Microbial Gene Ontology informed deep neural network for microbe functionality discovery in human diseases.

The human microbiome plays a crucial role in human health and is associated with a number of human diseases. Determining microbiome functional roles in human diseases remains a biological challenge due to the high dimensionality of metagenome gene features. However, existing models were limited in providing biological interpretability, where the functional role of microbes in human diseases is unexplored. Here we propose to utilize a neural network-based model incorporating Gene Ontology (GO) relationship network to discover the microbe functionality in human diseases. We use four benchmark datasets, including diabetes, liver cirrhosis, inflammatory bowel disease, and colorectal cancer, to explore the microbe functionality in the human diseases. Our model discovered and visualized the novel candidates' important microbiome genes and their functions by calculating the important score of each gene and GO term in the network. Furthermore, we demonstrate that our model achieves a competitive performance in predicting the disease by comparison with other non-Gene Ontology informed models. The discovered candidates' important microbiome genes and their functions provide novel insights into microbe functional contribution.

A novel approach to quantify calcifications of thyroid nodules in US images based on deep learning: predicting the risk of cervical lymph node metastasis in papillary thyroid cancer patients.

OBJECTIVE: Based on ultrasound (US) images, this study aimed to detect and quantify calcifications of thyroid nodules, which are regarded as one of the most important features in US diagnosis of thyroid cancer, and to further investigate the value of US calcifications in predicting the risk of lymph node metastasis (LNM) in papillary thyroid cancer (PTC). METHODS: Based on the DeepLabv3+ networks, 2992 thyroid nodules in US images were used to train a model to detect thyroid nodules, of which 998 were used to train a model to detect and quantify calcifications. A total of 225 and 146 thyroid nodules obtained from two centers, respectively, were used to test the performance of these models. A logistic regression method was used to construct the predictive models for LNM in PTCs. RESULTS: Calcifications detected by the network model and experienced radiologists had an agreement degree of above 90%. The novel quantitative parameters of US calcification defined in this study showed a significant difference between PTC patients with and without cervical LNM (p < 0.05). The calcification parameters were beneficial to predicting the LNM risk in PTC patients. The LNM prediction model using these calcification parameters combined with patient age and other US nodular features showed a higher specificity and accuracy than the calcification parameters alone. CONCLUSIONS: Our models not only detect the calcifications automatically, but also have value in predicting cervical LNM risk of PTC patients, thereby making it possible to investigate the relationship between calcifications and highly invasive PTC in detail. CLINICAL RELEVANCE STATEMENT: Due to the high association of US microcalcifications with thyroid cancers, our model will contribute to the differential diagnosis of thyroid nodules in daily practice. KEY POINTS: • We developed an ML-based network model for automatically detecting and quantifying calcifications within thyroid nodules in US images. • Three novel parameters for quantifying US calcifications were defined and verified. • These US calcification parameters showed value in predicting the risk of cervical LNM in PTC patients.

An integrated AI model to improve diagnostic accuracy of ultrasound and output known risk features in suspicious thyroid nodules.

OBJECTIVES: From the viewpoint of ultrasound (US) physicians, an ideal thyroid US computer-assisted diagnostic (CAD) system for thyroid cancer should perform well in suspicious thyroid nodules with atypical risk features and be able to output explainable results. This study aims to develop an explainable US CAD model for suspicious thyroid nodules. METHODS: A total of 2992 solid or almost-solid thyroid nodules were analyzed retrospectively. All nodules had pathological results (1070 malignancies and 1992 benignities) confirmed by ultrasound-guided fine-needle aspiration cytology and histopathology after thyroidectomy. A deep learning model (ResNet50) and a multiple risk features learning ensemble model (XGBoost) were used to train the US images of 2794 thyroid nodules. Then, an integrated AI model was generated by combining both models. The diagnostic accuracies of the three AI models (ResNet50, XGBoost, and the integrated model) were predicted in a testing set including 198 thyroid nodules and compared to the diagnostic efficacy of five ultrasonographers. RESULTS: The accuracy of the integrated model was 76.77%, while the mean accuracy of the ultrasonographers was 68.38%. Of the risk features, microcalcifications showed the highest contribution to the diagnosis of malignant nodules. CONCLUSIONS: The integrated AI model in our study can improve the diagnostic accuracy of suspicious thyroid nodules and output the known risk features simultaneously, thus aiding in training young ultrasonographers by linking the explainable results to their clinical experience and advancing the acceptance of AI diagnosis for thyroid cancer in clinical practice. KEY POINTS: • We developed an artificial intelligence (AI) diagnosis model based on both deep learning and multiple risk feature ensemble learning methods. • The AI diagnosis model showed higher diagnostic accuracy for suspicious thyroid nodules than ultrasonographers. • The AI diagnosis model showed partial explainability by outputting the known risk features, thus aiding young ultrasonic doctors in increasing the diagnostic level for thyroid cancer.

