MX2: Identification and systematic mechanistic analysis of a novel immune-related biomarker for systemic lupus erythematosus

Systemic lupus in the era of machine learning medicine

Artificial intelligence and machine learning applications are emerging as transformative technologies in medicine. With greater access to a diverse range of big datasets, researchers are turning to these powerful techniques for data analysis. Machine learning can reveal patterns and interactions between variables in large and complex datasets more accurately and efficiently than traditional statistical methods. Machine learning approaches open new possibilities for studying SLE, a multifactorial, highly heterogeneous and complex disease. Here, we discuss how machine learning methods are rapidly being integrated into the field of SLE research. Recent reports have focused on building prediction models and/or identifying novel biomarkers using both supervised and unsupervised techniques for understanding disease pathogenesis, early diagnosis and prognosis of disease. In this review, we will provide an overview of machine learning techniques to discuss current gaps, challenges and opportunities for SLE studies. External validation of most prediction models is still needed before clinical adoption. Utilisation of deep learning models, access to alternative sources of health data and increased awareness of the ethics, governance and regulations surrounding the use of artificial intelligence in medicine will help propel this exciting field forward.

Systematic analysis of MASP-1 serves as a novel immune-related biomarker in sepsis and trauma followed by preliminary experimental validation

Background

Dysregulated immune response in trauma and sepsis leads to the abnormal activation of the complement and coagulation systems. Mannose-binding lectin (MBL)-associated serine protease-1 (MASP-1) activates the lectin pathway of the complement system and mediates proinflammatory and procoagulant reactions. However, the potential effects of MASP-1 in trauma and sepsis have not yet been explored.

Methods

We obtained five sepsis, two trauma, and one sepsis and trauma RNA-sequencing dataset from the Gene Expression Omnibus (GEO) database and conducted a comprehensive evaluation of the expression pattern, biological functions, and diagnostic value of MASP-1 in trauma and sepsis. Additionally, we investigated the association between MASP-1 expression and clinicopathological characteristics of trauma and sepsis. Furthermore, we collected clinical specimens to preliminarily validate the expression level and diagnostic efficacy of MASP-1 as well as the correlation of MASP-1 with clinical features of trauma and sepsis. Subsequently, we conducted a correlation analysis among MASP-1, immune cell infiltration, and immune and molecular pathways. Finally, we mechanistically analyzed the relationship among MASP-1, specific immune cells, and pivotal molecular pathways.

Results

MASP-1 expression was significantly upregulated in the trauma/sepsis samples compared to the control samples in the GEO datasets. MASP-1 exhibited excellent diagnostic values (AUC > 0.7) in multiple datasets and at multiple time points and could efficiently distinguish trauma/sepsis samples from the control samples. Moreover, MASP-1 expression was significantly positively correlated with the severity of the disease (APACHE-II, CRP, and neutrophil levels). These results were further validated by real-time quantitative polymerase chain reaction and enzyme-linked immunosorbent assay. Functional enrichment analysis revealed that MASP-1 primarily promotes trauma and sepsis via the immune-related signaling pathway. MASP-1 was significantly correlated with the infiltration of specific immune cells (such as B cells, CD8 T cells, neutrophils, macrophages, and infiltrating lymphocytes) and immune and molecular pathways (such as checkpoint, HLA, IL6/JAK/STAT3 signaling, necrosis, T-cell co-inhibition, and T-cell co-stimulation). Finally, analysis of the transcription and single-cell data revealed that MASP-1 was specifically expressed in T cells, and further correlation analysis revealed a close correlation between MASP-1 expression, proportion of CD8 T cells, and IL6/JAK/STAT3 signaling scores.

Conclusion

Our results suggest that MASP-1 can serve as an immune-related biomarker for the diagnosis and disease severity of trauma and sepsis. It may activate the IL6 JAK-STAT3 signaling pathway and promote CD8 T-cell depletion to trigger traumatic sepsis.

