Multi-center validation of machine learning model for preoperative prediction of postoperative mortality.

Accurate prediction of postoperative mortality is important for not only successful postoperative patient care but also for information-based shared decision-making with patients and efficient allocation of medical resources. This study aimed to create a machine-learning prediction model for 30-day mortality after a non-cardiac surgery that adapts to the manageable amount of clinical information as input features and is validated against multi-centered rather than single-centered data. Data were collected from 454,404 patients over 18 years of age who underwent non-cardiac surgeries from four independent institutions. We performed a retrospective analysis of the retrieved data. Only 12-18 clinical variables were used for model training. Logistic regression, random forest classifier, extreme gradient boosting (XGBoost), and deep neural network methods were applied to compare the prediction performances. To reduce overfitting and create a robust model, bootstrapping and grid search with tenfold cross-validation were performed. The XGBoost method in Seoul National University Hospital (SNUH) data delivers the best performance in terms of the area under receiver operating characteristic curve (AUROC) (0.9376) and the area under the precision-recall curve (0.1593). The predictive performance was the best when the SNUH model was validated with Ewha Womans University Medical Center data (AUROC, 0.941). Preoperative albumin, prothrombin time, and age were the most important features in the model for each hospital. It is possible to create a robust artificial intelligence prediction model applicable to multiple institutions through a light predictive model using only minimal preoperative information that can be automatically extracted from each hospital.

Identification of tissue of origin in cancer of unknown primary using a targeted bisulfite sequencing panel.

Aim: To construct a targeted bisulfite sequencing panel predicting origin of cancer of unknown primary. Methods: A bisulfite sequencing panel targeting 2793 tissue-specific markers was performed in 100 clinical samples. Results: The authors' prediction model showed 0.85 accuracy for the 'first-ranked' tissue type and 0.93 accuracy for the 'second-ranked' tissue type using 2793 tissue-specific markers and 0.84 accuracy for the 'first-ranked' tissue type and 0.92 accuracy for the 'second-ranked' tissue type when the number of tissue-specific markers was reduced to 514. Conclusion: Targeted bisulfite sequencing is a useful method for predicting the tissue of origin in patients with cancer of unknown primary.

When patients with cancer present with tumors that have migrated from elsewhere in the body, it is difficult for clinicians to identify where the cancer originated. DNA methylation profiling is a promising test to help identify where the cancer originated because it reflects cell of origin and is compatible with formalin-fixed, paraffin-embedded tissues. Because next-generation sequencing has already been implemented in clinical laboratories, the authors developed a targeted bisulfite sequencing panel that could predict the tissue of origin using genomic DNA extracted from formalin-fixed, paraffin-embedded tissues. The authors found that a hybrid capture-based targeted bisulfite sequencing panel is a useful method for predicting the tissue of origin in patients with cancer of unknown primary origin in clinical practice.

Targeted Sequencing Revealed Distinct Mutational Profiles of Ocular and Extraocular Sebaceous Carcinomas.

The biological behavior of sebaceous carcinoma (SeC) is relatively indolent; however, local invasion or distant metastasis is sometimes reported. Nevertheless, a lack of understanding of the genetic background of SeC makes it difficult to apply effective systemic therapy. This study was designed to investigate major genetic alterations in SeCs in Korean patients. A total of 29 samples, including 20 ocular SeCs (SeC-Os) and 9 extraocular SeCs (SeC-EOs), were examined. Targeted next-generation sequencing tests including 171 cancer-related genes were performed. TP53 and PIK3CA genes were frequently mutated in both SeC-Os and SeC-EOs with slight predominance in SeC-Os, whereas the NOTCH1 gene was more commonly mutated in SeC-EOs. In clinical correlation, mutations in RUNX1 and ATM were associated with development of distant metastases, and alterations in MSH6 and BRCA1 were associated with inferior progression-free survival (all p < 0.05). In conclusion, our study revealed distinct genetic alterations between SeC-Os and SeC-EOs and some important prognostic molecular markers. Mutations in potentially actionable genes, including EGFR, ERBB2, and mismatch repair genes, were noted, suggesting consideration of a clinical trial in intractable cases.

Transcriptomic Profiling Identifies a Risk Stratification Signature for Predicting Peritoneal Recurrence and Micrometastasis in Gastric Cancer.

