Computed tomography radiomic feature analysis of thymic epithelial tumors: Differentiation of thymic epithelial tumors from thymic cysts and prediction of histological subtypes.

PURPOSE: To investigate the value of computed tomography (CT) radiomic feature analysis for the differential diagnosis between thymic epithelial tumors (TETs) and thymic cysts, and prediction of histological subtypes of TETs. MATERIALS AND METHODS: Twenty-four patients with TETs (13 low-risk and 9 high-risk thymomas, and 2 thymic carcinomas) and 12 with thymic cysts were included in this study. For each lesion, the radiomic features of a volume of interest covering the lesion were extracted from non-contrast enhanced CT images. The Least Absolute Shrinkage and Selection Operator (Lasso) method was used for the feature selection. Predictive models for differentiating TETs from thymic cysts (model A), and high risk thymomas + thymic carcinomas from low risk thymomas (model B) were created from the selected features. The receiver operating characteristic curve was used to evaluate the effectiveness of radiomic feature analysis for differentiating among these tumors. RESULTS: In model A, the selected 5 radiomic features for the model A were NGLDM_Contrast, GLCM_Correlation, GLZLM_SZLGE, DISCRETIZED_HISTO_Entropy_log2, and DISCRETIZED_HUmin. In model B, sphericity was the only selected feature. The area under the curve, sensitivity, and specificity of radiomic feature analysis were 1 (95% confidence interval [CI]: 1-1), 100%, and 100%, respectively, for differentiating TETs from thymic cysts (model A), and 0.76 (95%CI: 0.53-0.99), 64%, and 100% respectively, for differentiating high-risk thymomas + thymic carcinomas from low-risk thymomas (model B). CONCLUSION: CT radiomic analysis could be utilized as a non-invasive imaging technique for differentiating TETs from thymic cysts, and high-risk thymomas + thymic carcinomas from low-risk thymomas.

An anoikis-related gene signature predicts prognosis and reveals immune infiltration in hepatocellular carcinoma.

Background: Hepatocellular carcinoma (HCC) is a global health burden with poor prognosis. Anoikis, a novel programmed cell death, has a close interaction with metastasis and progression of cancer. In this study, we aimed to construct a novel bioinformatics model for evaluating the prognosis of HCC based on anoikis-related gene signatures as well as exploring the potential mechanisms. Materials and methods: We downloaded the RNA expression profiles and clinical data of liver hepatocellular carcinoma from TCGA database, ICGC database and GEO database. DEG analysis was performed using TCGA and verified in the GEO database. The anoikis-related risk score was developed via univariate Cox regression, LASSO Cox regression and multivariate Cox regression, which was then used to categorize patients into high- and low-risk groups. Then GO and KEGG enrichment analyses were performed to investigate the function between the two groups. CIBERSORT was used for determining the fractions of 22 immune cell types, while the ssGSEA analyses was used to estimate the differential immune cell infiltrations and related pathways. The "pRRophetic" R package was applied to predict the sensitivity of administering chemotherapeutic and targeted drugs. Results: A total of 49 anoikis-related DEGs in HCC were detected and 3 genes (EZH2, KIF18A and NQO1) were selected out to build a prognostic model. Furthermore, GO and KEGG functional enrichment analyses indicated that the difference in overall survival between risk groups was closely related to cell cycle pathway. Notably, further analyses found the frequency of tumor mutations, immune infiltration level and expression of immune checkpoints were significantly different between the two risk groups, and the results of the immunotherapy cohort showed that patients in the high-risk group have a better immune response. Additionally, the high-risk group was found to have higher sensitivity to 5-fluorouracil, doxorubicin and gemcitabine. Conclusion: The novel signature of 3 anoikis-related genes (EZH2, KIF18A and NQO1) can predict the prognosis of patients with HCC, and provide a revealing insight into personalized treatments in HCC.

Identification of iron ore brands by multi-component analysis and chemometric tools.

Identification of iron ore brand is one of the most important precautions against fraud in the international iron ore trade. However, the identification of iron ore brand can be sophisticated, due to fact that the role played by multi-component in iron ore brand identification was unclear. This study aims to establish an objective approach to identify iron ore brands based on their multi-component content. A total of 1469 batches of iron ore samples, covering 16 commonly consumed iron ore brands from 3 countries, were analyzed for multi-component content. It was investigated that 10 primary, minor, and trace chemical components varied significantly in contents according to different iron ore brands. This prospective relationship between the multi-component contents and the iron ore brand was then used to place 16 brands into 12 groups and 8 brands of them were correctly identified by a flowchart. Furthermore, chemometric tools such as linear discriminant analysis (LDA), k-nearest neighbor (k-NN), and support vector machine (SVM) were applied to construct models to simultaneously discriminate 16 iron ore brands. Both the training and test results proved that LDA performed best in this circumstance. In the LDA method, MgO, Fe, SiO2, and P are the feature components contributing the most to the identification of 16 brands of iron ore. Based on the findings, the multi-components are distinct variables to establish an internationally recognized model of iron ore brand identification.

Interpretable deep learning-assisted laser-induced breakdown spectroscopy for brand classification of iron ores.

