Large multicenter validation of urine RNA profile for urothelial carcinoma detection and surveillance
Kathleen E. Mach, Zachary Kornberg, Eugene Shkolyar, Jin Long, Timothy J. Lee, Vinh La, Ihna Yoo, Gabriela Rodriguez, Alan E. Thong, Kris B. Prado, Jay B. Shah, John T. Leppert, Eila C. Skinner, Joseph C. Liao
Kathleen E. Mach, Zachary Kornberg, Eugene Shkolyar, Jin Long, Timothy J. Lee, Vinh La, Ihna Yoo, Gabriela Rodriguez, Alan E. Thong, Kris B. Prado, Jay B. Shah, John T. Leppert, Eila C. Skinner, Joseph C. Liao
View: Text | PDF
Clinical Research and Public Health Clinical Research Oncology

Large multicenter validation of urine RNA profile for urothelial carcinoma detection and surveillance

Abstract

BACKGROUND Current diagnosis and surveillance of bladder cancer rely on cystoscopy, which is invasive and user dependent. The urine mRNA panel, uRNAp, measures expression of 3 genes for identification of bladder cancer. Here we report validation of uRNAp for patients undergoing initial workup for suspected bladder cancer and surveillance for bladder cancer.METHODS Urine specimens were prospectively collected prior to cystoscopy at 2 health care systems from patients without (detection cohort) or with (surveillance cohort) a history of bladder cancer. RNA was isolated from urine sediment for RT-qPCR to determine roundabout guidance receptor 1, corticotropin releasing hormone, and insulin-like growth factor 2 expression and calculate the uRNAp bladder cancer probability score.RESULTS In the detection cohort, 547 samples were collected from 529 patients. There were 123 new diagnoses of bladder cancer in the detection cohort, and uRNAp demonstrated 98% sensitivity and 51% specificity for identification of bladder cancer. In the surveillance cohort, 1,543 samples were collected from 447 patients with 286 recurrences. uRNAp demonstrated 94% overall sensitivity with 43% specificity and 99% sensitivity for high-grade recurrence. The receiver operating characteristic area under the curve was 0.92 in the detection and 0.81 in the surveillance cohort. uRNAp scores significantly increased with tumor size and grade.CONCLUSION Prospective validation of uRNAp demonstrated a strong potential clinical utility as a noninvasive adjunct to cystoscopy for management of bladder cancer. uRNAp may be a useful triage tool to defer or expedite cystoscopy for patients undergoing detection or surveillance of bladder cancer.FUNDING Department of Veterans Affairs BLR&D Merit Review I01 BX004962 to JCL.

Authors

Kathleen E. Mach, Zachary Kornberg, Eugene Shkolyar, Jin Long, Timothy J. Lee, Vinh La, Ihna Yoo, Gabriela Rodriguez, Alan E. Thong, Kris B. Prado, Jay B. Shah, John T. Leppert, Eila C. Skinner, Joseph C. Liao

×

Figure 2

uRNAp score for UC detection and surveillance by tumor grade and aggregate size.

Options: View larger image (or click on image) Download as PowerPoint
uRNAp score for UC detection and surveillance by tumor grade and aggrega...
(A and B) Waterfall plots of uRNAp scores by tumor grade from patient samples without UC (No UC, beige) and with low-grade (LG; pink) or high-grade (HG; red) UC. A uRNAp score of 0.23 (horizontal dashed line) was the cutoff for UC diagnosis in (A) patients without a prior history of UC (detection cohort, n = 547 samples) and (B) patients with a history of bladder cancer (surveillance cohort, n = 1,543 samples). (C and D) uRNAp scores increase with increasing tumor size and grade for both detection and surveillance cohorts. P values for ANOVA comparisons are shown above brackets. Significant P values that are below the Bonferroni-adjusted value for multiple comparisons of 0.002 are marked with an asterisk. (C) In the detection cohort 11 tumors were defined as small (<1 cm; n = 7 LG; n = 4 HG), 26 were medium (between 1 and 3 cm; n = 13 LG; n = 13 HG), and 84 were large (>3 cm; n = 10 LG; n = 74 HG). Tumor size information was not available in 2 cases. (D) In the surveillance cohort 82 tumors were defined as small (<1 cm; n = 50 LG; n = 32 HG), 89 were medium (between 1 and 3 cm; n = 33 LG; n = 56 HG), and 77 were large (>3 cm; n = 14 LG; n = 63 HG). Tumor size information was not available in 38 cases. Box plots show the interquartile range, median (line), and minimum and maximum (whiskers).