Purpose: New methods to accurately predict an individual tumor behavior are urgently required to improve the treatment of cancer. We previously found that promoter hypermethylation can be an accurate predictor of bladder cancer progression, but it is not cancer specific. Here, we investigate a panel of methylated loci in a prospectively collected cohort of bladder tumors to determine whether hypermethylation has a useful role in the management of patients with bladder cancer. Experimental Design: Quantitative methylation-specific PCR was done at 17 gene promoters, suspected to be associated with tumor progression, in 96 malignant and 30 normal urothelial samples. Statistical analysis and artificial intelligence techniques were used to interrogate the results. Results: Using log-rank analysis, five loci were associated with progression to more advanced disease (RASSF1a, E-cadherin, TNFSR25, EDNRB, and APC ; P < 0.05). Multivariate analysis revealed that the overall degree of methylation was more significantly associated with subsequent progression and death (Cox, P = 0.002) than tumor stage (Cox, P = 0.008). Neuro-fuzzy modeling confirmed that these five loci were those most associated with tumor progression. Epigenetic predictive models developed using artificial intelligence techniques identified the presence and timing of tumor progression with 97% specificity and 75% sensitivity. Conclusion: Promoter hypermethylation seems a reliable predictor of tumor progression in bladder cancer. It is associated with aggressive tumors and could be used to identify patients with either superficial disease requiring radical treatment or a low progression risk suitable for less intensive surveillance. Multicenter studies are warranted to validate this marker.

No abstract

No abstract

No abstract

SUMMARY The myoelectrical activity of human colon and rectum has been studied by three types of electrode in man-intraluminal (suction), serosal and cutaneous. The patterns obtained indicate a high degree of consistency between the methods and the value of surface electrodes is emphasized.Gradients along the large bowel of both frequency and percentage electrical activity have been observed and possible physiological roles are postulated for them. By correlating the features of regular electrical and corresponding regular motor waves an alteration in the myoelectrical pattern is observed in the region of the rectosigmoid junction.The main electrical wave form detected in the smooth muscle of the stomach and small bowel is known as the slow wave or basic electrical rhythm and is present all the time. Slow waves are thought to coordinate gastrointestinal motility and their characteristics have been described both in vivo and in vitro (Daniel, Wachter, Honour, and Bogoch, 1960;Bass, Code, and Lambert, 1961; Duthie, Brown, Robertson-Dunn, Kwong, Whittaker, and . The situation in the colon and rectum is more complex. In a previous study of the rectosigmoid region in man we described long periods of electrical silence, particularly in the lower sigmoid region, although in the lower rectum slow waves were more constant and regular (Taylor, Duthie, Smallwood, Brown, and Linkens, 1974). These observations agree with other work (Provenzale and Pisano, 1971). In addition two distinct rhythms of slow waves were recognized in this region: a predominant higher frequency one (6-10 c/m), and a less commonly observed lower frequency one (2.5-4 c/m) with a greater amplitude. We have now extended these initial studies to the more proximal large bowel. MethodsElectrical slow waves of the smooth muscle of the colon and rectum were recorded in 108 subjects with no known pathology in this area, using three types of electrodes-intraluminal, serosal and cutaneous.

No abstract

Compared to traditional regression statistics artificial intelligence methods appear to be accurate and more explorative for analyzing large data cohorts. Furthermore, they allow individualized prediction of disease behavior. Each artificial intelligence method has characteristics that make it suitable for different tasks. The lack of transparency of artificial neural networks hinders global scientific community acceptance of this method but this can be overcome by neuro-fuzzy modeling systems.

No abstract