3037 - Urine Cell-Free RNA Enables Tumor Tissue-of-Origin Prediction in Genitourinary Cancers

Purpose/Objective(s): Urine is a promising ultra-non-invasive analyte for genitourinary (GU) cancer detection; however, its cell-free RNA (cfRNA) snapshot of transcriptional activity remains largely untapped in diagnostics, treatment monitoring, and tumor tissue-of-origin prediction. Here, we characterize the urine cfRNA transcriptome to identify tumor-specific signatures and leverage urine cfRNA for non-invasive tumor tissue-of-origin prediction, advancing GU cancer classification and precision oncology.

Materials/Methods: A total of 100 GU cancer patients and 24 healthy adults were enrolled in this prospective study. Preoperative urine samples were collected from 50 bladder cancer (BC) patients undergoing cystectomy, 16 renal cell carcinoma (RCC) patients undergoing nephrectomy, and 34 metastatic prostate cancer (mPC) patients. Urine cfRNA was isolated, DNase-treated and purified, followed by Illumina RNA prep with enrichment for library preparation. Sequencing was performed on an Illumina NovaSeq X with ~100 million reads per sample. Differentially expressed genes (DEGs) were identified using DESeq2, while transcription factor (TF) activity was inferred via collecTRI and pathway activation using PROGENy. MSigDB was used for gene set enrichment analysis. A logistic regression LASSO model classified tumor tissue of origin based on cfRNA profiles.

Results: Urine cell-free RNA pathway analysis revealed upregulation of PI3K-AKT-mTOR, TRAIL, WNT and TGF-ß signaling in bladder cancer patients, representing transcriptional alterations driving tumor progression, survival and EMT. Concurrently, androgen response, TNFa and p53 pathways in urine cfRNA were downregulated in bladder cancer patients compared to the other groups. Furthermore, TF analysis identified E2F1, STAT2 and FOXM1 as upregulated in bladder cancer urine cfRNA, reinforcing roles in cell cycle progression and immune modulation. In mPC, MYC, AR, NFKB, REST and JUN were upregulated, driving cell cycle progression, survival and transcriptional reprogramming, with activation of JAK-STAT, androgen, VEGF, and hypoxia signaling, supporting metastatic progression and tumor adaptation. RCC exhibited upregulation of MYC, JUN, STAT1, NFKB, GATA3, ESR1, HIF1A, MYB and STAT5A, alongside activation of MAPK, hypoxia, androgen, JAK-STAT, and TGF-ß signaling, with estrogen and TRAIL pathways downregulated. Finally, logistic regression LASSO-based machine learning classification using the top 250 coding transcripts between the groups achieved an AUC of 0.93, with accuracies of 85% for mPC, 86% for BC, and 72% for RCC, further demonstrating urine cfRNA’s potential for non-invasive GU cancer diagnosis and classification.

Conclusion: Urine cfRNA reflects GU cancer-specific transcriptomes and enables tumor tissue-of-origin prediction, laying the foundation for its potential as an ultra-non-invasive liquid biopsy analyte for GU cancer detection, profiling, and radiotherapy response monitoring.