Preview

PDF, English - main document Download (49MB) | Terms of use

Abstract

Prostate cancer (PCa) is a biologically and clinically heterogeneous disease. Current diagnostic strategies, including prostate-specific antigen (PSA) testing, imaging-based clinical staging, and histopathological assessment, are limited in their ability to fully capture this heterogeneity and to reliably support individualized risk stratification. Consequently, there is a critical need for additional molecular biomarkers and integrative diagnostic approaches that enable personalized, patient-oriented treatment decisions. Liquid biopsy analyses allow the minimally invasive assessment of tumor-derived molecules in body fluids, providing a comprehensive view of the entire tumor burden and its heterogeneity. The analysis of cell-free DNA (cfDNA), and in particular circulating tumor DNA (ctDNA), enables genomic and epigenomic tumor profiling. However, ctDNA detection in PCa is challenging due to low ctDNA shedding and a low genomic alteration burden, especially in localized disease. The aim of this PhD project was to establish a multimodal liquid biopsy framework that integrates genomic and epigenomic cfDNA analyses from matched plasma and urine samples of PCa patients at initial diagnosis. The goal was to improve ctDNA detection and to achieve a comprehensive molecular characterization of heterogeneous PCa, with a particular focus on identifying localized disease and stratifying aggressive tumors. To this end, I applied low-coverage whole-genome sequencing and cell-free methylated DNA immunoprecipitation sequencing to plasma and urinary cfDNA to assess chromosomal instability, copy number variations, cfDNA fragmentation patterns, and aberrant DNA methylation. I performed these analyses on 109 plasma samples and 102 urine samples obtained from 73 newly diagnosed PCa patients and 36 cancer-free controls. Most PCa patients presented with localized disease, while approximately one quarter had lymph node or distant metastases. In addition, I analyzed eight fresh-frozen PCa tissue samples with matched buffy coat DNA to enable comparisons with liquid biopsy data and to identify tumor-informative characteristics. Furthermore, I compared results from genomic and epigenomic cfDNA profiling between plasma and urine to identify shared and complementary characteristics across both biofluids, and between PCa patients and controls to distinguish tumor-specific patterns. I subsequently selected representative cfDNA features for ctDNA detection, with detection thresholds defined based on control samples. Finally, I assessed ctDNA positivity in relation to clinical and pathological characteristics. Fragmentation analyses revealed distinct and tumor-informative cfDNA profiles in plasma and urine, with biofluid-specific fragmentation features distinguishing PCa patients from controls. Genomic profiling identified copy number variations and increased chromosomal instability in both biofluids, providing complementary evidence of ctDNA presence. Samples with higher proportions of detectable ctDNA harbored recurrent genomic alterations consistent with those observed in PCa tissue samples and with known alterations in primary PCa reported in the literature. Epigenomic analyses identified differentially methylated regions in both plasma and urinary cfDNA that distinguished metastatic PCa patients from PCa patients without distant metastases and from cancer-free controls. Furthermore, plasma and urinary cfDNA from PCa patients showed increased methylation levels in PCa tissue-derived methylation regions, which were validated using an external PCa tissue dataset, further supporting ctDNA detection. Overall, genomic and epigenomic differences were most pronounced in advanced PCa, while cfDNA features related to DNA methylation and chromosomal instability also enabled discrimination between advanced and localized PCa. The integrated assessment of cfDNA fragmentation, copy number variations, chromosomal instability, and methylation in plasma and urine substantially increased ctDNA detection rates compared to single-parameter or single-biofluid analyses. CtDNA was detected in 45% of newly diagnosed PCa patients, including 42% of localized cases and 56% of advanced cases. Importantly, ctDNA was detectable in a considerable proportion of patients with low to intermediate PSA levels (< 10 ng/ml), a clinical scenario in which risk stratification remains challenging. These findings underscore the potential of cfDNA-based genomic and epigenomic markers to address an unmet clinical need and to complement established diagnostics for refined risk stratification at initial diagnosis. In conclusion, this proof-of-concept study introduces the first reported multimodal liquid biopsy framework integrating genomic and epigenomic cfDNA analyses from both plasma and urine in newly diagnosed PCa patients. The results demonstrate that combining multiple cfDNA features across two biofluids enables a more comprehensive molecular characterization of PCa and improves ctDNA detection, even in localized disease. Despite these promising results, several limitations must be acknowledged. CtDNA detection rates, although improved, remained moderate, reflecting both biological constraints and current technical limitations. The cohort size was moderate, with limited representation of advanced and metastatic disease. Further studies with larger, independent cohorts, as well as prospective clinical trials, will be required to validate multimodal liquid biopsy approaches and to evaluate their clinical applicability. Future perspectives include the integration of multimodal liquid biopsy data with additional diagnostic modalities, supported by machine learning-based approaches for data integration, to achieve comprehensive, multimodal representations of PCa biology and ultimately support improved risk stratification and personalized treatment decisions.