Brand classification of iron ores using laser-induced breakdown spectroscopy (LIBS) combined with artificial neural networks can quickly realize the compliance verification and guarantee the interests of both trading partners. However, its practical application is impeded by complex pretreatments and unexplained feature learning problems. According to the LIBS data characteristics of iron ores, a convolutional neural network (CNN) is designed to predict 16 types of brand iron ores from Australia, Brazil, and South Africa. The accuracies of the calibration set and the prediction set with five-fold cross-validation (5-CV) were 99.86% and 99.88%, and the value of loss function was 0.0356. Meanwhile, the established CNN method was also compared with common machine learning methods using raw spectra as input variables, and it outperformed other methods. For the first time, this work interprets the CNN's effectiveness layer by layer in self-adaptively extracting LIBS features through t-distributed stochastic neighbor embedding (t-SNE) and the quantitative data of major chemical components in iron ores. Our approach shows that deep learning assisted LIBS is able to significantly reduce manual factors in preprocessing and feature selection and has broad application prospects in the brand classification of iron ores.

A novel serum based biomarker panel has complementary ability to preclude presence of early lung cancer for low dose CT (LDCT).

Low Dosage Computerized Tomography (LDCT) has been shown to improve early detection of lung cancer and mortality rates in high-risk individuals, which was, however, limited by specifically coverage for heavy smokers and high rates of false positivity. Here, we aim to investigate a novel biomarker for early detection of lung cancer, and further extend to concentrate high-risk subjects for increasing specificity and coverage of LDCT. We performed retrospective blinded evaluation of lung cancer and healthy controls in training and validation cohorts. Macrophage inhibitory cytokine 1 (MIC-1) alone and panel were assessed. Our data showed the sensitivity of MIC-1 was 72.2% and 67.1% for lung cancer diagnosis and early diagnosis respectively, at 96.6% specificity, which were significantly higher than Cyfra21-1, NSE CA125, CEA and SCC. At 90% specificity, the panel of MIC-1, Cyfra21-1, CA125 and CEA provided 89.5% sensitivity for early diagnosis of lung cancer, which could be used to concentrate the high-risk subjects for further LDCT screening. We conclude that MIC-1 have great capacity in early lung cancer diagnosis. The algorithmic panel of MIC-1, Cyfra21-1, CA125 and CEA could be used to refine the preselection criteria of high-risk subjects, and thus might facilitate the widespread implementation of LDCT screening.

Circulating MIC-1/GDF15 is a complementary screening biomarker with CEA and correlates with liver metastasis and poor survival in colorectal cancer.

Macrophage inhibitory cytokine 1 (MIC-1/GDF15) has been characterized as a candidate biomarker for colorectal cancer (CRC) recently. However, the role of serum MIC-1 in screening patients with early stage CRC and monitoring therapeutic response have not been well-established, particularly in the combination with CEA for the screening and the prejudgment of occurrence with liver metastasis. In this study, we performed a retrospective blinded evaluation of 987 serum samples from 473 individuals with CRC, 25 with adenomatous polyps, and 489 healthy individuals using ELISA or immunoassay. The sensitivity of serum MIC-1 was 43.8% and 38.5% for CRC diagnosis and early diagnosis, respectively, which were independent of and comparatively higher than for CEA (36.6% and 27.3%) at comparable specificity. Serum MIC-1 after surgery were significantly elevated at the time of tumor recurrence, and notable increase were observed in 100% patients with liver metastasis. Besides the TNM classification and differentiation grade, MIC-1 was an independent prognostic factor contributing to overall survival. We conclude that MIC-1 can act as a candidate complementary biomarker for screening early-stage CRC by combination with CEA, and furthermore, for the first time, identify a promising prognostic indicator for monitoring recurrence with liver metastasis, to support strategies towards personalized therapy.

Macrophage inhibitory cytokine 1 (MIC-1/GDF15) as a novel diagnostic serum biomarker in pancreatic ductal adenocarcinoma.

BACKGROUND: Macrophage inhibitory cytokine 1 (MIC-1/GDF15) has been identified as a potential novel biomarker for detection of pancreatic cancer (PCa). However, the diagnostic value of serum MIC-1 for pancreatic ductal adenocarcinoma (PDAC), particularly for those at the early stage, and the value for treatment response monitoring have not yet been investigated. METHODS: MIC-1 expression in tumor tissue was analyzed by RT-PCR from 64 patients with PDAC. Serum MIC-1 levels were detected by ELISA in 1472 participants including PDAC, benign pancreas tumor, chronic pancreatitis and normal controls. The diagnostic performance of MIC-1 was assessed and compared with CA19.9, CEA and CA242, and the value of it as a predictive indicator for therapeutic response and tumor recurrence was also evaluated. RESULTS: MIC-1 levels were significantly elevated in PDAC tissues as well as serum samples. The sensitivity of serum MIC-1 for PDAC diagnosis was much higher than that of CA19.9 (65.8% vs. 53.3%) with similar specificities. Furthermore, serum MIC-1 detected 238 out of 377 (63.1%) CA19.9-negative PDAC. Moreover, receiver operating characteristic (ROC) curve analysis also showed that serum MIC-1 had a better performance compared with CA19.9 in distinguishing early-stage PDAC from normal serum with a higher sensitivity (62.5% vs. 25.0% respectively). Notably, serum MIC-1 level was significantly decreased in patients with PDAC after curative resection and returned to elevated levels when tumor relapse occurred. CONCLUSIONS: Serum MIC-1 is significantly elevated in most PDAC, including those with negative CA19.9 and early stage disease, and thus may serve as a novel diagnostic marker in early diagnosis and postoperative monitoring of PDAC.