Machine learning-based integrative analysis of methylome and transcriptome identifies novel prognostic DNA methylation signature in uveal melanoma

Study on the mechanism of S100A4-mediated cancer oncogenesis in uveal melanoma cells through the integration of bioinformatics and in vitro experiments

BACKGROUND

The elevated metastasis rate of uveal melanoma (UM) is intricately correlated with patient prognosis, significantly affecting the quality of life. S100 calcium-binding protein A4 (S100A4) has tumorigenic properties; therefore, the present study investigated the impact of S100A4 on UM cell proliferation, apoptosis, migration, and invasion using bioinformatics and in vitro experiments.

METHODS

Bioinformatic analysis was used to screen S100A4 as a hub gene and predict its possible mechanism in UM cells, and the S100A4 silencing cell line was constructed. The impact of S100A4 silencing on the proliferative ability of UM cells was detected using the Cell Counting Kit-8 and colony formation assays. Annexin V-FITC/PI double fluorescence and Hoechst 33342 staining were used to observe the effects of apoptosis on UM cells. The effect of S100A4 silencing on the migratory and invasive capabilities of UM cells was assessed using wound healing and Transwell assays. Western blotting was used to detect the expression of related proteins.

RESULTS

The present study found that S100A4 is a biomarker of UM, and its high expression is related to poor prognosis. After constructing the S100A4 silencing cell line, cell viability, clone number, proliferating cell nuclear antigen, X-linked inhibitor of apoptosis protein, and survivin expression were decreased in UM cells. The cell apoptosis rate and relative fluorescence intensity increased, accompanied by increased levels of Bax and caspase-3 and decreased levels of Bcl-2. Additionally, a decrease in the cell migration index and relative invasion rate was observed with increased E-cadherin expression and decreased N-cadherin and vimentin protein expression.

CONCLUSION

S100A4 silencing can inhibit the proliferation, migration, and invasion and synchronously induces apoptosis in UM cells.

Single-cell characterization of macrophages in uveal melanoma uncovers transcriptionally heterogeneous subsets conferring poor prognosis and aggressive behavior

Uveal melanoma (UM) is the most frequent primary intraocular malignancy with high metastatic potential and poor prognosis. Macrophages represent one of the most abundant infiltrating immune cells with diverse functions in cancers. However, the cellular heterogeneity and functional diversity of macrophages in UM remain largely unexplored. In this study, we analyzed 63,264 single-cell transcriptomes from 11 UM patients and identified four transcriptionally distinct macrophage subsets (termed MΦ-C1 to MΦ-C4). Among them, we found that MΦ-C4 exhibited relatively low expression of both M1 and M2 signature genes, loss of inflammatory pathways and antigen presentation, instead demonstrating enhanced signaling for proliferation, mitochondrial functions and metabolism. We quantified the infiltration abundance of MΦ-C4 from single-cell and bulk transcriptomes across five cohorts and found that increased MΦ-C4 infiltration was relevant to aggressive behaviors and may serve as an independent prognostic indicator for poor outcomes. We propose a novel subtyping scheme based on macrophages by integrating the transcriptional signatures of MΦ-C4 and machine learning to stratify patients into MΦ-C4-enriched or MΦ-C4-depleted subtypes. These two subtypes showed significantly different clinical outcomes and were validated through bulk RNA sequencing and immunofluorescence assays in both public multicenter cohorts and our in-house cohort. Following further translational investigation, our findings highlight a potential therapeutic strategy of targeting macrophage subsets to control metastatic disease and consistently improve the outcome of patients with UM.

Accurate prediction of pan-cancer types using machine learning with minimal number of DNA methylation sites

DNA methylation analysis has been applied to determine the primary site of cancer; however, robust and accurate prediction of cancer types with a minimum number of sites is still a significant scientific challenge. To build an accurate and robust cancer type prediction tool with a minimum number of DNA methylation sites, we internally benchmarked different DNA methylation site selection and ranking procedures, as well as different classification models. We used The Cancer Genome Atlas dataset (26 cancer types with 8296 samples) to train and test models and used an independent dataset (17 cancer types with 2738 samples) for model validation. A deep neural network model using a combined feature selection procedure (named MethyDeep) can predict 26 cancer types using 30 methylation sites with superior performance compared with the known methods for both primary and metastatic cancers in independent validation datasets. In conclusion, MethyDeep is an accurate and robust cancer type predictor with the minimum number of DNA methylation sites; it could help the cost-effective clarification of cancer of unknown primary patients and the liquid biopsy-based early screening of cancers.

