Development and Validation of an Autophagy-Stroma-Based Microenvironment Gene Signature for Risk Stratification in Colorectal Cancer

Experimental prognostic model integrating N6-methyladenosine-related programmed cell death genes in colorectal cancer

Colorectal cancer (CRC) intricacies, involving dysregulated cellular processes and programmed cell death (PCD), are explored in the context of N6-methyladenosine (m6A) RNA modification. Utilizing the TCGA-COADREAD/CRC cohort, 854 m6A-related PCD genes are identified, forming the basis for a robust 10-gene risk model (CDRS) established through LASSO Cox regression. qPCR experiments using CRC cell lines and fresh tissues was performed for validation. The CDRS served as an independent risk factor for CRC and showed significant associations with clinical features, molecular subtypes, and overall survival in multiple datasets. Moreover, CDRS surpasses other predictors, unveiling distinct genomic profiles, pathway activations, and associations with the tumor microenvironment. Notably, CDRS exhibits predictive potential for drug sensitivity, presenting a novel paradigm for CRC risk stratification and personalized treatment avenues.

Integration of bulk RNA sequencing data and single-cell RNA sequencing analysis on the heterogeneity in patients with colorectal cancer

The cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) pathway has emerged as a critical innate immune pathway that could virtually impact nearly all aspects of tumorigenesis including colorectal cancer. This work aimed to develop and validate molecular subtypes related to cGAS-STING pathways for colorectal cancer using Bulk RNA-seq and single-cell RNA-seq (scRNA-seq) data. Bulk RNA-seq data were acquired from The Cancer Genome Atlas dataset (training dataset) and Gene Expression Omnibus dataset (validation dataset). Univariate COX survival analysis was utilized to identify prognostic differentially expressed genes (DEGs) from 6 immune pathways related to cGAS-STING. ConsensusClusterPlus package was used to classify different subtypes based on DEGs. scRNA-seq data were used to validate differences in immune status between different subtypes. Two clusters with distinct prognosis were identified based on 27 DEGs. The six cGAS-STING-related pathways had different levels of significance between the two clusters. Clust1 had most number of amplified CNVs and clust2 had the most number of loss CNVs. TP53 was the top mutated gene of which missense mutations contributed the most of single-nucleotide variants. Immune score of clust1 was higher than that in clust2, as reflected in macrophages, T cells, and natural killer cells. Three unfavorable genes and 31 protection factors were screened between the two clusters in three datasets. ScRNA-seq data analysis demonstrated that macrophages were more enriched in clust1, and tumor cells and immune cells had close interaction. We classified two distinct subtypes with different prognosis, mutation landscape, and immune characteristics.

A Novel Gene List Identifies Tumors with a Stromal-Mesenchymal Phenotype and Worse Prognosis in Gastric Cancer

BACKGROUND

Molecular biomarkers that predict disease progression can help identify tumor subtypes and shape treatment plans. In this study, we aimed to identify robust biomarkers of prognosis in gastric cancer based on transcriptomic data obtained from primary gastric tumors.

METHODS

Microarray, RNA sequencing, and single-cell RNA sequencing-based gene expression data from gastric tumors were obtained from public databases. Freshly frozen gastric tumors (n = 42) and matched FFPE (formalin-fixed, paraffin-embedded) (n = 40) tissues from a Turkish gastric cancer cohort were used for quantitative real-time PCR and immunohistochemistry-based assessments of gene expression, respectively.

RESULTS

A novel list of 20 prognostic genes was identified and used for the classification of gastric tumors into two major tumor subgroups with differential stromal gene expression ("Stromal-UP" (SU) and "Stromal-DOWN" (SD)). The SU group had a more mesenchymal profile with an enrichment of extracellular matrix-related gene sets and a poor prognosis compared to the SD group. Expression of the genes within the signature correlated with the expression of mesenchymal markers ex vivo. A higher stromal content in FFPE tissues was associated with shorter overall survival.

CONCLUSIONS

A stroma-rich, mesenchymal subgroup among gastric tumors identifies an unfavorable clinical outcome in all cohorts tested.

