Use of Artificial Intelligence and Machine Learning Algorithms with Gene Expression Profiling to Predict Recurrent Nonmuscle Invasive Urothelial Carcinoma of the Bladder.

PURPOSE: Due to the high recurrence risk of nonmuscle invasive urothelial carcinoma it is crucial to distinguish patients at high risk from those with indolent disease. In this study we used a machine learning algorithm to identify the genes in patients with nonmuscle invasive urothelial carcinoma at initial presentation that were most predictive of recurrence. We used the genes in a molecular signature to predict recurrence risk within 5 years after transurethral resection of bladder tumor. MATERIALS AND METHODS: Whole genome profiling was performed on 112 frozen nonmuscle invasive urothelial carcinoma specimens obtained at first presentation on Human WG-6 BeadChips (Illumina®). A genetic programming algorithm was applied to evolve classifier mathematical models for outcome prediction. Cross-validation based resampling and gene use frequencies were used to identify the most prognostic genes, which were combined into rules used in a voting algorithm to predict the sample target class. Key genes were validated by quantitative polymerase chain reaction. RESULTS: The classifier set included 21 genes that predicted recurrence. Quantitative polymerase chain reaction was done for these genes in a subset of 100 patients. A 5-gene combined rule incorporating a voting algorithm yielded 77% sensitivity and 85% specificity to predict recurrence in the training set, and 69% and 62%, respectively, in the test set. A singular 3-gene rule was constructed that predicted recurrence with 80% sensitivity and 90% specificity in the training set, and 71% and 67%, respectively, in the test set. CONCLUSIONS: Using primary nonmuscle invasive urothelial carcinoma from initial occurrences genetic programming identified transcripts in reproducible fashion, which were predictive of recurrence. These findings could potentially impact nonmuscle invasive urothelial carcinoma management.

A Prospective Blinded Evaluation of Urine-DNA Testing for Detection of Urothelial Bladder Carcinoma in Patients with Gross Hematuria.

Retrospective studies have provided proof of principle that bladder cancer can be detected by testing for the presence of tumor DNA in urine. We have conducted a prospective blinded study to determine whether a urine-based DNA test can replace flexible cystoscopy in the initial assessment of gross hematuria. A total of 475 consecutive patients underwent standard urological examination including flexible cystoscopy and computed tomography urography, and provided urine samples immediately before (n=461) and after (n=444) cystoscopy. Urine cells were collected using a filtration device and tested for eight DNA mutation and methylation biomarkers. Clinical evaluation identified 99 (20.8%) patients with urothelial bladder tumors. With this result as a reference and based on the analysis of all urine samples, the DNA test had a sensitivity of 97.0%, a specificity of 76.9%, a positive predictive value of 52.5%, and a negative predictive value of 99.0%. In three patients with a positive urine-DNA test without clinical evidence of cancer, a tumor was detected at repeat cystoscopy within 16 mo. Our results suggest that urine-DNA testing can be used to identify a large subgroup of patients with gross hematuria in whom cystoscopy is not required. PATIENT SUMMARY: We tested the possibility of using a urine-based DNA test to check for bladder cancer in patients with visible blood in the urine. Our results show that the test efficiently detects bladder cancer and therefore may be used to greatly reduce the number of patients who would need to undergo cystoscopy.

Predicting recurrence and progression of noninvasive papillary bladder cancer at initial presentation based on quantitative gene expression profiles.

BACKGROUND: Currently, tumor grade is the best predictor of outcome at first presentation of noninvasive papillary (Ta) bladder cancer. However, reliable predictors of Ta tumor recurrence and progression for individual patients, which could optimize treatment and follow-up schedules based on specific tumor biology, are yet to be identified. OBJECTIVE: To identify genes predictive for recurrence and progression in Ta bladder cancer at first presentation using a quantitative, pathway-specific approach. DESIGN, SETTING, AND PARTICIPANTS: Retrospective study of patients with Ta G2/3 bladder tumors at initial presentation with three distinct clinical outcomes: absence of recurrence (n=16), recurrence without progression (n=16), and progression to carcinoma in situ or invasive disease (n=16). MEASUREMENTS: Expressions of 24 genes that feature in relevant pathways that are deregulated in bladder cancer were quantified by real-time polymerase chain reaction on tumor biopsies from the patients at initial presentation. RESULTS AND LIMITATIONS: CCND3 (p=0.003) and HRAS (p=0.01) were predictive for recurrence by univariate analysis. In a multivariable model based on CCND3 expression, sensitivity and specificity for recurrence were 97% and 63%, respectively. HRAS (p<0.001), E2F1 (p=0.017), BIRC5/Survivin (p=0.038), and VEGFR2 (p=0.047) were predictive for progression by univariate analysis. Multivariable analysis based on HRAS, VEGFR2, and VEGF identified progression with 81% sensitivity and 94% specificity. Since this is a small retrospective study using medium-throughput profiling, larger confirmatory studies are needed. CONCLUSIONS: Gene expression profiling across relevant cancer pathways appears to be a promising approach for Ta bladder tumor outcome prediction at initial diagnosis. These results could help differentiate between patients who need aggressive versus expectant management.