Comparison of Gene Enrichment between Prostate Tissue and Urine: Epithelial and Stromal Signature Genes And their Implications to Urine Biomarker Development

Thumbnail Image
Xing, Yuandong
prostate cancer , epithelium , stroma , urine , biomarker , expression data alignment , gene enrichment comparison
Background: For prostate cancer (PCa), early diagnosis and patient stratification based on biomarkers dramatically improve prognosis after treatment. PCa biomarkers are best credentialled in tissue after comprehensive histopathologic review. However, urinary biomarkers are obtainable non-invasively and thus safter to obtain. It is warranted to understand how to develop urinary biomarkers based on gene expression in tissue. As a critical step, comparison of gene enrichment in prostate tissue versus urine may help in selecting transcripts highly enriched in urine as biomarker candidates. Here, we focus on prostate epithelial and stromal signature genes for gene set enrichment analysis. Methods: Epithelial and stromal genes were identified, filtered, and validated based on public data from Human Protein Atlas (HPA) and Gene Expression Omnibus (GEO). Next, we profiled 367 transcripts in prostate tissue (n=9) and urine (n=9) samples from patients undergoing cancer surgery. We developed a Tissue-Urine Comparability Index (TUCI) to evaluate the efficacy of different data processing methods in aligning gene expression profiles from tissue and urine. Processed by the best method, tissue and urine data were employed to compare the enrichment of epithelial and stromal gene sets, as well as gene sets available in the Molecular Signature Database (MSigDB). Results: Epithelial and stromal gene sets, including nineteen and thirteen genes respectively, reliably represented epithelial and stromal signatures and had potential diagnostic value. Correcting batch effect followed by normalizing to GNAS and TAF6L transcripts achieved the highest (best) TUCI. Both epithelial and stromal gene sets were significantly enriched in tissue compared to urine, while the stromal gene set was enriched to a larger extent in tissue than epithelial gene set. Moreover, compared to tissue, immune gene sets were more enriched in urine. Tissue-enriched gene sets include gene sets targeted by miRNA (e.g., miR-148/152 family, miR-302 family, miR-520 family, and miR-372/373 cluster), gene sets targeted by transcription factors (e.g., STAT5), and gene sets involved in biological processes (e.g., cell growth and organ development). Conclusion: Prostate tissue-enriched gene sets differed from urine-enriched ones. These differences provide new insights into prostate tissue and urine contents and direct future work on developing urine based PCa biomarkers.
External DOI