Machine learning algorithm-generated and multi-center validated melanoma prognostic signature with inspiration for treatment management

Integrative single-cell and bulk transcriptome analyses identify a distinct pro-tumor macrophage signature that has a major prognostic impact on glioblastomas

Glioblastoma (GBM) is a highly heterogeneous disease with poor clinical outcomes. To comprehensively dissect the molecular landscape of GBM and heterogeneous macrophage clusters in the progression of GBM, this study integrates single-cell and bulk transcriptome data to recognize a distinct pro-tumor macrophage cluster significantly associated with the prognosis of GBM and develop a GBM prognostic signature to facilitate prior subtypes. Leveraging glioma single-cell sequencing data, we identified a novel pro-tumor macrophage subgroup, marked by S100A9, which might interact with endothelial cells to facilitate tumor progression via angiogenesis. To further benefit clinical application, a prognostic signature was established with the genes associated with pro-tumor macrophages. Patients classified within the high-risk group characterized with enrichment in functions related to tumor progression, including epithelial-mesenchymal transition and hypoxia, displays elevated mutations in the TERT promoter region, reduced methylation in the MGMT promoter region, poorer prognoses, and diminished responses to temozolomide therapy, thus effectively discriminating between the prognostic outcomes of GBM patients. Our research sheds light on the intricate microenvironment of gliomas and identifies potential molecular targets for the development of novel therapeutic approaches.

MOCS, a novel classifier system integrated multimoics analysis refining molecular subtypes and prognosis for skin melanoma

PURPOSE

The present investigation focuses on Skin Cutaneous Melanoma (SKCM), a melanocytic carcinoma characterized by marked aggression, significant heterogeneity, and a complex etiological background, factors which collectively contribute to the challenge in prognostic determinations. We defined a novel classifier system specifically tailored for SKCM based on multiomics.

METHODS

We collected 423 SKCM samples with multi omics datasets to perform a consensus cluster analysis using 10 machine learning algorithms and verified in 2 independent cohorts. Clinical features, biological characteristics, immune infiltration pattern, therapeutic response and mutation landscape were compared between subtypes.

RESULTS

Based on consensus clustering algorithms, we identified two Multi-Omics-Based-Cancer-Subtypes (MOCS) in SKCM in TCGA project and validated in GSE19234 and GSE65904 cohorts. MOCS2 emerged as a subtype with poor prognosis, characterized by a complex immune microenvironment, dysfunctional anti-tumor immune state, high cancer stemness index, and genomic instability. MOCS2 exhibited resistance to chemotherapy agents like erlotinib and sunitinib while sensitive to rapamycin, NSC87877, MG132, and FH355. Additionally, ELSPBP1 was identified as the target involving in glycolysis and M2 macrophage infiltration in SKCM.

CONCLUSIONS

MOCS classification could stably predict prognosis of SKCM; patients with a high cancer stemness index combined with genomic instability may be predisposed to an immune exhaustion state.Communicated by Ramaswamy H. Sarma.

Development of a prognostic Neutrophil Extracellular Traps related lncRNA signature for soft tissue sarcoma using machine learning

Background

Soft tissue sarcoma (STS) is a highly heterogeneous musculoskeletal tumor with a significant impact on human health due to its high incidence and malignancy. Long non-coding RNA (lncRNA) and Neutrophil Extracellular Traps (NETs) have crucial roles in tumors. Herein, we aimed to develop a novel NETsLnc-related signature using machine learning algorithms for clinical decision-making in STS.

Methods

We applied 96 combined frameworks based on 10 different machine learning algorithms to develop a consensus signature for prognosis and therapy response prediction. Clinical characteristics, univariate and multivariate analysis, and receiver operating characteristic curve (ROC) analysis were used to evaluate the predictive performance of our models. Additionally, we explored the biological behavior, genomic patterns, and immune landscape of distinct NETsLnc groups. For patients with different NETsLnc scores, we provided information on immunotherapy responses, chemotherapy, and potential therapeutic agents to enhance the precision medicine of STS. Finally, the gene expression was validated through real-time quantitative PCR (RT-qPCR).

