An Adaptive Transfer-Learning-Based Deep Cox Neural Network for Hepatocellular Carcinoma Prognosis Prediction

Epigenetic reactivation of tumor suppressor genes with CRISPRa technologies as precision therapy for hepatocellular carcinoma

BACKGROUND

Epigenetic silencing of tumor suppressor genes (TSGs) is a key feature of oncogenesis in hepatocellular carcinoma (HCC). Liver-targeted delivery of CRISPR-activation (CRISPRa) systems makes it possible to exploit chromatin plasticity, by reprogramming transcriptional dysregulation.

RESULTS

Using The Cancer Genome Atlas HCC data, we identify 12 putative TSGs with negative associations between promoter DNA methylation and transcript abundance, with limited genetic alterations. All HCC samples harbor at least one silenced TSG, suggesting that combining a specific panel of genomic targets could maximize efficacy, and potentially improve outcomes as a personalized treatment strategy for HCC patients. Unlike epigenetic modifying drugs lacking locus selectivity, CRISPRa systems enable potent and precise reactivation of at least 4 TSGs tailored to representative HCC lines. Concerted reactivation of HHIP, MT1M, PZP, and TTC36 in Hep3B cells inhibits multiple facets of HCC pathogenesis, such as cell viability, proliferation, and migration.

CONCLUSIONS

By combining multiple effector domains, we demonstrate the utility of a CRISPRa toolbox of epigenetic effectors and gRNAs for patient-specific treatment of aggressive HCC.

Screening for Lipid-Metabolism-Related Genes and Identifying the Diagnostic Potential of ANGPTL6 for HBV-Related Early-Stage Hepatocellular Carcinoma

Lipid metabolic reprogramming is one of the hallmarks of hepatocarcinogenesis and development. Therefore, lipid-metabolism-related genes may be used as potential biomarkers for hepatocellular carcinoma (HCC). This study aimed to screen for genes with dysregulated expression related to lipid metabolism in HCC and explored the clinical value of these genes. We screened differentially expressed proteins between tumorous and adjacent nontumorous tissues of hepatitis B virus (HBV)-related HCC patients using a Nanoscale Liquid Chromatography–Tandem Mass Spectrometry platform and combined it with transcriptomic data of lipid-metabolism-related genes from the GEO and HPA databases to identify dysregulated genes that may be involved in lipid metabolic processes. The potential clinical values of these genes were explored by bioinformatics online analysis tools (GEPIA, cBioPortal, SurvivalMeth, and TIMER). The expression levels of the secreted protein (angiopoietin-like protein 6, ANGPTL6) in serum were analyzed by ELISA. The ability of serum ANGPTL6 to diagnose early HCC was assessed by ROC curves. The results showed that serum ANGPTL6 could effectively differentiate between HBV-related early HCC patients with normal serum alpha-fetoprotein (AFP) levels and the noncancer group (healthy participants and chronic hepatitis B patients) (AUC = 0.717, 95% CI: from 0.614 to 0.805). Serum ANGPTL6 can be used as a potential second-line biomarker to supplement serum AFP in the early diagnosis of HBV-related HCC.

Study on the Prognostic Values of TTC36 Correlated with Immune Infiltrates and Its Methylation in Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) remains an incurable disease with a very poor clinical outcome. The purpose of this article was to investigate whether the expression or methylation of tetrapeptide repeat domain 36 (TTC36) could be used as a prognostic marker in hepatocellular carcinoma. TCGA database was used to obtain information on HCC gene expression and the associated clinical features of HCC patients. Differentially expressed genes (DEGs) were screened between 374 HCC specimens and 50 nontumor specimens. The expression and prognostic value of TTC36 were analyzed. The correlations between TTC36 and cancer immune infiltrates were investigated via TIMER. In this study, HCC specimens and nontumor specimens were compared and 35 DEGs were found between them. Among the 35 DEGs, the expression of TTC36 was significantly reduced in HCC samples compared with nontumor samples. Survival tests revealed that patients with low TTC36 expression had a shorter overall survival than patients with high TTC36 expression. TTC36 was found to be an independent predictive factor for HCC in both univariate and multivariate regression analyses. TTC36 was negatively regulated by TTC36 methylation, leading to its low expression in HCC tissues. Immune analysis revealed that TTC36 expression has significant correlations with B cell, T cell CD4+, neutrophil, macrophage, and myeloid dendritic cell. Finally, TTC36 expression was dramatically reduced in HCC cells, and overexpression greatly suppressed HCC cell proliferation and invasion, according to our experimental results. Overall, our data suggested that TTC36 could be applied as a prognostic marker for predicting outcome and immune infiltration in HCC.

