Prima Dewi Sinawang PhD Thesis Defense

Prima Dewi Sinawang
PhD Candidate
Chemical Engineering
Academic advisor: Professor Utkan Demirci

Abstract: "Late-stage diagnosis remains the dominant driver of poor outcomes in cancers, and minimally invasive tools that detect, stratify, or monitor disease from biological samples are of sustained clinical interest. Liquid biopsy—the analysis of tumor-derived molecular signals in blood and other biofluids—is one such tool, with extracellular vesicles (EVs) emerging as analytes of growing attention alongside circulating tumor cells, cell-free nucleic acids, and soluble proteins. The way an EV biomarker workflow is engineered upstream (e.g., in how samples are prepared, EVs are isolated, and the workflow is matched to the biological questions that are being asked) shapes what analytical readouts it can support downstream. The work in this dissertation builds on that principle.

This dissertation advances engineering strategies for EV isolation and analysis from biofluids, bridging engineering and biology to extend the molecular, biophysical, and biological dimensions along which EVs can be investigated across cancer contexts and beyond.

In metastatic castration-resistant prostate cancer, controlled plasma preparation enables recovery of EV-associated miR-375-3p cargo that tracks treatment response to docetaxel chemotherapy. In a separate prostate cancer cohort, EV isolation from serum enables quantification of EV-Trop2 protein, which distinguishes high-risk disease from low-risk and cancer-free groups. A machine learning classifier combining EV-Trop2 with PSA improves the risk group stratification beyond PSA alone. In prostate cancer cell-line models, ExoTIC fluidic size fractionation cascade enables multi-omic profiling across three EV size classes and reveals that cancer cargo distributes across the size axis. Morphology, RNA, and protein—each carrying complementary information at different EV size classes—position vesicle size as a biophysical dimension that has been little characterized. In pancreatic ductal adenocarcinoma, the same workflow supports a functional readout where the EV fraction carries active protease signal; EV-associated matrix metalloproteinase (MMP) activity differentiates patients from controls, showing that compositional and functional measurements can be made from the same isolated sample. Extending beyond cancer, the workflow robustly isolates EVs from seminal plasma, including from dried samples on substrates relevant to casework, yielding sufficient EV-associated DNA to generate complete forensic STR profiles for genetic identification.

Together, these chapters bridge engineering and biology: an engineered EV workflow, applied across diverse biofluids and cargo classes, supports biological investigations that have been constrained, in part, by the methodological capabilities of available EV isolation approaches. Across the cancer chapters, the workflow extends what EV cargo can be measured for (protein, RNA, DNA, and enzymatic activity) and further reveals an underexplored analytical axis that has been limited by engineering rather than biology: size fractionation of heterogeneous EV subpopulations, where vesicle size and morphology emerge as biophysical dimensions along which cancer-related features can be observed. These contributions position the workflow as a framework on which future EV biomarker investigation can be built, and point toward broader effort and applications to engineer how biological signals are accessed, isolated, and analyzed for earlier and more reliable disease detection."