Results

Using the weighted gene co-expression network analysis (WGCNA) algorithm, we identified NETsLncs. Subsequently, we constructed a prognostic NETsLnc signature with the highest mean c-index by combining machine learning algorithms. The NETsLnc-related features showed excellent and stable performance for survival prediction in STS. Patients in the low NETsLnc group, associated with improved prognosis, exhibited enhanced immune activity, immune infiltration, and tended toward an immunothermal phenotype with a potential immunotherapy response. Conversely, patients with a high NETsLnc score showed more frequent genomic alterations and demonstrated a better response to vincristine treatment. Furthermore, RT-qPCR confirmed abnormal expression of several signature lncRNAs in STS.

Conclusion

In conclusion, the NETsLnc signature shows promise as a powerful approach for predicting the prognosis of STS. which not only deepens our understanding of STS but also opens avenues for more targeted and effective treatment strategies.

Artificial intelligence for predictive biomarker discovery in immuno-oncology: a systematic review

BACKGROUND

The widespread use of immune checkpoint inhibitors (ICIs) has revolutionised treatment of multiple cancer types. However, selecting patients who may benefit from ICI remains challenging. Artificial intelligence (AI) approaches allow exploitation of high-dimension oncological data in research and development of precision immuno-oncology.

MATERIALS AND METHODS

We conducted a systematic literature review of peer-reviewed original articles studying the ICI efficacy prediction in cancer patients across five data modalities: genomics (including genomics, transcriptomics, and epigenomics), radiomics, digital pathology (pathomics), and real-world and multimodality data.

RESULTS

A total of 90 studies were included in this systematic review, with 80% published in 2021-2022. Among them, 37 studies included genomic, 20 radiomic, 8 pathomic, 20 real-world, and 5 multimodal data. Standard machine learning (ML) methods were used in 72% of studies, deep learning (DL) methods in 22%, and both in 6%. The most frequently studied cancer type was non-small-cell lung cancer (36%), followed by melanoma (16%), while 25% included pan-cancer studies. No prospective study design incorporated AI-based methodologies from the outset; rather, all implemented AI as a post hoc analysis. Novel biomarkers for ICI in radiomics and pathomics were identified using AI approaches, and molecular biomarkers have expanded past genomics into transcriptomics and epigenomics. Finally, complex algorithms and new types of AI-based markers, such as meta-biomarkers, are emerging by integrating multimodal/multi-omics data.

CONCLUSION

AI-based methods have expanded the horizon for biomarker discovery, demonstrating the power of integrating multimodal data from existing datasets to discover new meta-biomarkers. While most of the included studies showed promise for AI-based prediction of benefit from immunotherapy, none provided high-level evidence for immediate practice change. A priori planned prospective trial designs are needed to cover all lifecycle steps of these software biomarkers, from development and validation to integration into clinical practice.

CRISPR-Cas9 identifies growth-related subtypes of glioblastoma with therapeutical significance through cell line knockdown

BACKGROUND

Glioblastoma (GBM) is a type of highly malignant brain tumor that is known for its significant intratumoral heterogeneity, meaning that there can be a high degree of variability within the tumor tissue. Despite the identification of several subtypes of GBM in recent years, there remains to explore a classification based on genes related to proliferation and growth.

METHODS

The growth-related genes of GBM were identified by CRISPR-Cas9 and univariate Cox regression analysis. The expression of these genes in the Cancer Genome Atlas cohort (TCGA) was used to construct growth-related genes subtypes (GGSs) via consensus clustering. Validation of this subtyping was performed using the nearest template prediction (NTP) algorithm in two independent Gene Expression Omnibus (GEO) cohorts and the ZZ cohort. Additionally, copy number variations, biological functions, and potential drugs were analyzed for each of the different subtypes separately.

RESULTS

Our research established multicenter-validated GGSs. GGS1 exhibits the poorest prognosis, with the highest frequency of chr 7 gain & chr 10 loss, and the lowest frequency of chr 19 & 20 co-gain. Additionally, GGS1 displays the highest expression of EGFR. Furthermore, it is significantly enriched in metabolic, stemness, proliferation, and signaling pathways. Besides we showed that Foretinib may be a potential therapeutic agent for GGS1, the worst prognostic subtype, through data screening and in vitro experiments. GGS2 has a moderate prognosis, with a slightly higher proportion of chr 7 gain & chr 10 loss, and the highest proportion of chr 19 & 20 co-gain. The prognosis of GGS3 is the best, with the least chr 7 gain & 10 loss and EGFR expression.

CONCLUSIONS

These results enhance our understanding of the heterogeneity of GBM and offer insights for stratified management and precise treatment of GBM patients.