Screening Biomarkers for Systemic Lupus Erythematosus Based on Machine Learning and Exploring Their Expression Correlations With the Ratios of Various Immune Cells

Abstract

BackgroundSystemic lupus erythematosus (SLE) is an autoimmune illness caused by a malfunctioning immunomodulatory system. China has the second highest prevalence of SLE in the world, from 0.03% to 0.07%. SLE is diagnosed using a combination of immunological markers, clinical symptoms, and even invasive biopsy. As a result, genetic diagnostic biomarkers for SLE diagnosis are desperately needed.MethodFrom the Gene Expression Omnibus (GEO) database, we downloaded three array data sets of SLE patients’ and healthy people’s peripheral blood mononuclear cells (PBMC) (GSE65391, GSE121239 and GSE61635) as the discovery metadata (nSLE = 1315, nnormal = 122), and pooled four data sets (GSE4588, GSE50772, GSE99967, and GSE24706) as the validate data set (nSLE = 146, nnormal = 76). We screened the differentially expressed genes (DEGs) between the SLE and control samples, and employed the least absolute shrinkage and selection operator (LASSO) regression, and support vector machine recursive feature elimination (SVM-RFE) analyze to discover possible diagnostic biomarkers. The candidate markers’ diagnostic efficacy was assessed using the receiver operating characteristic (ROC) curve. The reverse transcription quantitative polymerase chain reaction (RT-qPCR) was utilized to confirm the expression of the putative biomarkers using our own Chinese cohort (nSLE = 13, nnormal = 10). Finally, the proportion of 22 immune cells in SLE patients was determined using the CIBERSORT algorithm, and the correlations between the biomarkers’ expression and immune cell ratios were also investigated.ResultsWe obtained a total of 284 DEGs and uncovered that they were largely involved in several immune relevant pathways, such as type І interferon signaling pathway, defense response to virus, and inflammatory response. Following that, six candidate diagnostic biomarkers for SLE were selected, namely ABCB1, EIF2AK2, HERC6, ID3, IFI27, and PLSCR1, whose expression levels were validated by the discovery and validation cohort data sets. As a signature, the area under curve (AUC) values of these six genes reached to 0.96 and 0.913, respectively, in the discovery and validation data sets. After that, we checked to see if the expression of ABCB1, IFI27, and PLSCR1 in our own Chinese cohort matched that of the discovery and validation sets. Subsequently, we revealed the potentially disturbed immune cell types in SLE patients using the CIBERSORT analysis, and uncovered the most relevant immune cells with the expression of ABCB1, IFI27, and PLSCR1.ConclusionOur study identified ABCB1, IFI27, and PLSCR1 as potential diagnostic genes for Chinese SLE patients, and uncovered their most relevant immune cells. The findings in this paper provide possible biomarkers for diagnosing Chinese SLE patients.

Alterations of the gut virome in patients with systemic lupus erythematosus

“…Until now, many attempts to screen SLE biomarkers have been made. Based on bacterial 16S rRNA gene ( 56 ) and various immune cells ( 57 ), 0.79 and 0.96 of AUCs were achieved, respectively. We obtained a more accurate and higher AUC of 0.98 ( Figure 5 ) using only 426 gut vOTUs.…”

Identification of Predicting Diagnostic Gene Biomarkers through Machine Learning combined with Pan-cancer in Patients with Systemic Lupus Erythematosus

“…SLE is a common autoimmune disease, often associated with vital organ involvement. Although symptoms and signs are the basis for the diagnostic and disease assessment of SLE, due to the complexity and diversity of its clinical manifestations, early de nitive diagnosis, disease assessment and treatment monitoring still require reference to laboratory indicators [13,24]. Therefore, the search for genetic markers is important for the diagnosis of SLE.…”

Editorial: New biomarkers for the diagnosis and treatment of systemic lupus erythematosus

“…Patients with LN exhibit a profound depletion of atypical age-associated B-cell (ABC) like CD11c + T-bet + CD21 hi B cells in comparison with healthy individuals and SLE patients without LN ( 19 ). Selected from 284 DEGs identified in two independent SLE cohorts in the Gene Expression Omnibus (GEO) database, machine learning validated ABCB1, IFI27, and PLSCR1 as top predictors of SLE in a Chinese validation cohort of patients over ethnically matched controls ( 13 ). Expression of each these genes was correlated with the expansion of pro-inflammatory lineages of the adaptive and innate immune systems ( 13 ).…”

“…Selected from 284 DEGs identified in two independent SLE cohorts in the Gene Expression Omnibus (GEO) database, machine learning validated ABCB1, IFI27, and PLSCR1 as top predictors of SLE in a Chinese validation cohort of patients over ethnically matched controls ( 13 ). Expression of each these genes was correlated with the expansion of pro-inflammatory lineages of the adaptive and innate immune systems ( 13 ). Separately, a comprehensive analysis of gut microbiome genome databases newly identified Bacilli and Lactobacillales as promoters of SLE and Bacillales , Coprobacter and Lachnospira as protectors from SLE ( 9 ).…”

“…Therefore, integrating biomarkers into the signaling networks that connect the adaptive and innate arms of a dysfunctional immune system is critical for appreciating their significance for controlling pathogenesis, predicting flares or serving as target for treatment in SLE. Notably, only few of these studies involved cellular biomarkers ( 13 , 19 ) that can be connected to central pathways of lupus pathogenesis. Nevertheless, certain easily detectable biomarkers may also serve other purposes, such as sensing organ damage and obviating the need for invasive procedures, i.e., renal biopsy in LN ( 15 , 17 – 19 ).…”