotopically implanted tissues more closely reflected the stroma and transcriptome of their derived tissues than ectopic implants. We observed organ-specific influence on the upregulated gene expression at each mouse location. Complementary development and optimization of a USGI cross-xenograft model was used to overcome the imaging artifact left by surgical scarring and to reduce the invasion and risk of surgeries. Together these results define a liver metastasis collagen signature, demonstrate chemotherapy-associated extracellular matrix (ECM) remodeling, and provide accessible and reproducible in vivo platforms for testing ECM targeting strategies and spatiotemporal hypotheses about metastatic niche adaptation.

Comments

2025 Copyright, the authors

Download

Available for download on Friday, December 10, 2027

Share

COinS