PURPOSE: Gastric cancer peritoneal carcinomatosis is fatal. Delay in detection of peritoneal metastases contributes to high mortality, highlighting the need to develop biomarkers that can help identify patients at high risk for peritoneal recurrence or metastasis. EXPERIMENTAL DESIGN: We performed a systematic discovery and validation for the identification of peritoneal recurrence prediction and peritoneal metastasis detection biomarkers by analyzing expression profiling datasets from 249 patients with gastric cancer, followed by analysis of 426 patients from three cohorts for clinical validation. RESULTS: Genome-wide expression profiling identified a 12-gene panel for robust prediction of peritoneal recurrence in patients with gastric cancer (AUC = 0.95), which was successfully validated in a second dataset (AUC = 0.86). Examination of 216 specimens from a training cohort allowed us to establish a six gene-based risk-prediction model [AUC = 0.72; 95% confidence interval (CI): 0.66-0.78], which was subsequently validated in an independent cohort of 111 patients with gastric cancer (AUC = 0.76; 95% CI: 0.67-0.83). In both cohorts, combining tumor morphology and depth of invasion further improved the predictive accuracy of the prediction model (AUC = 0.84). Thereafter, we evaluated the performance of the identical six-gene panel for its ability to detect peritoneal metastasis by analyzing 210 gastric cancer specimens (prior 111 patients plus additional 99 cases), which discriminated patients with and without peritoneal metastasis (AUC = 0.72). Finally, our biomarker panel was also remarkably effective for identifying peritoneal micrometastasis (AUC = 0.72), and its diagnostic accuracy was significantly enhanced when depth of invasion was included in the model (AUC = 0.85). CONCLUSIONS: Our novel transcriptomic signature for risk stratification and identification of high-risk patients with peritoneal carcinomatosis might serve as an important clinical decision making in patients with gastric cancer.

Moesin (MSN) as a Novel Proteome-Based Diagnostic Marker for Early Detection of Invasive Bladder Urothelial Carcinoma in Liquid-Based Cytology.

Bladder urothelial carcinoma (BUC) is the most lethal malignancy of the urinary tract. Treatment for the disease highly depends on the invasiveness of cancer cells. Therefore, a predictive biomarker needs to be identified for invasive BUC. In this study, we employed proteomics methods on urine liquid-based cytology (LBC) samples and a BUC cell line library to determine a novel predictive biomarker for invasive BUC. Furthermore, an in vitro three-dimensional (3D) invasion study for biological significance and diagnostic validation through immunocytochemistry (ICC) were also performed. The proteomic analysis suggested moesin (MSN) as a potential biomarker to predict the invasiveness of BUC. The in vitro 3D invasion study showed that inhibition of MSN significantly decreased invasiveness in BUC cell lines. Further validation using ICC ultimately confirmed moesin (MSN) as a potential biomarker to predict the invasiveness of BUC (p = 0.023). In conclusion, we suggest moesin as a potential diagnostic marker for early detection of BUC with invasion in LBC and as a potential therapeutic target.

A Comprehensive Proteomic and Phosphoproteomic Analysis of Retinal Pigment Epithelium Reveals Multiple Pathway Alterations in Response to the Inflammatory Stimuli.

Overwhelming and persistent inflammation of retinal pigment epithelium (RPE) induces destructive changes in the retinal environment. However, the precise mechanisms remain unclear. In this study, we aimed to investigate RPE-specific biological and metabolic responses against intense inflammation and identify the molecular characteristics determining pathological progression. We performed quantitative analyses of the proteome and phosphoproteome of the human-derived RPE cell line ARPE-19 after treatment with lipopolysaccharide (LPS) for 45 min or 24 h using the latest isobaric tandem-mass tags (TMTs) labeling approach. This approach led to the identification of 8984 proteins, of which 261 showed a 1.5-fold change in abundance after 24 h of treatment with LPS. A parallel phosphoproteome analysis identified 20,632 unique phosphopeptides from 3207 phosphoproteins with 3103 phosphorylation sites. Integrated proteomic and phosphoproteomic analyses showed significant downregulation of proteins related to mitochondrial respiration and cell cycle checkpoint, while proteins related to lipid metabolism, amino acid metabolism, cell-matrix adhesion, and endoplasmic reticulum (ER) stress were upregulated after LPS stimulation. Further, phosphorylation events in multiple pathways, including MAPKK and Wnt/ß-catenin signalings, were identified as involved in LPS-triggered pathobiology. In essence, our findings reveal multiple integrated signals exerted by RPE under inflammation and are expected to give insight into the development of therapeutic interventions for RPE disorders.

Comparative Gene Expression Profiles in Parathyroid Adenoma and Normal Parathyroid Tissue.

Parathyroid adenoma is the main cause of primary hyperparathyroidism, which is characterized by enlarged parathyroid glands and excessive parathyroid hormone secretion. Here, we performed transcriptome analysis, comparing parathyroid adenomas with normal parathyroid gland tissue. RNA extracted from ten parathyroid adenoma and five normal parathyroid samples was sequenced, and differentially expressed genes (DEGs) were identified using strict cut-off criteria. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using DEGs as the input, and protein-protein interaction (PPI) networks were constructed using Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) and visualized in Cytoscape. Among DEGs identified in parathyroid adenomas (n = 247; 45 up-regulated, 202 down-regulated), the top five GO terms for up-regulated genes were nucleoplasm, nucleus, transcription DNA-template, regulation of mRNA processing, and nucleic acid binding, while those for down-regulated genes were extracellular exosome, membrane endoplasmic reticulum (ER), membrane, ER, and melanosome. KEGG enrichment analysis revealed significant enrichment of five pathways: protein processing in ER, protein export, RNA transport, glycosylphosphatidylinositol-anchor biosynthesis, and pyrimidine metabolism. Further, PPI network analysis identified a densely connected sub-module, comprising eight hub molecules: SPCS2, RPL23, RPL26, RPN1, SEC11C, SEC11A, RPS25, and SEC61G. These findings may be helpful in further analysis of the mechanisms underlying parathyroid adenoma development.