Previous reports have shown favorable performance of artificial intelligence (AI) systems for diagnosing esophageal squamous cell carcinoma (ESCC) compared with endoscopists. However, these findings don’t reflect performance in clinical situations, as endoscopists classify lesions based on both magnified and non-magnified videos, while AI systems often use only a few magnified narrow band imaging (NBI) still images. We evaluated the performance of the AI system in simulated clinical situations. We used 25,048 images from 1433 superficial ESCC and 4746 images from 410 noncancerous esophagi to construct our AI system. For the validation dataset, we took NBI videos of suspected superficial ESCCs. The AI system diagnosis used one magnified still image taken from each video, while 19 endoscopists used whole videos. We used 147 videos and still images including 83 superficial ESCC and 64 non-ESCC lesions. The accuracy, sensitivity and specificity for the classification of ESCC were, respectively, 80.9% [95% CI 73.6–87.0], 85.5% [76.1–92.3], and 75.0% [62.6–85.0] for the AI system and 69.2% [66.4–72.1], 67.5% [61.4–73.6], and 71.5% [61.9–81.0] for the endoscopists. The AI system correctly classified all ESCCs invading the muscularis mucosa or submucosa and 96.8% of lesions ≥ 20 mm, whereas even the experts diagnosed some of them as non-ESCCs. Our AI system showed higher accuracy for classifying ESCC and non-ESCC than endoscopists. It may provide valuable diagnostic support to endoscopists.

No abstract

Objectives Endoscopic resection (ER) is a minimally invasive treatment for esophageal squamous cell carcinoma (ESCC). However, stricture may develop after ER for widespread lesions. Application of ER is justified if these cancers are pathological T1a‐epithelial/lamina propria (pEP/LPM) cancers that can be cured by ER. We conducted a study to clarify the association between pathological invasion depth and lesion size or circumference in clinical (c) EP/LPM cancers. Methods From our database, we identified patients diagnosed with cEP/LPM ESCC via endoscopic examination who underwent endoscopic or surgical tumor resection. The accuracy of the cEP/LPM ESCC diagnosis was determined by histologically diagnosing cancer invasion depth as a reference standard. Results Between January 2015 and December 2019, 1271 cancer patients were diagnosed with cEP/LPM ESCC, of which 1195 (94.0%) were correctly diagnosed with pEP/LPM cancer. The positive predictive value (PPV) classified according to lesion sizes of ≤25, 26–49, and ≥50 mm was 95.8% (981/1024 lesions), 89.7% (191/213 lesions), and 67.6% (23/34 lesions), respectively. PPV according to the circumferential extent of <3/4, ≥3/4, and <1, and whole was 94.6% (1164/1230 lesions), 75.0% (24/32 lesions), and 77.8% (7/9 lesions), respectively. In multivariate analysis, the PPV of cEP/LPM ESCC was significantly associated with lesion size (P < 0.001) and male sex. Conclusions Between January 2015 and December 2019, 1271 cancer patients were diagnosed with cEP/LPM ESCC, of which 1195 (94.0%) were correctly diagnosed with pEP/LPM cancer. The PPV of cEP/LPM ESCC was related to lesion size. Treatment should be determined considering the high risk of cancer invasion into the muscularis mucosa or deeper in cEP/LPM cancers with a lesion size of ≥50 mm.

Organic anion transporting polypeptide 2B1 (OATP2B1, SLCO2B1) is an uptake transporter expressed in several tissues, including the liver, intestine, brain, kidney, and skeletal muscle. Hepatocyte nuclear factor 4 alpha (HNF4α) is known as an important transcriptional factor of OATP2B1 in the liver. It has been reported that there are large interindividual differences in OATP2B1 mRNA and protein expressions in human livers. The mechanism causing the interindividual differences in OATP2B1 expression is still unclear. MicroRNAs (miRNAs) control gene expression by leading translational repression and/or degradation of the target mRNA. There is no significant correlation between OATP2B1 mRNA and protein expression, suggesting that post-transcriptional regulating mechanisms, such as miRNAs, play an important role in the interindividual differences in OATP2B1 expression. In this study, we hypothesized that certain miRNAs cause the interindividual differences in OATP2B1 expression in the human liver. In silico analysis showed that miR-24 was a candidate miRNA regulating OATP2B1 expression. It has been reported that miR-24 degrades HNF4α mRNA expression. We revealed that the miR-24 expression level was negatively correlated with OATP2B1 mRNA, protein, and HNF4α mRNA expression levels in human livers. Transfection by the miR-24 precursor decreased the luciferase activity in the transfected cells with the vector containing the OATP2B1 3′ untranslated region (3′UTR) or SLCO2B1 promoter region. In HepaRG cells, miR-24 decreased the OATP2B1 and HNF4α expression levels. These results suggest that miR-24 represses not only the translation of OATP2B1 but also the transcription of OATP2B1 by HNF4α mRNA degradation.

Background:Previous reports have shown favorable performance of artificial intelligence (AI) systems for diagnosing esophageal squamous cell carcinoma (ESCC) compared with endoscopists. However, these findings don’t reflect performance in clinical situations, as endoscopists classify lesions based on both magnified and non-magnified videos, while AI systems often use only a few magnified narrow band imaging (NBI) still images. We evaluated the performance of the AI system in simulated clinical situations.Methods:We used 25,048 images from 1,433 superficial ESCC and 4,746 images from 410 noncancerous esophagi to construct our AI system. For the validation dataset, we took NBI videos of suspected superficial ESCCs. The AI system diagnosis used one magnified still image taken from each video, while 19 endoscopists used whole videos. Results:We used 147 datasets including 83 superficial ESCC and 64 non-ESCC lesions. The accuracy, sensitivity and specificity for the classification of ESCC were, respectively, 80.9%, 85.5%, and 75.0% for the AI system and 69.2%, 67.5%, and 71.5% for the endoscopists. The AI system correctly classified all ESCCs invading the muscularis mucosa or submucosa and 96.8% of lesions ≥ 20 mm, whereas even the experts misdiagnosed some of them.Conclusions:Our AI system showed higher diagnostic ability for classifying ESCC and non-ESCC than endoscopists. It may provide valuable diagnostic support to endoscopists.