Construction and validation of a bladder cancer risk model based on autophagy-related genes

Autophagy has an important association with tumorigenesis, progression, and prognosis. However, the mechanism of autophagy-regulated genes on the risk prognosis of bladder cancer (BC) patients has not been fully elucidated yet. In this study, we created a prognostic model of BC risk based on autophagy-related genes, which further illustrates the value of genes associated with autophagy in the treatment of BC. We first downloaded human autophagy-associated genes and BC datasets from Human Autophagy Database and The Cancer Genome Atlas (TCGA) database, and finally obtained differential prognosis-associated genes for autophagy by univariate regression analysis and differential analysis of cancer versus normal tissues. Subsequently, we downloaded two datasets from Gene Expression Omnibus (GEO), GSE31684 and GSE15307, to expand the total number of samples. Based on these genes, we distinguished the molecular subtypes (C1, C2) and gene classes (A, B) of BC by consistent clustering analysis. Using the genes merged from TCGA and the two GEO datasets, we conducted least absolute shrinkage and selection operator (LASSO) and multivariate Cox regression analysis to obtain risk genes and construct autophagy-related risk prediction models. The accuracy of this risk prediction model was assessed by receiver operating characteristic (ROC) and calibration curves, and then nomograms were constructed to predict the survival of bladder cancer patients at 1, 3, and 5 years, respectively. According to the median value of the risk score, we divided BC samples into the high- and low-risk groups. Kaplan-Meier (K-M) survival analysis was performed to compare survival differences between subgroups. Then, we used single sample gene set enrichment analysis (ssGSEA) for immune cell infiltration abundance, immune checkpoint genes, immunotherapy response, gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis, and tumor mutation burden (TMB) analysis for different subgroups. We also applied quantitative real-time polymerase chain reaction (PCR) and immunohistochemistry (IHC) techniques to verify the expression of these six genes in the model. Finally, we chose the IMvigor210 dataset for external validation. Six risk genes associated with autophagy (SPOCD1, FKBP10, NAT8B, LDLR, STMN3, and ANXA2) were finally screened by LASSO regression algorithm and multivariate Cox regression analysis. ROC and calibration curves showed that the model established was accurate and reliable. Univariate and multivariate regression analyses were used to verify that the risk model was an independent predictor. K-M survival analysis indicated that patients in the high-risk group had significantly worse overall survival than those in the low-risk group. Analysis by algorithms such as correlation analysis, gene set variation analysis (GSVA), and ssGSEA showed that differences in immune microenvironment, enrichment of multiple biologically active pathways, TMB, immune checkpoint genes, and human leukocyte antigens (HLAs) were observed in the different risk groups. Then, we constructed nomograms that predicted the 1-, 3-, and 5-year survival rates of different BC patients. In addition, we screened nine sensitive chemotherapeutic drugs using the correlation between the obtained expression status of risk genes and drug sensitivity results. Finally, the external dataset IMvigor210 verified that the model is reliable and efficient. We established an autophagy-related risk prognostic model that is accurate and reliable, which lays the foundation for future personalized treatment of bladder cancer.

Integrating machine learning to construct aberrant alternative splicing event related classifiers to predict prognosis and immunotherapy response in patients with hepatocellular carcinoma

Introduction: In hepatocellular carcinoma (HCC), alternative splicing (AS) is related to tumor invasion and progression.

Methods: We used HCC data from a public database to identify AS subtypes by unsupervised clustering. Through feature analysis of different splicing subtypes and acquisition of the differential alternative splicing events (DASEs) combined with enrichment analysis, the differences in several subtypes were explored, cell function studies have also demonstrated that it plays an important role in HCC.

Results: Finally, in keeping with the differences between these subtypes, DASEs identified survival-related AS times, and were used to construct risk proportional regression models. AS was found to be useful for the classification of HCC subtypes, which changed the activity of tumor-related pathways through differential splicing effects, affected the tumor microenvironment, and participated in immune reprogramming.

Conclusion: In this study, we described the clinical and molecular characteristics providing a new approach for the personalized treatment of HCC patients.