Predicting disease outcome of non‐invasive transitional cell carcinoma of the urinary bladder using an artificial neural network model: Results of patient follow‐up for 15 years or longer

Abstract

Background : Patients with non-invasive (Ta/T1) transitional cell carcinoma (TCC) of the urinary bladder are often observed without progression in the long-term follow-up period, although many of them experience recurrence of disease. It is difficult to accurately predict the disease outcome of each patient with Ta/T1 TCC using conventional prognostic criteria. In this study, we examined the usefulness of artificial neural networks (ANNs) to predict the long-term disease outcome of patients with TCC of the urinary bladder. Methods : A retrospective, prognostic study of 90 patients with Ta/T1 TCC of the urinary bladder, diagnosed by transurethral resection of the bladder tumor between April 1981 and March 1985, and then followed up for 15 years or longer, was carried out. Data were analyzed using the Bayesian network tool of SPSS Neural Connection 2.1. The input neural data consisted of tumor stage, grade, tumor number, age, gender, tumor architecture and estimates of mean nuclear volume. The data set was randomly divided into 68 training and 22 testing examples for the prediction of disease progression and tumor recurrence within 15 years. Results : During 15 years follow-up, tumor recurrence was noted in 42/90 (47%) Ta/T1 tumors. The ANN model could not predict tumor recurrence. Conversely, disease progression was noted in 17/90 (19%) Ta/T1 tumors, and, in the test set, 4/22 (18%) Ta/T1 tumors underwent disease progression. The sensitivity of the ANN model to predict progression was 100% (specificity 67%; positive predictive value 40%; negative predictive value 100%). Patients who were judged to have a favorable prognosis using ANN analysis did not progress within the 15-year follow-up period. Conclusion : The results of the ANN study indicate that long-term progression-free survival of patients with non-invasive TCC of the urinary bladder can be precisely predicted. A favorable prognosis using ANNs would be one of the exclusion criteria for immediate or future total cystectomy.

The miR-143/-145 cluster regulates plasminogen activator inhibitor-1 in bladder cancer

“…Reduced expression of the miR-143/-145 cluster, as determined by miR-145 in situ hybridisation in T1 tumours, is also a predictor of poor prognosis (Ostenfeld et al , 2010). A critical obstacle in bladder cancer treatment is to characterise the carcinogenic potential of Ta and T1 tumours for which it is intrinsically difficult to predict disease outcome (Fujikawa et al , 2003). Thus, PAI-1 and miR-143/-145 deregulation have potential clinical significance for early diagnosis and treatment assessment of bladder cancer.…”

Prognostic accuracy of an artificial neural network in patients undergoing radical cystectomy for bladder cancer: a comparison with logistic regression analysis

“…Our results agree with those reported in other urological reports supporting the satisfactory diagnostic and prognostic performance of ANNs. Most trials have been focused on prostate cancer, but occasionally ANNs were applied to diagnostic and prognostic problems of kidney, testicle or bladder tumours [9,11–33]. In these studies ANNs were an optimal analytical statistical method to analyse complex problems, e.g.…”

“…However, when analysing the entire Ta/T1 cohort, there was no significant difference. In another report [32] an ANN was used to predict long‐term progression‐free survival (negative predictive value 100%) in patients with superficial bladder cancer followed for ≥15 years, but it could not predict tumour recurrence. Catto et al.…”

“…In oncological urology ANNs have been investigated specifically to resolve the diagnostic, staging and prognostic problems of prostate cancer, with very promising results [11–24]. Conversely, studies exploring the use of ANNs for kidney, testicle and bladder cancers are uncommon [25–33] (Table 1).…”

Cálculo del riesgo biológico de multirrecidiva y progresión del carcinoma urotelial no músculo-invasivo mediante nuevos modelos matemáticos

“…Nuestro objetivo es establecer el pronóstico y tratamiento una vez se realiza la RTU y se dispone del estudio anatomo-patológico que siempre es previo a la cistoscopia de los tres meses tras RTU. Otros modelos pronósticos al igual que nuestro estudio no incluyeron esta variable en su análisis [20,27,28]. Otra limitación viene determinada por la variable fragilidad del modelo de fragilidad conjunta.…”