Artificial intelligence has enabled the automated diagnosis of several cancer types. We aimed to develop and validate deep learning models that automatically classify cervical intraepithelial neoplasia (CIN) based on histological images. Microscopic images of CIN3, CIN2, CIN1, and non-neoplasm were obtained. The performances of two pre-trained convolutional neural network (CNN) models adopting DenseNet-161 and EfficientNet-B7 architectures were evaluated and compared with those of pathologists. The dataset comprised 1106 images from 588 patients; images of 10% of patients were included in the test dataset. The mean accuracies for the four-class classification were 88.5% (95% confidence interval [CI], 86.3–90.6%) by DenseNet-161 and 89.5% (95% CI, 83.3–95.7%) by EfficientNet-B7, which were similar to human performance (93.2% and 89.7%). The mean per-class area under the receiver operating characteristic curve values by EfficientNet-B7 were 0.996, 0.990, 0.971, and 0.956 in the non-neoplasm, CIN3, CIN1, and CIN2 groups, respectively. The class activation map detected the diagnostic area for CIN lesions. In the three-class classification of CIN2 and CIN3 as one group, the mean accuracies of DenseNet-161 and EfficientNet-B7 increased to 91.4% (95% CI, 88.8–94.0%), and 92.6% (95% CI, 90.4–94.9%), respectively. CNN-based deep learning is a promising tool for diagnosing CIN lesions on digital histological images.

Kim JW, et al. J Korean Med Sci. 2015 Jan;30(1):16-23. http://dx.doi.org/10.3346/jkms.2015.30.1.16

Hong M, et al. J Korean Med Sci. 2017 Oct;32(10):1595-1602. https://doi.org/10.3346/jkms.2017.32.10.1595

The adhesion molecule Nectin-4 is a new potential therapeutic target for different types of cancer; however, little is known about its diagnosis significance in endometrial cancer (EC). We found that Nectin-4 expression was significantly higher in EC tissues than in nonadjacent normal tissue. The area under the receiver operating characteristic curve value of 0.922 indicated good diagnostic accuracy for Nectin-4 expression in EC. Furthermore, Nectin-4 expression was associated with DNA mismatch repair (MMR) protein deficiency. Notably, the high Nectin-4 expression group of patients with MSH2/6-deficient EC had shorter progression-free survival than that of the low Nectin-4 expression group. The number of lymphovascular space invasion-positive patients in groups with MMR deficiency and high Nectin-4 expression was also increased compared with that in the low Nectin-4 expression group. Bioinformatics analysis revealed that alteration in Nectin-4 and MMR genes is associated with Nectin-4 expression in EC. To the best of our knowledge, this is the first study to show that Nectin-4 expression may be a potential biomarker for EC diagnosis and that high Nectin-4 expression in MMR-deficient patients with EC can predict short progression-free survival, thus providing clues to identify patients for adjuvant therapy.

Upregulation of the expression of Delta/notch-like epidermal growth factor-related receptor (DNER) and its oncogenic role have been reported in several cancers, including gastric, breast, and prostate cancers. This study aimed to investigate the oncogenic role of DNER and the mechanisms behind its oncogenic role in gastric cancer. Analysis of the RNASeq data of gastric cancer tissues obtained from the TCGA database revealed that the expression of DNER was associated with the pathology of advanced gastric cancer and the prognosis of patients. DNER expression was increased upon stem cell-enriching cancer spheroid culture. Knockdown of DNER expression inhibited cell proliferation and invasion, induced apoptosis, enhanced chemosensitivity, and decreased spheroid formation of SNU-638 gastric cancer cells. DNER silencing elevated the expression of p53, p21^cip/waf, and p27, and increased G₁ phase cells at the expense of S phase cells. Knockdown of p21^cip/waf expression in the DNER-silenced cells partially restored cell viability and S phase progression. DNER silencing also induced the apoptosis of SNU-638 cells. While both cleaved caspases-8 and 9 were detected in adherent cells, only cleaved caspase-8 was found to have increased in spheroid-cultured cells, suggesting a distinct activation pattern of caspase activation depending on the growth condition. Knockdown of p53 expression rescued the DNER-silenced cells from apoptosis and partially restored cell viability. In contrast, overexpression of the Notch intracellular domain (NICD) decreased the expression of p53, p21^cip/waf, and cleaved caspase-3 in DNER-silenced cells. Moreover, NICD expression fully reverted the cell viability reduction, arrest in the G₁ phase, and elevated apoptosis caused by DNER silencing, thereby suggesting activation of Notch signaling by DNER. Expression of a membrane-unbound mutant of mDNER also decreased cell viability and induced apoptosis. On the other hand, TGF-β signals were found to be involved in DNER expression in both adherent and spheroid-cultured cells. DNER could therefore be a link connecting TGF-β signaling to Notch signaling. Taken together, DNER regulates cell proliferation, survival, and invasive capacity of the gastric cancer cells through the activation of Notch signaling, which may facilitate tumor progression into an advanced stage. This study provides evidences suggesting that DNER could be a potential prognostic marker, a therapeutic target, and a drug candidate in the form of a cell-free mutant.