HEAL: an automated deep learning framework for cancer histopathology image analysis.

MOTIVATION: Digital pathology supports analysis of histopathological images using deep learning methods at a large-scale. However, applications of deep learning in this area have been limited by the complexities of configuration of the computational environment and of hyperparameter optimization, which hinder deployment and reduce reproducibility. RESULTS: Here, we propose HEAL, a deep learning-based automated framework for easy, flexible and multi-faceted histopathological image analysis. We demonstrate its utility and functionality by performing two case studies on lung cancer and one on colon cancer. Leveraging the capability of Docker, HEAL represents an ideal end-to-end tool to conduct complex histopathological analysis and enables deep learning in a broad range of applications for cancer image analysis. AVAILABILITY AND IMPLEMENTATION: The docker image of HEAL is available at https://hub.docker.com/r/docurdt/heal and related documentation and datasets are available at http://heal.erc.monash.edu.au. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The clinicopathologic and immunohistochemical features of villoglandular adenocarcinoma of uterine cervix.

AIM: Villoglandular adenocarcinoma (VGA) of the uterine cervix is a variant of endocervical adenocarcinoma. However, the clinicopathologic and immunohistochemical features of VGA are still unclear. The aim of this study was to investigate the clinicopathologic and immunohistochemical features of VGA. MATERIALS AND METHODS: A total of 20 VGA patients were identified among 852 patients diagnosed with cervical cancer and enrolled in this study. The immunohistochemical levels of Ki-67, P53, P16, progesterone receptor (PR), carcinoembryonic antigen (CEA), vimentin (Vim), and estrogen receptor (ER) were measured by immunohistochemistry. RESULTS: VGA was prevalent in younger women and presented favorable prognosis. Ki-67, P16, and CEA were highly expressed in VGA tissues, while PR expression was hardly to be detected. The positive rates of Ki-67, CEA, and P16 were 90.0%, 90.0%, and 85.0%, respectively, which were significantly higher compared with PR (5.0%, P < 0.001). In addition, the positive rates of P53, Vim, and ER in VGA tissues were 55.0%, 50.0%, and 40.0%, respectively. However, the expression levels of Ki-67, P53, P16, PR, CEA, Vim, and ER were not significantly associated with clinical features (P > 0.05). CONCLUSION: These data indicate that VGA is a rare cervical adenocarcinoma, which is prevalent in younger women, and presents favorable prognosis. Detection of Ki-67, P53, P16, PR, CEA, Vim, and ER would be beneficial for the diagnosis of VGA.

A novel ASXL1-OGT axis plays roles in H3K4 methylation and tumor suppression in myeloid malignancies.

ASXL1 plays key roles in epigenetic regulation of gene expression through methylation of histone H3K27, and disruption of ASXL1 drives myeloid malignancies, at least in part, via derepression of posterior HOXA loci. However, little is known about the identity of proteins that interact with ASXL1 and about the functions of ASXL1 in modulation of the active histone mark, such as H3K4 methylation. In this study, we demonstrate that ASXL1 is a part of a protein complex containing HCFC1 and OGT; OGT directly stabilizes ASXL1 by O-GlcNAcylation. Disruption of this novel axis inhibited myeloid differentiation and H3K4 methylation as well as H2B glycosylation and impaired transcription of genes involved in myeloid differentiation, splicing, and ribosomal functions; this has implications for myelodysplastic syndrome (MDS) pathogenesis, as each of these processes are perturbed in the disease. This axis is responsible for tumor suppression in the myeloid compartment, as reactivation of OGT induced myeloid differentiation and reduced leukemogenecity both in vivo and in vitro. Our data also suggest that MLL5, a known HCFC1/OGT-interacting protein, is responsible for gene activation by the ASXL1-OGT axis. These data shed light on the novel roles of the ASXL1-OGT axis in H3K4 methylation and activation of transcription.

Identification of a p53-repressed gene module in breast cancer cells.