Identification and validation of immunogenic cell death-related score in uveal melanoma to improve prediction of prognosis and response to immunotherapy

BACKGROUND

Immunogenic cell death (ICD) could activate innate and adaptive immune response. In this work, we aimed to develop an ICD-related signature in uveal melanoma (UVM) patients and facilitate assessment of their prognosis and immunotherapy.

METHODS

A set of machine learning methods, including non-negative matrix factorization (NMF) method and least absolute shrinkage and selection operator (LASSO) logistic regression model, and bioinformatics analytic tools were integrated to construct an ICD-related risk score (ICDscore). CIBERSORT and ESTIMATE algorithms were used to evaluate the infiltration of immune cells. The Genomics of Drug Sensitivity in Cancer (GDSC), cellMiner and tumor immune dysfunction and exclusion (TIDE) databases were used for therapy sensitivity analyses. The predictive performance between ICDscore with other mRNA signatures was also compared.

RESULTS

The ICDscore could predict the prognosis of UVM patients in both the training and four validating cohorts. The ICDscore outperformed 19 previously published signatures. Patients with high ICDscore exhibited a substantial increase in immune cell infiltration and expression of immune checkpoint inhibitor-related genes, leading to a higher response rate to immunotherapy. Furthermore, the downregulation of poly (ADP-ribose) polymerase family member 8 (PARP8), a critical gene involved in the development of the ICDscore, resulted in decreased cell proliferation and slower migration of UVM cells.

CONCLUSION

In conclusion, we developed a robust and powerful ICD-related signature for evaluating the prognosis and benefits of immunotherapy that could serve as a promising tool to guide decision-making and surveillance for UVM patients.

Identification of a novel immune-related gene signature for prognosis and the tumor microenvironment in patients with uveal melanoma combining single-cell and bulk sequencing data

Introduction

Uveal melanoma (UVM) is the most invasive intraocular malignancy in adults with a poor prognosis. Growing evidence revealed that immune-related gene is related to tumorigenesis and prognosis. This study aimed to construct an immune-related prognostic signature for UVM and clarify the molecular and immune classification.

Methods

Based on The Cancer Genome Atlas (TCGA) database, single-sample gene set enrichment (ssGSEA) and hierarchical clustering analysis were performed to identify the immune infiltration pattern of UVM and classify patients into two immunity clusters. Then, we proposed univariate and multivariate Cox regression analysis to identify immune-related genes that related to overall survival (OS) and validated in the Gene Expression Omnibus (GEO) external validation cohort. The molecular and immune classification in the immune-related gene prognostic signature defined subgroups were analyzed.

Results

The immune-related gene prognostic signature was constructed based on S100A13, MMP9, and SEMA3B genes. The prognostic value of this risk model was validated in three bulk RNA sequencing datasets and one single-cell sequencing dataset. Patients in the low-risk group had better OS than those in the high-risk group. The receiver-operating characteristic (ROC) analysis revealed its strong predictive ability for UVM patients. Lower expression of immune checkpoint genes was presented in the low-risk group. Functional studies showed that S100A13 knockdown via siRNA inhibited UVM cell proliferation, migration, and invasion in vitro, with the increased expression of reactive oxygen species (ROS) related markers in UVM cell lines.

Discussion

The immune-related gene prognostic signature is an independent predictive factor for the survival of patients with UVM and provides new information about cancer immunotherapy in UVM.

Genome-wide DNA methylation profile analysis identifies an individualized predictive signature for melanoma immune response

PURPOSE

The current evaluation methods for tumor infiltrating lymphocytes (TILs), particularly CD8 + TILs, mainly rely on semiquantitative immunohistochemistry with high variability. We aimed to construct an individualized DNA methylation-based signature for CD8 + TILs (CD8 + MeTIL) that may characterize melanoma immune microenvironment and guide therapeutic selection.

METHODS

The transcriptome profiles and DNA methylation data of 457 melanoma patients from The Cancer Genome Atlas (TCGA) database were analyzed. Differential methylation analysis between groups with high and low CD8 + TILs was performed to select differentially methylated positions (DMPs) and define CD8 + MeTIL. The prognostic value of CD8 + MeTIL and its predictive value for immunotherapy response were investigated using multiple melanoma cohorts.

