Machine learning revealed stemness features and a novel stemness-based classification with appealing implications in discriminating the prognosis, immunotherapy and temozolomide responses of 906 glioblastoma patients.
Glioblastoma (GBM) is the most malignant and lethal intracranial tumor, with extremely limited treatment options. Immunotherapy has been widely studied in GBM, but none can significantly prolong the overall survival (OS) of patients without selection. Considering that GBM cancer stem cells (CSCs) play a non-negligible role in tumorigenesis and chemoradiotherapy resistance, we proposed a novel stemness-based classification of GBM and screened out certain population more responsive to immunotherapy. The one-class logistic regression algorithm was used to calculate the stemness index (mRNAsi) of 518 GBM patients from The Cancer Genome Atlas (TCGA) database based on transcriptomics of GBM and pluripotent stem cells. Based on their stemness signature, GBM patients were divided into two subtypes via consensus clustering, and patients in Stemness Subtype I presented significantly better OS but poorer progression-free survival than Stemness Subtype II. Genomic variations revealed patients in Stemness Subtype I had higher somatic mutation loads and copy number alteration burdens. Additionally, two stemness subtypes had distinct tumor immune microenvironment patterns. Tumor Immune Dysfunction and Exclusion and subclass mapping analysis further demonstrated patients in Stemness Subtype I were more likely to respond to immunotherapy, especially anti-PD1 treatment. The pRRophetic algorithm also indicated patients in Stemness Subtype I were more resistant to temozolomide therapy. Finally, multiple machine learning algorithms were used to develop a 7-gene Stemness Subtype Predictor, which were further validated in two external independent GBM cohorts. This novel stemness-based classification could provide a promising prognostic predictor for GBM and may guide physicians in selecting potential responders for preferential use of immunotherapy.
Wang Z
，Wang Y
，Yang T
，Xing H
，Wang Y
，Gao L
，Guo X
，Xing B
，Wang Y
，Ma W
... -
《-》
Predictive value of a stemness-based classifier for prognosis and immunotherapy response of hepatocellular carcinoma based on bioinformatics and machine-learning strategies.
Significant advancements have been made in hepatocellular carcinoma (HCC) therapeutics, such as immunotherapy for treating patients with HCC. However, there is a lack of reliable biomarkers for predicting the response of patients to therapy, which continues to be challenging. Cancer stem cells (CSCs) are involved in the oncogenesis, drug resistance, and invasion, as well as metastasis of HCC cells. Therefore, in this study, we aimed to create an mRNA expression-based stemness index (mRNAsi) model to predict the response of patients with HCC to immunotherapy.
We retrieved gene expression and clinical data of patients with HCC from the GSE14520 dataset and the Cancer Genome Atlas (TCGA) database. Next, we used the "one-class logistic regression (OCLR)" algorithm to obtain the mRNAsi of patients with HCC. We performed "unsupervised consensus clustering" to classify patients with HCC based on the mRNAsi scores and stemness subtypes. The relationships between the mRNAsi model, clinicopathological features, and genetic profiles of patients were compared using various bioinformatic methods. We screened for differentially expressed genes to establish a stemness-based classifier for predicting the patient's prognosis. Next, we determined the effect of risk scores on the tumor immune microenvironment (TIME) and the response of patients to immune checkpoint blockade (ICB). Finally, we used qRT-PCR to investigate gene expression in patients with HCC.
We screened CSC-related genes using various bioinformatics tools in patients from the TCGA-LIHC cohort. We constructed a stemness classifier based on a nine-gene (PPARGC1A, FTCD, CFHR3, MAGEA6, CXCL8, CABYR, EPO, HMMR, and UCK2) signature for predicting the patient's prognosis and response to ICBs. Further, the model was validated in an independent GSE14520 dataset and performed well. Our model could predict the status of TIME, immunogenomic expressions, congenic pathway, and response to chemotherapy drugs. Furthermore, a significant increase in the proportion of infiltrating macrophages, Treg cells, and immune checkpoints was observed in patients in the high-risk group. In addition, tumor cells in patients with high mRNAsi scores could escape immune surveillance. Finally, we observed that the constructed model had a good expression in the clinical samples. The HCC tumor size and UCK2 genes expression were significantly alleviated and decreased, respectively, by treatments of anti-PD1 antibody. We also found knockdown UCK2 changed expressions of immune genes in HCC cell lines.
The novel stemness-related model could predict the prognosis of patients and aid in creating personalized immuno- and targeted therapy for patients in HCC.
Chen E
，Zou Z
，Wang R
，Liu J
，Peng Z
，Gan Z
，Lin Z
，Liu J
... -
《Frontiers in Immunology》
Integrative stemness characteristics associated with prognosis and the immune microenvironment in esophageal cancer.
Cancer stem cells (CSCs) induces tumor metastasis and recurrence. However, the role of CSCs in molding the tumor immune microenvironment (TIME) is largely inexplicit. This study aimed to comprehensively characterize the stemness of esophageal cancer (EC) and correlate the stemness patterns with TIME.
A trained stemness index model was used to score EC patients based on the one-class logistic regression (OCLR) machine-learning algorithm. Gene expression-based stemness index (mRNAsi) and DNA methylation-based stemness index (mDNAsi) were calculated for integrative analyses of EC stemness in the training cohort (n = 182) and validation cohort (n = 179). Intrinsic stemness patterns were estimated to determine its association with clinical features, biological pathways, prognosis, and potential inhibitors. Additionally, the dynamic interplay between EC stemness and TIME was integrally characterized.
Analyses of EC stemness and clinical characteristics indicated that higher-stage and metastatic tumors featured more dedifferentiated phenotypically. Univariate and multivariate Cox regression analyses revealed that mRNAsi was significantly associated with overall survival (OS) of EC patients, whereas no relationship was observed between mDNAsi and OS. Notably, prolonged OS was observed with esophageal squamous cell carcinoma (ESCC) in low versus high mRNAsi groups, whereas the OS was equivalent between the two groups for esophageal adenocarcinoma (ESAD). The mRNAsi may thus recapitulate prognostic molecular subgroups of EC. The prognostic model comprising 14 stemness signatures was constructed using combined Cox and Lasso regression analyses which effectively distinguished individual survival of ESCC in two cohorts. Nevertheless, no significant differences in OS was observed when the same prognostic model of ESCC was applied to ESAD. Gene Set Enrichment Analysis (GSEA) of selected stemness signatures indicated that ESCC stemness is involved in immune-related pathways. Furthermore, ESCC stemness and stemness-related signatures were associated with tumor-infiltrating immune cells, immunoscore, and PD-L1 expression. Compounds specific to the selected stemness signatures were detected using the CMap database.
This study determined integrated characteristics of EC stemness. The identified mRNAsi-based signatures conferred with the predictive ability of personalized ESCC prognosis and highlighted the potential targets for CSC-mediated immunotherapy. Analyses of the interface between ESCC stemness and TIME may help in predicting the efficacy of CSC-specific immunotherapy and provide insight into combinatorial therapy by targeting ESCC stem cells and TIME.
Yi L
，Huang P
，Zou X
，Guo L
，Gu Y
，Wen C
，Wu G
... -
《-》
ResearchGate
(全网免费下载)
钛学术
(全网免费下载)