The p53 protein is a sophisticated transcription factor that regulates dozens of target genes simultaneously in accordance with the cellular circumstances. Although considerable efforts have been made to elucidate the functions of p53-induced genes, a holistic understanding of the orchestrated signaling network repressed by p53 remains elusive. Here, we performed a systematic analysis to identify simultaneously regulated p53-repressed genes in breast cancer cells. Consequently, 28 genes were designated as the p53-repressed gene module, whose gene components were simultaneously suppressed in breast cancer cells treated with Adriamycin. A ChIP-seq database showed that p53 does not preferably bind to the region around the transcription start site of the p53-repressed gene module elements compared with that of p53-induced genes. Furthermore, we demonstrated that p21/CDKN1A plays a pivotal role in the suppression of the p53-repressed gene module in breast cancer cells. Finally, we showed that appropriate suppression of some genes belonging to the p53-repressed gene module contributed to a better prognosis of breast cancer patients. Taken together, these findings disentangle the gene regulatory network underlying the built-in p53-mediated tumor suppression system.

Identification of a p53 target, CD137L, that mediates growth suppression and immune response of osteosarcoma cells.

p53 encodes a transcription factor that transactivates downstream target genes involved in tumour suppression. Although osteosarcoma frequently has p53 mutations, the role of p53 in osteosarcomagenesis is not fully understood. To explore p53-target genes comprehensively in calvarial bone and find out novel druggable p53 target genes for osteosarcoma, we performed RNA sequencing using the calvarial bone and 23 other tissues from p53 +/+ and p53 -/- mice after radiation exposure. Of 23,813 genes, 69 genes were induced more than two-fold in irradiated p53 +/+ calvarial bone, and 127 genes were repressed. Pathway analysis of the p53-induced genes showed that genes associated with cytokine-cytokine receptor interactions were enriched. Three genes, CD137L, CDC42 binding protein kinase gamma and Follistatin, were identified as novel direct p53 target genes that exhibited growth-suppressive effects on osteosarcoma cell lines. Of the three genes, costimulatory molecule Cd137l was induced only in calvarial bone among the 24 tissues tested. CD137L-expressing cells exhibited growth-suppressive effects in vivo. In addition, recombinant Fc-fusion Cd137l protein activated the immune response in vitro and suppressed osteosarcoma cell growth in vivo. We clarified the role of CD137L in osteosarcomagenesis and its potential therapeutic application. Our transcriptome analysis also indicated the regulation of the immune response through p53.

The Transcriptional Landscape of p53 Signalling Pathway.

Although recent cancer genomics studies have identified a large number of genes that were mutated in human cancers, p53 remains as the most frequently mutated gene. To further elucidate the p53-signalling network, we performed transcriptome analysis on 24 tissues in p53+/+ or p53-/- mice after whole-body X-ray irradiation. Here we found transactivation of a total of 3551 genes in one or more of the 24 tissues only in p53+/+ mice, while 2576 genes were downregulated. p53 mRNA expression level in each tissue was significantly associated with the number of genes upregulated by irradiation. Annotation using TCGA (The Cancer Genome Atlas) database revealed that p53 negatively regulated mRNA expression of several cancer therapeutic targets or pathways such as BTK, SYK, and CTLA4 in breast cancer tissues. In addition, stomach exhibited the induction of Krt6, Krt16, and Krt17 as well as loricrin, an epidermal differentiation marker, after the X-ray irradiation only in p53+/+ mice, implying a mechanism to protect damaged tissues by rapid induction of differentiation. Our comprehensive transcriptome analysis elucidated tissue specific roles of p53 and its signalling networks in DNA-damage response that will enhance our understanding of cancer biology.

[Expression of YY 1 protein in human insulinoma and its clinical implication].

OBJECTIVE: To investigate the expression of Yin Yang 1 (YY1) protein in human insulinoma and explore its clinical significance. METHODS: Nineteen pancreatic neuroendocrine tumor tissue were collected from patients treated in Nanfang Hospital between 2000 and 2014. The protein expression of YY1 in benign and malignant insulinoma tissues were detected by immunohistochemistry. RESULTS: Positive expression for YY1 protein was detected in both benign and malignant tumor tissues, but the malignant tissues had a significantly greater intensity of YY1 expression than the benign tissues (P=0.042). The intensity of YY1 expression was positively correlated with the nature of the tumor, and the insulinomas with high expressions of YY1 had significantly greater malignant potentials (P=0.037). CONCLUSION: The high expression of YY1 protein is associated with the development of insulinima. YY1 may serve as a new tumor marker for detecting the malignant transformation of insulinoma.