RESULTS

We successfully constructed the CD8 + MeTIL signature based on four DMPs. The survival analyses showed that higher CD8 + MeTIL score was associated with worse survival outcomes in TCGA-SKCM and GSE144487 cohorts. The ROC curve for the predictive analysis revealed that the survival prediction of CD8 + MeTIL score was superior compared with CD8 + TILs (CIBERSORT) and CD8B mRNA expression. Furthermore, we founded that tumors with higher CD8 + MeTIL score were marked with immunosuppressive characteristics, including low immune score and downregulated immune-related pathways. More importantly, the CD8 + MeTIL score showed a potential predictive value for the benefit from immunotherapy in two published cohorts. When combined CD8 + MeTIL with PD-L1 expression, the patient classification showed significantly different immunotherapy response rates and long-term survival outcomes.

CONCLUSIONS

The CD8 + MeTIL signature might be as a novel method to evaluate CD8 + TILs and guide immunotherapy approaches.

Monoacylglycerol lipase regulates macrophage polarization and cancer progression in uveal melanoma and pan-cancer

Background

Although lipid metabolism has been proven to play a key role in the development of cancer, its significance in uveal melanoma (UM) has not yet been elucidated in the available literature.

Methods

To identify the expression patterns of lipid metabolism in 80 UM patients from the TCGA database, 47 genes involved in lipid metabolism were analyzed. Consensus clustering revealed two distinct molecular groups. ESTIMATE, TIMER, and ssGSEA analyses were done to identify the differences between the two subgroups in tumor microenvironment (TME) and immune state. Using Cox regression and Lasso regression analysis, a risk model based on differentially expressed genes (DEGs) was developed. To validate the expression of monoacylglycerol lipase (MGLL) and immune infiltration in diverse malignancies, a pan-cancer cohort from the UCSC database was utilized. Next, a single-cell sequencing analysis on UM patients from the GEO data was used to characterize the lipid metabolism in TME and the role of MGLL in UM. Finally, in vitro investigations were utilized to study the involvement of MGLL in UM.

Results

Two molecular subgroups of UM patients have considerably varied survival rates. The majority of DEGs between the two subgroups were associated with immune-related pathways. Low immune scores, high tumor purity, a low number of immune infiltrating cells, and a comparatively low immunological state were associated with a more favorable prognosis. An examination of GO and KEGG data demonstrated that the risk model based on genes involved with lipid metabolism can accurately predict survival in patients with UM. It has been demonstrated that MGLL, a crucial gene in this paradigm, promotes the proliferation, invasion, and migration of UM cells. In addition, we discovered that MGLL is strongly expressed in macrophages, specifically M2 macrophages, which may play a function in the M2 polarization of macrophages and M2 macrophage activation in cancer cells.

Conclusion

This study demonstrates that the risk model based on lipid metabolism may be useful for predicting the prognosis of patients with UM. By promoting macrophage M2 polarization, MGLL contributes to the evolution of malignancy in UM, suggesting that it may be a therapeutic target for UM.

Melanoma 2.0. Skin cancer as a paradigm for emerging diagnostic technologies, computational modelling and artificial intelligence

We live in an unprecedented time in oncology. We have accumulated samples and cases in cohorts larger and more complex than ever before. New technologies are available for quantifying solid or liquid samples at the molecular level. At the same time, we are now equipped with the computational power necessary to handle this enormous amount of quantitative data. Computational models are widely used helping us to substantiate and interpret data. Under the label of systems and precision medicine, we are putting all these developments together to improve and personalize the therapy of cancer. In this review, we use melanoma as a paradigm to present the successful application of these technologies but also to discuss possible future developments in patient care linked to them. Melanoma is a paradigmatic case for disruptive improvements in therapies, with a considerable number of metastatic melanoma patients benefiting from novel therapies. Nevertheless, a large proportion of patients does not respond to therapy or suffers from adverse events. Melanoma is an ideal case study to deploy advanced technologies not only due to the medical need but also to some intrinsic features of melanoma as a disease and the skin as an organ. From the perspective of data acquisition, the skin is the ideal organ due to its accessibility and suitability for many kinds of advanced imaging techniques. We put special emphasis on the necessity of computational strategies to integrate multiple sources of quantitative data describing the tumour at different scales and levels.

