Unraveling tumor- and tissue-specific gene expression patterns from plasma-derived cfRNA

Minimally invasive plasma-derived cell-free RNA (cfRNA) profiling captures blood-derived clinical markers when white blood cells (WBCs) are unavailable, harboring potential for molecular diagnostics and disease monitoring. Our study explored cfRNA analysis as a promising method for deconvolving signals from tumors and various tissues from a single blood draw, expanding its utility to include functional genomics that reveals gene expression patterns and their impact on disease status. Plasma cfRNA was obtained from 207 healthy donors and 214 cancer patients. For comparison, 96 paired tumor whole exome sequencing (WES) and 208 paired WBC RNA-seq samples were used. Gene expression was quantified using pseudo-alignment with Kallisto and signature scores were computed by single-sample gene set enrichment analysis. Statistical significance was assessed with the Mann-Whitney U test. Somatic single nucleotide variants (SNVs) identified by tumor WES were cross-checked for detection in matched plasma cfRNA. Extracellular matrix organization, cell proliferation, and T- and B-cell recruitment signatures were significantly elevated in cfRNA compared with WBCs (p values 1.1×10⁻⁵⁵, 3.8×10⁻⁶³, and 2.8×10⁻⁶², respectively). These cfRNA (but not WBC) signatures effectively distinguished cancer patients from healthy donors (p value 6.5×10⁻¹¹, 1.1×10⁻⁴, and 2.3×10⁻⁵, respectively). The epithelial signature was enriched in carcinoma cfRNA, with higher scores corresponding to larger numbers of metastatic sites (p value 1.5×10⁻⁷). Conversely, this score remained low in sarcomas originating from non-epithelial tissues. SNVs identified by tumor WES were detected in cfRNA in 20/29 (68.9%) of breast, 8/14 (57.1%) of colorectal, 7/13 (53.8%) of lung, and 5/11 (45.4%) of pancreatic cancer cases. The number of variants detected per patient trended positively with cancer stage. In colorectal cancer, samples with detected variants had higher epithelial and colon tissue signature scores, which aligns with reported cfDNA shedding levels for the diagnosis . Fisher's exact test showed the cfRNA-tumor variant overlap to occur significantly more often (p-value = 9×10⁻¹²⁹) in matched pairs than in random sample comparisons, indicative of the reliability of our approach for detecting tumor-related signals. This proof-of-concept study shows that cfRNA harbors tumor- and tissue-specific expression patterns that complement blood cell-based RNA-seq, providing additional biological and clinical information. Our framework integrates genomic and transcriptomic data to deconvolve tumor-related and systemic tissue signals from plasma, supporting cfRNA signatures as a clinically relevant resource for patient stratification, discovery of novel drug targets and plasma-based biomarkers, and disease progression and treatment response monitoring, with prospects for better trial design and treatment optimization.