Emerging artificial intelligence applications in gastroenterology: A review of the literature

Artificial intelligence (AI) allows machines to provide disruptive value in several industries and applications. Applications of AI techniques, specifically machine learning and more recently deep learning, are arising in gastroenterology. Computer-aided diagnosis for upper gastrointestinal endoscopy has growing attention for automated and accurate identification of dysplasia in Barrett’s esophagus, as well as for the detection of early gastric cancers (GCs), therefore preventing esophageal and gastric malignancies. Besides, convoluted neural network technology can accurately assess Helicobacter pylori (H. pylori) infection during standard endoscopy without the need for biopsies, thus, reducing gastric cancer risk. AI can potentially be applied during colonoscopy to automatically discover colorectal polyps and differentiate between neoplastic and non-neoplastic ones, with the possible ability to improve adenoma detection rate, which changes broadly among endoscopists performing screening colonoscopies. In addition, AI permits to establish the feasibility of curative endoscopic resection of large colonic lesions based on the pit pattern characteristics. The aim of this review is to analyze current evidence from the literature, supporting recent technologies of AI both in upper and lower gastrointestinal diseases, including Barrett's esophagus, GC, H. pylori infection, colonic polyps and colon cancer.

A different perspective on sofosbuvir-ledipasvir treatment of patients with HCV genotype 1b cirrhosis: The ital-c network study

The effectiveness of a 12-week course of sofosbuvir-ledipasvir in treatment-experienced HCV genotype 1b-infected patients with cirrhosis is still under debate. Our primary endpoint was to compare the sustained virological response at post-treatment week 12 (SVR12) of sofosbuvir-ledipasvir in combination with ribavirin for 12 weeks, and sofosbuvir-ledipasvir alone for 24 weeks. This was a prospective observational study that enrolled 424 (195 naive, 229 experienced; 164 treated for 12 weeks with Ribavirin and 260 with sofosbuvir-ledipasvir alone for 24 weeks) consecutive HCV genotype 1b-infected patients with cirrhosis. The SVR12 rates were 93.9% and 99.2% in patients treated for 12 and 24 weeks, respectively (P = .002). The baseline characteristics of patients treated for 12 weeks were significantly different from those treated for 24 weeks as regards their younger age (P = .002), prevalence of Child-Pugh class A (P = .002), lower MELD scores (P = .001) and smaller number of nonresponders (P = .04). The shorter treatment was significantly associated with a lower SVR12 in univariate and multivariate analyses (P = .007 and P = .008, respectively). The SVR rate was unaffected by age, gender, BMI, Child-Pugh class, MELD score or previous antiviral treatment. Patients receiving ribavirin experienced more episodes of ascites and headache but less recurrence of hepatocellular carcinoma (HCC), and were prescribed more diuretics and cardiopulmonary drugs. No patient discontinued treatment. The therapeutic regimen of sofosbuvir-ledipasvir plus ribavirin administered for 12 weeks was less effective than sofosbuvir-ledipasvir alone given for 24 weeks. At odds with European guidelines, the recommended 12-week treatment with sofosbuvir-ledipasvir alone might be suboptimal for this setting of patients.

Small intestinal bacterial overgrowth and celiac disease: A systematic review with pooled-data analysis

BACKGROUND

A link between small intestinal bacterial overgrowth (SIBO) and celiac disease (CD) has been hypothesized.

METHODS

Literature search was performed in main medical databases. Methods of analysis/inclusion criteria were based on Preferred Reporting Items for Systematic reviews and Meta-Analyses recommendations. The end-point was to estimate, by a pooled-data analysis, SIBO prevalence in CD. Proportions/percentages and their 95% confidence intervals (CI) were calculated by inverse variance method, whereas odd ratios (OR) and their 95% CI were estimated, where available, based on the Mantel-Haenszel method. Data were entered into the RevMan 5.3 software.

KEY RESULTS

Eleven articles fulfilled considered criteria. The pooled mean prevalence of SIBO in CD was 20% (95% CI of 10%-30%). In comparison to asymptomatic controls, CD was associated to higher risk of SIBO, with an OR of 10.52 (95% CI 2.69-41.21, P=.0007). Jejunal aspirate culture assessed SIBO prevalence of 11% (95% CI 3%-19%) in CD, whereas breath tests detected a higher value (23%, 95% CI 10%-37%). The pooled prevalence of SIBO in CD patients who were symptomatic despite a GFD was 28% (95% CI 10%-47%), higher than in asymptomatic celiac patients (pooled prevalence of 10%, with a 95% CI of 3%-16%), despite not statistically significant (P=.06). When GFD-unresponsive CD was defined only by clinical persistence of symptoms, the prevalence of SIBO was higher than in the case of villous atrophy association (31% vs 16% P=.33).

CONCLUSIONS

The heterogeneity of available studies may not support a relationship SIBO-CD. Nevertheless, SIBO could be more common in CD when symptoms do not improve after GFD.