Presentation Authors: Aaron Berkenwald*, Erik Katz, Brendan Brown, Chintan Patel, Travis Sullivan, Burlington, MA, Erik Burks, Boston, MA, Jay Raman, Joshua Warrick, Hershey, PA, David Canes, Kimberly Rieger-Christ, Burlington, MA
Introduction: Radical Nephroureterectomy (RNU) is the gold standard treatment for Upper Tract Urothelial Carcinoma (UTUC). However, less invasive treatment modalities exist for low grade, non-invasive tumors. Determination of tumor characteristics are currently based on endoscopic biopsies, which often supply insufficient tissue for accurate diagnosis. Biochemical analysis of UTUC biopsies may enable practitioners to make more informed clinical decisions and avoid overtreating less aggressive tumors. We propose that by comparing microRNA (miRNA) expression patterns from UTUC biopsies to their final pathology from RNU samples we may provide a framework for more effective diagnosis, and possibly help predict tumor behavior.
Methods: Under an IRB-approved study, total RNA was extracted from formalin-fixed, paraffin-embedded UTUC biopsy samples from 64 patients who subsequently underwent RNU from 2005-2018 at two high-volume institutions. Twenty samples were profiled via miRNA RT-qPCR and screened with an array for 752 unique miRNAs. Differentially expressed miRNAs were then selected for validation using 44 UTUC biopsy samples. In total, 28 high grade (HG) and 16 low grade (LG) samples were analyzed using ROC curves and logistic regression analysis to establish a predictive miRNA model corresponding to final pathological grade and stage after RNU.
Results: Screening array analysis identified 26 miRNAs differentially expressed between LG and HG tumors (p < 0.05 and FDR < 0.1). Of these, four were up-regulated and 22 were down-regulated in the HG, invasive tumors. Hierarchical clustering analysis yielded two distinct groups with miRNA expression patterns corresponding to final RNU pathology (p=0.029). Validation of these miRNAs revealed upregulation of miR-21-5p in invasive tumors (p=0.026) and downregulation of four miRNAs (miR-141-3b, miR-26b-5p, miR-29c-3p, let-7b-5p) in HG tumors (p < 0.05). Using miR-141-3p and miR-29c-3p in combination as a predictive model for grade resulted in a sensitivity and specificity of 89.3% and 62.5% respectively, with an area under the curve of 0.812.
Conclusions: We present distinct miRNA expression profiles of UTUC biopsies that are associated with HG, invasive tumors compared to LG, non-invasive lesions. We further highlight validated UTUC miRNAs that show a statistically significant correlation with tumor invasion and grade.