Presentation Authors: Takahiro Osawa*, Hiroshi Harada, Keita Minami, Nobuo Shinohara, Sapporo, Japan, Taku Murakami, Cindy M. Yamamoto, Irvine, CA, Hiroshi Tanaka, Toshimori Seki, Sapporo, Japan
Introduction: High-risk non-muscle invasive bladder cancer (HR-NMIBC) (high grade Ta, T1, and any Tis in NCCN category) patients are at higher risk of both recurrence and progression compared to low-risk (LR) NMIBC (low grade Ta). Cystoscopy itself does not allow us to distinguish HR- and LR-NMIBC, therefore pathological analyses of tumors are crucial. The objective of this study is to develop urine based molecular diagnostics to identify and distinguish HR- and LR-NMIBC preoperatively.
Methods: We conducted a marker discovery using urinary extracellular vesicles (EV) from bladder cancer patients. Candidate markers were identified through RNA-seq analysis of 12 urinary EV from urothelial carcinoma patients, those with a history of urothelial carcinoma and healthy controls. Diagnostic performance of these genes was obtained in a larger cohort of 254 urine samples from urothelial carcinoma patients. EV mRNA were isolated using ExoComplete (Hitachi Chemical Diagnostics, Inc., CA) following the manufacturer's protocol. Statistical analysis was conducted by R and sparse logistic regression analysis (SLR) was employed to identify the coefficients for each gene and cytology and develop scoring formulas for HR- and LR-NMIBC detection.
Results: Among the differentially expressed genes in urinary EV through RNA-seq, we confirmed that SLC2A1 (basal type), GPRC5A (luminal type) and KRT17 (basal type) were elevated in HR-NMIBC (N=74) compared to LR-NMIBC (N=106) by 5.1-fold, 8.6-fold and 4.2-fold, respectively. Indeed, diagnostic performance of these markers was better for HR-NMIBC (AUC 0.739 to 0.802) than those for LR-NMIBC (AUC 0.55 to 0.625). These markers may be useful as prognostic markers since overexpression of these genes in tumors indicated poor prognosis in several types of cancer including renal, liver, pancreatic, lung, and urothelial cancer in the Cancer Genome Atlas (TCGA). Sparse logistic regression analysis through 10-fold cross validation with 50,000 bootstrap re-sampling allowed us to develop scoring formulas comprised of the expression levels of EV SLC2A1, GPRC5A and KRT17 as well as urine cytology with better diagnostic performance (AUC 0.883 for HR-NMIBC, 0.635 for LR-NMIBC) compared to conventional urine cytology alone (AUC 0.785 for HR-NMIBC, 0.504 for LR NMIBC). In addition, this combination was also applicable to detect muscle invasive bladder cancer (MIBC) with AUC 0.830.
Conclusions: Taken together, our findings suggest that these three urinary EV mRNA were promising biomarkers for risk stratification of NMIBC as well as MIBC.
Source of Funding: TM and CMY are the employees of Hitachi Chemical Co. America, Ltd.