Exploration of programmed cell death-associated characteristics and immune infiltration in neonatal sepsis: new insights from bioinformatics analysis and machine learning

BACKGROUND

Neonatal sepsis, a perilous medical situation, is typified by the malfunction of organs and serves as the primary reason for neonatal mortality. Nevertheless, the mechanisms underlying newborn sepsis remain ambiguous. Programmed cell death (PCD) has a connection with numerous infectious illnesses and holds a significant function in newborn sepsis, potentially serving as a marker for diagnosing the condition.

METHODS

From the GEO public repository, we selected two groups, which we referred to as the training and validation sets, for our analysis of neonatal sepsis. We obtained PCD-related genes from 12 different patterns, including databases and published literature. We first obtained differential expressed genes (DEGs) for neonatal sepsis and controls. Three advanced machine learning techniques, namely LASSO, SVM-RFE, and RF, were employed to identify potential genes connected to PCD. To further validate the results, PPI networks were constructed, artificial neural networks and consensus clustering were used. Subsequently, a neonatal sepsis diagnostic prediction model was developed and evaluated. We conducted an analysis of immune cell infiltration to examine immune cell dysregulation in neonatal sepsis, and we established a ceRNA network based on the identified marker genes.

RESULTS

Within the context of neonatal sepsis, a total of 49 genes exhibited an intersection between the differentially expressed genes (DEGs) and those associated with programmed cell death (PCD). Utilizing three distinct machine learning techniques, six genes were identified as common to both DEGs and PCD-associated genes. A diagnostic model was subsequently constructed by integrating differential expression profiles, and subsequently validated by conducting artificial neural networks and consensus clustering. Receiver operating characteristic (ROC) curves were employed to assess the diagnostic merit of the model, which yielded promising results. The immune infiltration analysis revealed notable disparities in patients diagnosed with neonatal sepsis. Furthermore, based on the identified marker genes, the ceRNA network revealed an intricate regulatory interplay.

CONCLUSION

In our investigation, we methodically identified six marker genes (AP3B2, STAT3, TSPO, S100A9, GNS, and CX3CR1). An effective diagnostic prediction model emerged from an exhaustive analysis within the training group (AUC 0.930, 95%CI 0.887-0.965) and the validation group (AUC 0.977, 95%CI 0.935-1.000).

Exploration of biomarkers for systemic lupus erythematosus by machine-learning analysis

BACKGROUND

In recent years, research on the pathogenesis of systemic lupus erythematosus (SLE) has made great progress. However, the prognosis of the disease remains poor, and high sensitivity and accurate biomarkers are particularly important for the early diagnosis of SLE.

METHODS

SLE patient information was acquired from three Gene Expression Omnibus (GEO) databases and used for differential gene expression analysis, such as weighted gene coexpression network (WGCNA) and functional enrichment analysis. Subsequently, three algorithms, random forest (RF), support vector machine-recursive feature elimination (SVM-REF) and least absolute shrinkage and selection operation (LASSO), were used to analyze the above key genes. Furthermore, the expression levels of the final core genes in peripheral blood from SLE patients were confirmed by real-time quantitative polymerase chain reaction (RT-qPCR) assay.

RESULTS

Five key genes (ABCB1, CD247, DSC1, KIR2DL3 and MX2) were found in this study. Moreover, these key genes had good reliability and validity, which were further confirmed by clinical samples from SLE patients. The receiver operating characteristic curves (ROC) of the five genes also revealed that they had critical roles in the pathogenesis of SLE.

CONCLUSION

In summary, five key genes were obtained and validated through machine-learning analysis, offering a new perspective for the molecular mechanism and potential therapeutic targets for SLE.

B cell subsets-related biomarkers and molecular pathways for systemic lupus erythematosus by transcriptomics analyses

BACKGROUND

Systemic lupus erythematosus (SLE), an autoimmune disease, is characterised by B-cell abnormalities and a loss of tolerance that can produce autoantibody. However, the imperative genes and molecular pathways involved in the change of B cell populations remain unclear.

