- This study was the first blinded clinical validation of UroAmp (Convergent Genomics), a genomic assay for the initial diagnosis of urothelial carcinoma (UC), prediction of recurrence and risk stratification
- UroAmp had high sensitivity and specificity (95% and 90%, respectively) in the initial detection of UC, including 100% sensitivity for high-grade and muscle-invasive tumors and 99% negative predictive value in patients with hematuria
- In the UC surveillance setting, UroAmp significantly predicted future recurrence with 89% specificity and a negative predictive value of 91%
- UroAmp provides a molecular grade assessment, and in this study, predictions of high-grade UC were confirmed by pathology 88% of the time
- UroAmp appears to have the potential to serve as both a "rule-out" and "rule-in" test for UC
Guidelines recommend transurethral white-light cystoscopy for diagnosis and surveillance of urothelial carcinoma (UC). However, adherence to this standard of care is poor. Cystoscopy is an invasive test, and beyond that, relatively few adults with hematuria prove to have UC, contributing to many patients never being referred to urologists for evaluation. A reliable noninvasive test could reduce the burden of evaluation and improve adherence to guidelines.
Subscribe to the latest updates from Urology Advances in Motion
Researchers at Massachusetts General Hospital recently collaborated with Convergent Genomics and helped validate UroAmp (Convergent Genomics)*, a urine-based test leveraging comprehensive genomic profiling that can accurately diagnose UC and predict the risk of recurrence.
Keyan Salari, MD, PhD, director of translational research and urologic surgeon in the Department of Urology, Adam S. Feldman, MD, MPH, associate chair for research and a urologic oncologist in the Department, Convergent Genomics scientists, and colleagues report specifics in Clinical Cancer Research.
UroAmp uses next-generation DNA sequencing to analyze 60 high-impact UC-associated genes for mutations while broadly measuring chromosomal changes across the whole genome. Urinary comprehensive genomic profiles are generated and serve as inputs to prediction algorithms.
At 10 urology clinics, the researchers performed a case–control study of UroAmp using urine specimens from 581 patients: 333 for disease classification and algorithm development and 248 for blinded validation.
The algorithm trained for initial diagnosis optimized specificity while maintaining the high sensitivity needed for screening. The cases were 78 patients who underwent cystoscopy leading to a de novo UC diagnosis, and the controls were 235 patients who had benign urologic conditions.
In the blinded validation cohort, the initial diagnosis algorithm demonstrated:
- Overall—95% sensitivity and 90% specificity; 99% negative predictive value in patients with hematuria
- High-grade and muscle-invasive tumors—100% sensitivity
- Non–muscle-invasive tumors—94% sensitivity
- Low-grade disease—87% sensitivity
A further analysis considered levels of disease prevalence across risk categories specified by the American Urological Association. For low- and intermediate-risk microhematuria, where the natural prevalence of UC is 0.5% and 1%, respectively, a negative result decreased the likelihood of UC to 0.03% and 0.05%, demonstrating the value of UroAmp as a rule-out test.
The algorithm for predicting recurrence during surveillance optimized specificity and positive predictive value. The cases were 86 patients who underwent cystoscopy leading to a diagnosis of recurrent UC and the 182 controls had prior definitive surgical treatment of UC, negative cystoscopy within one month of urine collection, and a second negative cystoscopy three to 12 months following collection.
The results of the predictive algorithm in the validation cohort were:
- Overall—89% specificity, 65% sensitivity
- High-grade disease—76% sensitivity
- Low-grade disease—58% sensitivity
- Overall negative predictive value—91%
- Overall positive predictive value—59%
A third algorithm was developed to predict the presence of high-grade UC, optimized for specificity. In the validation cohort, molecular prediction of high-grade disease achieved:
- Positive predictive value—88% (cases confirmed by pathology)
- Odds ratio—15.2 (P<0.0001)
In the aggregated set of 164 tumor-positive patients, UroAmp identified genomic features associated with grade and stage.
Toward the Future
Genomic profiles have the potential to identify molecular subtypes of cancer and predict response to FDA-approved therapies. The pattern recognition algorithms in UroAmp could learn to prioritize certain genes above others and discover complex interactions of genes in tumor development. Ultimately, the algorithms may identify previously unknown tumor subtypes and their association with prognosis or response to therapies.
*Mass General received research support from Convergent Genomics for this study.
Learn about research in the Department of Urology
Refer a patient to the Department of Urology