Adaptive genetic algorithms combined with high sensitivity single cell-based technology derived urine-based score to differentiate between high-grade and low-grade transitional cell carcinoma of the bladder.

572 Background: We previously showed that adaptive genetic algorithms (AGA), in combination with single-cell flow cytometry technology, can be used to develop a noninvasive urine-based score to detect bladder cancer with high accuracy. Our aim in this analysis was to investigate if that same score can differentiate between high grade (HG) and low grade (LG) transitional cell carcinoma of the bladder (BC). Methods: We collected urine samples from cystoscopy confirmed HG and LG superficial bladder cancer patients and healthy donors in an optimized urine collection media. We then examined these samples using an assay developed from AGA in combination with single-cell flow cytometry technology. Results: We examined 50 BC and 15 healthy donor urine samples. Patients were majorly White (59.2%), males (61.2%), and had HG BC (66.7%). AGA derived score of 1.1 differentiated between BCa and healthy patients with high precision (AUC 0.92). The median score was 2.8 for LG BC and 6 for LG BC. Mann-Whitney Rank Sum Test indicated that the difference between the median score of HG and LG BC was significant at P value = 0.003. The score performed well independent of patients’ sex or smoking history. Conclusions: Using single-cell technology and machine learning, we developed a new urine-based score that can potentially differentiate between HG and LG bladder cancer. Future studies are planned to validate this score.