Pan-Cancer Methylated Dysregulation of Long Non-coding RNAs Reveals Epigenetic Biomarkers

Different cancer types not only have common characteristics but also have their own characteristics respectively. The mechanism of these specific and common characteristics is still unclear. Pan-cancer analysis can help understand the similarities and differences among cancer types by systematically describing different patterns in cancers and identifying cancer-specific and cancer-common molecular biomarkers. While long non-coding RNAs (lncRNAs) are key cancer modulators, there is still a lack of pan-cancer analysis for lncRNA methylation dysregulation. In this study, we integrated lncRNA methylation, lncRNA expression and mRNA expression data to illuminate specific and common lncRNA methylation patterns in 23 cancer types. Then, we screened aberrantly methylated lncRNAs that negatively regulated lncRNA expression and mapped them to the ceRNA relationship for further validation. 29 lncRNAs were identified as diagnostic biomarkers for their corresponding cancer types, with lncRNA AC027601 was identified as a new KIRC-associated biomarker, and lncRNA ACTA2-AS1 was regarded as a carcinogenic factor of KIRP. Two lncRNAs HOXA-AS2 and AC007228 were identified as pan-cancer biomarkers. In general, the cancer-specific and cancer-common lncRNA biomarkers identified in this study may aid in cancer diagnosis and treatment.

A novel DNA methylation-driver gene signature for long-term survival prediction of hepatitis-positive hepatocellular carcinoma patients

BACKGROUND

Abnormal DNA methylation is one of the most general epigenetic modifications in hepatocellular carcinoma (HCC). Recent research showed that DNA methylation was a prognostic indicator of all-cause HCC and nonviral HCC. However, whether DNA methylation-driver genes could be used for predicting survival, the probability of hepatitis-positive HCC remains unclear.

METHODS

In this study, DNA methylation-driver genes (MDGs) were screened by a joint analysis of methylome and transcriptome data of 142 hepatitis-positive HCC patients. Subsequently, a prognostic risk score and nomogram were constructed. Finally, correlation analyses between the risk score and signaling pathways and immunity were conducted by GSVA and CIBERSORT.

RESULTS

Through random forest screening and Cox progression analysis, 10 prognostic methylation-driver genes (AC008271.1, C11orf53, CASP8, F2RL2, GBP5, LUCAT1, RP11-114B7.6, RP11-149I23.3, RP11-383 J24.1, and SLC35G2) were screened out. As a result, a prognostic risk score signature was constructed. The independent value of the risk score for prognosis prediction were addressed in the TCGA-HCC and the China-HCC cohorts. Next, clinicopathological features were analyzed and HBV status and histological grade were screened to construct a nomogram together with the risk score. The prognostic efficiency of the nomogram was validated by the calibration curves and the concordance index (C index: 0.829, 95% confidence interval: 0.794-0.864), while its clinical application ability was confirmed by decision curve analysis (DCA). At last, the relationship between the risk score and signaling pathways, as well as the correlations between immune cells were elucidated preliminary.

CONCLUSIONS

Taken together, our study explored a novel DNA methylation-driver gene risk score signature and an efficient nomogram for long-term survival prediction of hepatitis-positive HCC patients.

Novel technical and privacy-preserving technology for artificial intelligence in ophthalmology

PURPOSE OF REVIEW

The application of artificial intelligence (AI) in medicine and ophthalmology has experienced exponential breakthroughs in recent years in diagnosis, prognosis, and aiding clinical decision-making. The use of digital data has also heralded the need for privacy-preserving technology to protect patient confidentiality and to guard against threats such as adversarial attacks. Hence, this review aims to outline novel AI-based systems for ophthalmology use, privacy-preserving measures, potential challenges, and future directions of each.

RECENT FINDINGS

Several key AI algorithms used to improve disease detection and outcomes include: Data-driven, imagedriven, natural language processing (NLP)-driven, genomics-driven, and multimodality algorithms. However, deep learning systems are susceptible to adversarial attacks, and use of data for training models is associated with privacy concerns. Several data protection methods address these concerns in the form of blockchain technology, federated learning, and generative adversarial networks.

