Background and Aims: Recently, there have been some reports that image-enhanced endoscopy may improve detection of gastric intestinal metaplasia (GIM). Our aim was to determine the usefulness of linked color imaging (LCI) for detection of GIM. Methods: In prospectively recruited patients, the whole antrum was observed by white light imaging (WLI) followed by LCI. When a whitish flat elevation (WFE) in WLI and a lavender color sign (LCS) in LCI were detected, target biopsies were performed after LCI. Random biopsies were performed in patients who had neither WFE nor LCS. The primary endpoint was the diagnostic accuracy of GIM per patient in WLI and LCI and the secondary endpoints were that of GIM per biopsy and interobserver agreement. Results: Data from 128 patients were analyzed and 58 patients (45.3%) had histological GIM in the antrum. The per-patient yields of WLI and LCI to detect GIM were 19.0% (11/58) and 91.4% (53/58) respectively. Diagnostic accuracies of target biopsies for GIM were 23.7% in WLI and 84.2% in LCI. Kappa values among 3 doctors were 0.60 for WFE and 0.78 for LCS respectively. Conclusion: LCI could be a new diagnostic tool for detecting GIM during routine endoscopy.

Introduction: Atezolizumab + bevacizumab showed survival benefit in patients with unresectable hepatocellular carcinoma (HCC) versus sorafenib in the Phase III IMbrave150 study. This exploratory analysis examined the prognostic impact of a baseline albumin-bilirubin (ALBI) score. Methods: Patients with treatment-naïve unresectable HCC, ≥1 measurable untreated lesion, and Child-Pugh class A liver function were randomized 2:1 to receive atezolizumab 1,200 mg + bevacizumab 15 mg/kg every 3 weeks or sorafenib 400 mg twice daily. Overall survival (OS) and progression-free survival (PFS) were assessed in the intention-to-treat population by ALBI/modified (m)ALBI grade. Time to deterioration (TTD; defined as time to 0.5-point increase from the baseline ALBI score over 2 visits or death) of liver function and safety were investigated. Results: Of 501 enrolled patients, 336 were randomized to receive atezolizumab + bevacizumab (ALBI grade [G] 1: n = 191; G2: n = 144 [mALBI G2a: n = 72, G2b: n = 72]; missing ALBI grade: n = 1) and 165 to sorafenib (ALBI G1: n = 87; G2: n = 78 [mALBI G2a: n = 37; G2b: n = 41]). Median follow-up was 15.6 months. OS and PFS improved with atezolizumab + bevacizumab versus sorafenib in patients with ALBI G1 (OS HR: 0.50 [95% CI: 0.35, 0.72]; PFS HR: 0.61 [95% CI: 0.45, 0.82]). In patients with ALBI G2 or mALBI G2a or G2b, PFS was numerically longer with atezolizumab + bevacizumab versus sorafenib, but no OS benefit was seen. Median TTD in the intention-to-treat population was 10.2 months (95% CI: 8.0, 11.0) with atezolizumab + bevacizumab versus 8.6 months (95% CI: 6.2, 11.8) with sorafenib (HR: 0.82 [95% CI: 0.65, 1.03]). Safety profiles of atezolizumab and bevacizumab were consistent with previous analyses, regardless of ALBI grade. Conclusion: ALBI grade appeared to be prognostic for outcomes with both atezolizumab + bevacizumab and sorafenib treatment in patients with HCC. Atezolizumab + bevacizumab preserved liver function for a numerically longer duration than sorafenib.

Background: Ultrasonography (US) is widely used for the diagnosis of liver tumors. However, the accuracy of the diagnosis largely depends on the visual perception of humans. Hence, we aimed to construct artificial intelligence (AI) models for the diagnosis of liver tumors in US. Methods: We constructed three AI models based on still B-mode images: model-1 using 24,675 images, model-2 using 57,145 images, and model-3 using 70,950 images. A convolutional neural network was used to train the US images. The four-class liver tumor discrimination by AI, namely, cysts, hemangiomas, hepatocellular carcinoma, and metastatic tumors, was examined. The accuracy of the AI diagnosis was evaluated using tenfold cross-validation. The diagnostic performances of the AI models and human experts were also compared using an independent test cohort of video images. Results: The diagnostic accuracies of model-1, model-2, and model-3 in the four tumor types are 86.8%, 91.0%, and 91.1%, whereas those for malignant tumor are 91.3%, 94.3%, and 94.3%, respectively. In the independent comparison of the AIs and physicians, the percentages of correct diagnoses (accuracies) by the AIs are 80.0%, 81.8%, and 89.1% in model-1, model-2, and model-3, respectively. Meanwhile, the median percentages of correct diagnoses are 67.3% (range 63.6%–69.1%) and 47.3% (45.5%–47.3%) by human experts and non-experts, respectively. Conclusion: The performance of the AI models surpassed that of human experts in the four-class discrimination and benign and malignant discrimination of liver tumors. Thus, the AI models can help prevent human errors in US diagnosis.

Introduction: The diagnosis of Helicobacter pylori infection status with white light imaging (WLI) is difficult. We evaluated the accuracies of using WLI and linked color imaging (LCI) for diagnosing H. pylori-active gastritis in a multicenter prospective study setting. Methods: Patients who underwent esophagogastroduodenoscopy were prospectively included. The image collection process was randomized and anonymous, and the image set included 4 images with WLI or 4 images with LCI in the corpus that 5 reviewers separately evaluated. Active gastritis was defined as positive when there was diffuse redness in WLI and crimson coloring in LCI. The H. pylori infection status was determined by the urea breath test and the serum antibody test. Cases in which both test results were negative but atrophy or intestinal metaplasia was histologically confirmed were defined as past infections. The primary endpoint was the diagnostic accuracies of WLI and LCI, and the secondary endpoint was inter-observer agreement. Results: Data for 127 patients were analyzed. The endoscopic diagnostic accuracy for active gastritis was 79.5 (sensitivity of 84.4 and specificity of 74.6) with WLI and 86.6 (sensitivity of 84.4 and specificity of 88.9) with LCI (p = 0.029). LCI significantly improved the accuracy in patients with past infections over WLI (36.8 in WLI and 78.9 in LCI, p < 0.01). The κ values were 0.59 in WLI and 0.70 in LCI. Conclusions: LCI is useful for endoscopic diagnosis of H. pylori-active or inactive gastritis, and it is advantageous for patients with past infections of inactive gastritis.