Session: MP22: Bladder Cancer: Epidemiology & Evaluation I
MP22-19: Urinary comprehensive genomic profiling predicts urothelial cancer up to 12 years ahead of clinical diagnosis. An expanded analysis of the Golestan Cohort Study
Introduction: Detecting pre-clinical urothelial carcinoma (UC) using urinary comprehensive genomic profiling (uCGP) may provide a valuable opportunity for early detection and screening of high-risk populations. The UroAmp (Convergent Genomics) uCGP test uses DNA sequencing to identify mutations across 60 genes. We study a subset of these genes, considering 10 genes with the highest performance, testing the potential of a screening modified uCGP (muCGP) to detect preclinical UC. Methods: A UC screening model was developed using muCGP data from a training cohort consisting of 140 urology controls & 96 tumors (56 de novo, 40 recurrent). Model validation was performed in two studies: first a multi-institutional case-control design with 96 controls & 70 UC cases (22 de novo, 48 surveillance); a second using a nested case-control design within the prospective Golestan Cohort Study (50,045 participants). The nested cohort consisted of 29 asymptomatic individuals who subsequently developed primary UC (median time to UC 7.3 yrs) and 98 matched controls (median f/u 6.1 yrs). Results: The UC screening model was trained to a sensitivity of 88% (97% sensitivity for HG) and specificity of 94%. In the first validation, a sensitivity of 86% in de novo (87% for HG), 71% overall (de novo + recurrent tumors), and specificity 94% was observed. In the Golestan cohort, baseline muCGP had a prediction sensitivity of 66% (71% for HG) and specificity 94% was observed (Figure 1A). In contrast baseline TERT predicted 48% percent of cancers with a specificity of 100%. Cancer-free survival was significantly worse in muGCP-predicted positives vs. muGCP-predicted negatives (HR 8.5, 95% CI 3.8 – 18.4, p<0.0001). When limited to UC diagnosis within five years, UroAmp detected pre-clinical UC in 90% of future cancers (Fig. 1B), compared to a sensitivity of 57% using TERT mutations alone. Conclusions: Our results provide the first evidence from a population-based prospective cohort study of pre-clinical UC detection with muCGP. muCGP identifies 9 of 10 cancers that occur within the first 5 years while detecting 10 of 19 cancers that occur beyond 5 years. Further studies will determine the frequency of muCGP screening required to maximize cancer detection and refine clinical interventions to save lives. SOURCE OF Funding: Convergent Genomics with a grant from the National Cancer Institute Small Business Innovation Research Program, 5R44CA200174-05