SUMMARY

AI-applications have vast potential to meet many eyecare needs, consequently reducing burden on scarce healthcare resources. A pertinent challenge would be to maintain data privacy and confidentiality while supporting AI endeavors, where data protection methods would need to rapidly evolve with AI technology needs. Ultimately, for AI to succeed in medicine and ophthalmology, a balance would need to be found between innovation and privacy.

SLC25A38 as a novel biomarker for metastasis and clinical outcome in uveal melanoma

Risk of metastasis is increased by the presence of chromosome 3 monosomy in uveal melanoma (UM). This study aimed to identify more accurate biomarker for risk of metastasis in UM. A total of 80 patients with UM from TCGA were assigned to two groups based on the metastatic status, and bioinformatic analyses were performed to search for critical genes for risk of metastasis. SLC25A38, located on chromosome 3, was the dominant downregulated gene in metastatic UM patients. Low expression of SLC25A38 was an independent predictive and prognostic factor in UM. The predictive potential of SLC25A38 expression was superior to that of pervious reported biomarkers in both TCGA cohort and GSE22138 cohort. Subsequently, its role in promoting metastasis was explored in vitro and in vivo. Knock-out of SLC25A38 could enhance the migration ability of UM cells, and promote distant metastasis in mice models. Through the inhibition of CBP/HIF-mediated pathway followed by the suppression of pro-angiogenic factors, SLC25A38 was situated upstream of metastasis-related pathways, especially angiogenesis. Low expression of SLC25A38 promotes angiogenesis and metastasis, and identifies increased metastatic risk and worse survival in UM patients. This finding may further improve the accuracy of prognostic prediction for UM.

Characterization of DNA Methylation and Screening of Epigenetic Markers in Polycystic Ovary Syndrome

Polycystic ovary syndrome (PCOS) is a heterogeneous endocrine and metabolic disorder in women, which is characterized by androgen excess, ovulation dysfunction, and polycystic ovary. Although the etiology of PCOS is largely unknown, many studies suggest that aberrant DNA methylation is an important contributing factor for its pathological changes. In this study, we investigated DNA methylation characteristics and their impact on gene expression in granulosa cells obtained from PCOS patients. Transcriptome analysis found that differentially expressed genes were mainly enriched in pathways of insulin resistance, fat cell differentiation, and steroid metabolism in PCOS. Overall DNA methylation level in granulosa cells was reduced in PCOS, and the first introns were found to be the major genomic regions that were hypomethylated in PCOS. Integrated analysis of transcriptome, DNA methylation, and miRNAs in ovarian granulosa cells revealed a DNA methylation and miRNA coregulated network and identified key candidate genes for pathogenesis of PCOS, including BMP4, ETS1, and IRS1. Our study shed more light on epigenetic mechanism of PCOS and provided valuable reference for its diagnosis and treatment.

Integrative Analysis of DNA Methylation Data and Transcriptome Data Identified a DNA Methylation-Dysregulated Four-LncRNA Signature for Predicting Prognosis in Head and Neck Squamous Cell Carcinoma

Increasing evidence has demonstrated the crosstalk between DNA epigenetic alterations and aberrant expression of long non-coding RNAs (lncRNAs) during carcinogenesis. However, epigenetically dysregulated lncRNAs and their functional and clinical roles in Head and Neck Squamous Cell Carcinoma (HNSCC) are still not explored. In this study, we performed an integrative analysis of DNA methylation data and transcriptome data and identified a DNA methylation-dysregulated four-lncRNA signature (DNAMeFourLncSig) from 596 DNA methylation-dysregulated lncRNAs using a machine-learning-based feature selection method, which classified the patients of the discovery cohort into two risk groups with significantly different survival including overall survival, disease-specific survival, and progression-free survival. Then the DNAMeFourLncSig was implemented to another two HNSCC patient cohorts and showed similar prognostic values in both. Results from multivariable Cox regression analysis revealed that the DNAMeFourLncSig might be an independent prognostic factor. Furthermore, the DNAMeFourLncSig was substantially correlated with the complete response rate of chemotherapy and may predict chemotherapy response. Functional in silico analysis found that DNAMeFourLncSig-related mRNAs were mainly enriched in cell differentiation, tissue development and immune-related pathways. Overall, our study will improve our understanding of underlying transcriptional and epigenetic mechanisms in HNSCC carcinogenesis and provided a new potential biomarker for the prognosis of patients with HNSCC.