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

Background: Early diagnosis of systemic lupus erythematosus (SLE) is crucial for treatment and reducing mortality. In this research, we set out to explore several important biomarkers for the diagnosis of SLE through machine learning combined with pan-cancer, and to investigate the significance of immune cell infiltration. Methods: Gene expression profiles for three human SLE and control samples were downloaded from the GEO database. The LASSO regression model and support vector machine recursive feature elimination (SVM-RFE) analysiswere used to identify candidate biomarkers. The area under the receiver operating characteristic curve (AUC) value was obtained for the evaluation of the discriminations. The GSE20864 dataset was then further validated for the expression and diagnostic value of SLE biomarkers. The structure of 22 immune cell fractions in SLE was explored through the pooled cohort of CIBERSORT. The Cancer Genome Atlas (TCGA), Treatment-Applicable Research to Generate Effective Treatments (TARGET) and Genotype-Tissue Expression (GTEx) datasets are sources of pan-cancer and normal tissue data. And associated pan-cancer analysis was carried out. Results: We identified a total of 52 differently expressed genes, of which 23 were significantly upregulated and 29 were significantly downregulated. After further screening, we found four genes as candidate biomarkers (FOS, IFI27, ANKH, and IGF2BP2). Three signature genes (ANKH, IFI27, IGF2BP2) are excellent diagnostic values in treat and control groups. In immune cell analysis, 11 immune cells showed significant differences between SLE patients and normal humans. We also analysed the effect of IGF2BP2 in the tumour process. The IGF2BP2 gene was low expressed in 9 tumors and high expressed in 22 tumors. In addition, we also found a strong correlation between IGF2BP2 expression in tumour cells and patient prognosis. Our study also indicates that IGF2BP2 may have the ability to predict the efficacy of ICI in the corresponding cancer. Conclusion: We explored several important biomarkers for the diagnosis of SLE through machine learning and found that IGF2BP2 has a sufficient role in the immune microenvironment of most of the tumors.