Identification of novel snoRNA-based biomarkers for clear cell renal cell carcinoma from urine-derived extracellular vesicles.

Abstract

Background: Clear cell renal cell carcinoma (ccRCC) is the most common subtype of RCC with high rates of metastasis. Targeted therapies such as tyrosine kinase and checkpoint inhibitors have improved treatment success, but therapy-related side effects and tumor recurrence remain a challenge. As a result, ccRCC still have a high mortality rate. Early detection before metastasis has great potential to improve outcomes, but no suitable biomarker specific for ccRCC is available so far. Therefore, molecular biomarkers derived from body fluids have been investigated over the past decade. Among them, RNAs from urine-derived extracellular vesicles (EVs) are very promising.

Methods: RNA was extracted from urine-derived EVs from a cohort of 78 subjects (54 ccRCC patients, 24 urolithiasis controls). RNA-seq was performed on the discovery cohort, a subset of the whole cohort (47 ccRCC, 16 urolithiasis). Reads were then mapped to the genome, and expression was quantified based on 100 nt long contiguous genomic regions. Cluster analysis and differential region expression analysis were performed with adjustment for age and gender. The candidate biomarkers were validated by qPCR in the entire cohort. Receiver operating characteristic, area under the curve and odds ratios were used to evaluate the diagnostic potential of the models.

Results: An initial cluster analysis of RNA-seq expression data showed separation by the subjects' gender, but not by tumor status. Therefore, the following analyses were done, adjusting for gender and age. The regions differentially expressed between ccRCC and urolithiasis patients mainly overlapped with small nucleolar RNAs (snoRNAs). The differential expression of four snoRNAs (SNORD99, SNORD22, SNORD26, SNORA50C) was validated by quantitative PCR. Confounder-adjusted regression models were then used to classify the validation cohort into ccRCC and tumor-free subjects. Corresponding accuracies ranged from 0.654 to 0.744. Models combining multiple genes and the risk factors obesity and hypertension showed improved diagnostic performance with an accuracy of up to 0.811 for SNORD99 and SNORA50C (p = 0.0091).

Conclusions: Our study uncovered four previously unrecognized snoRNA biomarkers from urine-derived EVs, advancing the search for a robust, easy-to-use ccRCC screening method.