[Prognostic value of circulating tumor cells and disseminated tumor cells in patients with esophageal cancer: a meta-analysis].

OBJECTIVE: To explore the correlations of circulating tumor cells (CTCs) and disseminated tumor cells (DTCs) with the clinicopathological characteristics, prognostic events, and survival outcomes in esophageal cancer (EC) patients. METHODS: The PubMed, Web of Science, Embase database and Cochrane database were searched for studies reporting the outcomes of interest. The studies were selected according to established inclusion/exclusion criteria. Meta-analysis of the studies was performed using Review Manager 5.3 and Stata12.0 software with the odds ratio (OR), risk ratio (RR) , hazard ratio (HR) , and 95% confidence interval (95% CI) as the effect indexes. RESULTS: Nineteen studies involving a total of 1766 patients were included in the analysis. Significant correlations of CTCs and DTCs were found with the clinicopathological parameters including the tumor stage (OR=1.95), depth of invasion (OR=1.99), lymph node metastasis (OR=2.44SEN), distal metastasis (OR=5.98SEN), histological differentiation (OR=1.67) and lymphovascular invasion (OR=4.48). CTCs and DTCs were also correlated with the prognostic events including relapse (RR=6.86SEN) and metastasis (RR=3.22) and with the survival outcomes including the overall survival (OS) overall analysis (HR=3.46) and disease-free survival/progression-free survival (DFS/PFS) overall analysis (HR=3.00). CONCLUSION: CTCs and DTCs are significantly associated with an advanced tumor stage, depth of tumor invasion, lymph node metastasis, distant metastasis before therapy, differentiation, lymphovascular invasion, relapse and metastasis in patients with EC. They are also significantly correlated with a poorer survival for OS and DFS/PFS to serve as clinical and prognostic predictors in patients with EC.

Micro-ultrasonographic imaging of atherosclerotic progression and correlation with inflammatory markers in apolipoprotein-E knockout mice.

We studied prospectively whether atherosclerotic progression in apolipoprotein-E knockout mice could be noninvasively and accurately measured by use of high-resolution ultrasonographic biomicroscopy. We examined the correlation between the ultrasonographic characterization of ascending aortic atherosclerotic plaque and plasma C-reactive protein, interleukin-1, and interleukin-6 levels in these mice.In 4 age groups (8, 16, 24, and 32 wk) of 8 male knockout mice each (atherosclerotic groups) and age-matched male C57BL/6 mice (control groups), we used ultrasonographic biomicroscopy to measure maximal plaque thickness or intima-media thickness in the ascending aorta. We compared the findings with corresponding histologic measurements, and we measured plasma C-reactive protein, interleukin-1, and interleukin-6 levels in each group.Mean atherosclerotic thicknesses and C-reactive protein and interleukin levels were significantly higher in each atherosclerotic group than in the control groups (all P < 0.05). Ultrasonographically measured atherosclerotic thickness correlated well with histologic measurements of the same vascular regions (r = 0.81, P < 0.001). C-reactive protein levels increased concomitantly with age in the knockout mice, and ultrasonographically measured atherosclerotic thickness correlated with those levels (r = 0.626, P < 0.001). However, there was no correlation between plasma interleukin levels and atherosclerotic severity as measured by ultrasonographic biomicroscopy.In the apolipoprotein-E knockout mice, we found that measurements of intima-media or maximal plaque thickness by ultrasonographic biomicroscopy noninvasively and accurately detected atherosclerotic progression, that plasma C-reactive protein levels correlated with atherosclerosis, and that elevated plasma C-reactive protein levels correlated with atherosclerotic severity.

Determination of nine organophosphorus pesticides in cereals and kidney beans by capillary gas chromatography with flame photometric detection.

A method is developed for the determination of nine organophosphorus pesticide residues in cereals and kidney beans by capillary gas chromatography with flame-photometric detection. In this method, dichloromethane is used for clean-up after liquid-liquid extraction. It is shown that good separations are obtained using a fused-silica capillary column (DB-1701) by the optimized temperature program. In the spiked levels of 0.012-0.43 mg/kg, the recoveries are from 83.7% to 107%, with the relative standard deviation between 3.2% and 13% and limits of detection from 8.2 to 15 microg/kg. The method is rapid, sensitive, and practical.