The overall objective of this Phase I study is to establish the technological merit and feasibility of our proposed approach to developing and then commercializing a pair of companion prognostics for newly diagnosed non- muscle invasive Ta T1 urothelial carcinoma (UC) of the bladder that accurately predicts the relative chances of tumor recurrence and progression within 5 years after initial transurethral tumor resection (TURBT). By the end of Phase II, the prognostic classifiers created in Phase I will be externally validated using an independent blinded set of Ta T1 bladder UC tissues. Utilizing Genetics Squared's proprietary genetic programming (GP) analytic approach (Evolver(tm)) to define and quantify the interrelationships of programmatically selected tumor genes, we aim to create prognostic signatures that are more accurate and universal across the entire population of Ta T1 bladder UC than the best existent clinical/pathologic prognostic factors. Specifically, we aim to: 1. Demonstrate the successful extraction of sufficient amounts of intact RNA from the available archival frozen non-muscle invasive Ta T1 bladder UC tissue (n=178) for further evaluation. 2. Establish the feasibility of using the Evolver(tm) platform to analyze baseline human whole-genome expression data and integrate correlative patient-specific clinical and pathological data to unbiasely identify 'key tumor genes' that, when used together, best predict non-muscle invasive bladder UC time-to-first recurrence (TTR) and time-to-progression (TTP) within 5 years post-TURBT. 3. Generate candidate prognostic functions that best predict tumor recurrence and progression by Evolver(tm) analysis of the quantitative RT-PCR (qRT-PCR)-derived expression profiles of the 'key tumor genes' previously identified. 4. From amongst the candidate classifiers generated, select the most accurate predictor of non-muscle invasive bladder UC TTR and the most accurate predictor of TTP and confirm that both have statistically significant prognostic abilities. The entire project will be conducted in collaboration with Dr. Richard Cote's laboratory in the Department of Pathology, University of Southern California, Los Angeles, CA.
The need for this test emanates from the difficult decisions that bladder cancer patients and their clinicians must make that affect their survival, health and quality of life. One of the more difficult decisions after surgery is whether or not to undergo unpleasant and potentially toxic chemotherapy or even cystectomy after the tumor is removed at an early stage of development. Making a decision to avoid these procedures could dramatically improve a patient's quality of life. However, for certain individuals, that decision could drastically shorten their life-span. The key benefit of this diagnostic then would be to reduce the number of patients undergoing unnecessary chemotherapy and/or cystectomy and identify those patients with tumors that are highly likely to progress and would need more aggressive treatment and/or follow-up. ? ? ?
| Bartsch Jr, Georg; Mitra, Anirban P; Mitra, Sheetal A et al. (2016) Use of Artificial Intelligence and Machine Learning Algorithms with Gene Expression Profiling to Predict Recurrent Nonmuscle Invasive Urothelial Carcinoma of the Bladder. J Urol 195:493-8 |