Screening potential immune signatures for early-stage basal-like/triple-negative breast cancer

Background

Breast cancer (BC) is a highly heterogeneous disease. Among the BC molecular subtypes, basal-like/triple-negative BC (TNBC) is characterized by a high propensity for relatively early metastases and a lack of available endocrine and targeted therapies. Therefore, this study aimed to discover potential signatures for predicting the immune response in early-stage basal-like/triple-negative BC.

Method

A total of 86 cases of early-stage TNBC from the TCGA and 459 cases of normal breast tissue from GTEx were enrolled and analyzed to screen out differentially expressed genes (DEGs). Then, the prognostic effect and tumor immune cell infiltration relationship with the basal-like-specific DEGs were also evaluated.

Results

A total of 1556 DEGs, including 929 upregulated genes and 627 downregulated genes, were screened in early-stage basal-like BC. Two prognosis-associated DEGs, GAL and TTC36, were finally found to be basal-like BC specific. However, only GAL was significantly correlated with tumor immune-infiltrating cells, especially CD8⁺ T cells. The expressions of GAL and TTC36 were revalidated by using the GEO dataset.

Conclusion

GAL might be an immune signature for the response to immune checkpoint therapy in early basal-like/triple-negative BC.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12957-022-02683-2.

Deep-Learning-Based Survival Prediction of Patients in Coronary Care Units

Background

A survival prediction model based on deep learning has higher accuracy than the CPH model in predicting the survival of CCU patients, and it also has a better discrimination ability. We collected information on patients with various diseases in coronary care units (CCUs) from the Medical Information Mart for Intensive Care III (MIMIC-III) database. The purpose of this study was to use this information to construct a neural-network model based on deep learning to predict the survival probabilities of patients with conditions that are common in CCUs.

Method

We collected information on patients in the United States with five common diseases in CCUs from 2001 to 2012. We randomly divided the patients into a training cohort and a testing cohort at a ratio of 7 : 3 and applied a survival prediction method based on deep learning to predict their survival probability. We compared our model with the Cox proportional-hazards regression (CPH) model and used the concordance indexes (C-indexes), receiver operating characteristic (ROC) curve, and calibration plots to evaluate the predictive performance of the model.

Results

The 3,388 CCU patients included in the study were randomly divided into 2,371 in the training cohort and 1,017 in the testing cohort. The stepwise regression results showed that the important factors affecting patient survival were the type of disease, age, race, anion gap, glucose, neutrophils, white blood cells, potassium, creatine kinase, and blood urea nitrogen (P < 0.05). We used the training cohort to construct a deep-learning model, for which the C-index was 0.833, or about 5% higher than that for the CPH model (0.786). The C-index of the deep-learning model for the test cohort was 0.822, which was also higher than that for the CPH model (0.782). The areas under the ROC curve for the 28-day, 90-day, and 1-year survival probabilities were 0.875, 0.865, and 0.874, respectively, in the deep-learning model, respectively, and 0.830, 0.843, and 0.806 in the CPH model. These values indicate that the survival analysis model based on deep learning is better than the traditional CPH model in predicting the survival of CCU patients.

Conclusion

A survival prediction model based on deep learning has higher accuracy than the CPH model in predicting the survival of CCU patients, and it also has a better discrimination ability.