METHODS

The expression of B cell subsets between SLE and healthy controls (HCs) was detected based on micro-array transcriptome data. The Weighted Gene Co-Expression Network Analysis (WGCNA) further revealed the co-expression modules of naïve and memory B cells. Whereafter, we performed the functional enrichment analysis, Protein-protein interaction (PPI) networks construction and feature selection to screen hub genes. Ultimately, we recruited SLE patients and HCs from the Second Hospital of Shanxi Medical University and further verified these genes in transcriptome sequencing samples.

RESULTS

Total of 1087 SLE patients and 86 HCs constituted in the study. Compared to HCs, the levels of peripheral naïve B cells of SLE patients decreased, while memory B cells increased. WGCNA identified two modules with the highest correlation for the subsequent analysis. The purple module was primarily in connection with naïve B cells, and the GO analysis indicated that these genes were mainly abundant in B cell activation. The blue module relevant to memory B cells was most significantly enriched in the "defence response to virus" correlation pathway. Then we screened six hub genes by PPI and feature selection. Finally, four biomarkers (IFI27, IFITM1, MX2, IRF7) were identified by transcriptome sequencing verification.

CONCLUSION

Our study identified hub genes and key pathways associated with the naïve and memory B cells respectively, which may offer novel insights into the behaviours of B cells and the pathogenesis of SLE.

Identification of predictors for neurological outcome after cardiac arrest in peripheral blood mononuclear cells through integrated bioinformatics analysis and machine learning

Neurological prognostication after cardiac arrest (CA) is important to avoid pursuing futile treatments for poor outcome and inappropriate withdrawal of life-sustaining treatment for good outcome. To predict neurological outcome after CA through biomarkers in peripheral blood mononuclear cells, four datasets were downloaded from the Gene Expression Omnibus database. GSE29546 and GSE74198 were used as training datasets, while GSE92696 and GSE34643 were used as verification datasets. The intersection of differentially expressed genes and hub genes from multiscale embedded gene co-expression network analysis (MEGENA) was utilized in the machine learning screening. Key genes were identified using support vector machine recursive feature elimination (SVM-RFE), least absolute shrinkage and selection operator (LASSO) logistic regression, and random forests (RF). The results were validated using receiver operating characteristic curve analysis. An mRNA-miRNA network was constructed. The distribution of immune cells was evaluated using cell-type identification by estimating relative subsets of RNA transcripts (CIBERSORT). Five biomarkers were identified as predictors for neurological outcome after CA, with an area under the curve (AUC) greater than 0.7: CASP8 and FADD-like apoptosis regulator (CFLAR), human protein kinase X (PRKX), miR-483-5p, let-7a-5p, and let-7c-5p. Interestingly, the combination of CFLAR minus PRKX showed an even higher AUC of 0.814. The mRNA-miRNA network consisted of 30 nodes and 76 edges. Statistical differences were found in immune cell distribution, including neutrophils, NK cells active, NK cells resting, T cells CD4 memory activated, T cells CD4 memory resting, T cells CD8, B cells memory, and mast cells resting between individuals with good and poor neurological outcome after CA. In conclusion, our study identified novel predictors for neurological outcome after CA. Further clinical and laboratory studies are needed to validate our findings.

Machine learning-based integration develops biomarkers initial the crosstalk between inflammation and immune in acute myocardial infarction patients

Great strides have been made in past years toward revealing the pathogenesis of acute myocardial infarction (AMI). However, the prognosis did not meet satisfactory expectations. Considering the importance of early diagnosis in AMI, biomarkers with high sensitivity and accuracy are urgently needed. On the other hand, the prevalence of AMI worldwide has rapidly increased over the last few years, especially after the outbreak of COVID-19. Thus, in addition to the classical risk factors for AMI, such as overwork, agitation, overeating, cold irritation, constipation, smoking, and alcohol addiction, viral infections triggers have been considered. Immune cells play pivotal roles in the innate immunosurveillance of viral infections. So, immunotherapies might serve as a potential preventive or therapeutic approach, sparking new hope for patients with AMI. An era of artificial intelligence has led to the development of numerous machine learning algorithms. In this study, we integrated multiple machine learning algorithms for the identification of novel diagnostic biomarkers for AMI. Then, the possible association between critical genes and immune cell infiltration status was characterized for improving the diagnosis and treatment of AMI patients.