Integrated Bioinformatics and Machine Learning Analysis Identify ACADL as a Potent Biomarker of Reactive Mesothelial Cells

Abstract

Mesothelial cells with reactive hyperplasia are difficult to distinguish from malignant mesothelioma cells based on cell morphology. This study aimed to identify and validate potential biomarkers that distinguish mesothelial cells from mesothelioma cells through machine learning combined with immunohistochemistry. It integrated the gene expression matrix from three Gene Expression Omnibus data sets (GSE2549, GSE12345, and GSE51024) to analyze the differently expressed genes between normal and mesothelioma tissues. Then, three machine learning algorithms, least absolute shrinkage and selection operator, support vector machine recursive feature elimination, and random forest were used to screen and obtain four shared candidate markers, including ACADL, EMP2, GPD1L, and HMMR. The receiver operating characteristic curve analysis showed that the area under the curve for distinguishing normal mesothelial cells from mesothelioma was 0.976, 0.943, 0.962, and 0.956, respectively. The expression and diagnostic performance of these candidate genes were validated in two additional independent data sets (GSE42977 and GSE112154), indicating that the performances of ACADL, GPD1L, and HMMR were consistent between the training and validation data sets. Finally, the optimal candidate marker ACADL was verified by immunohistochemistry assay. Acyl-CoA dehydrogenase long chain (ACADL) was stained strongly in mesothelial cells, especially for reactive hyperplasic mesothelial cells, but was negative in malignant mesothelioma cells. Therefore, ACADL has the potential to be used as a specific marker of reactive hyperplasic mesothelial cells in the differential diagnosis of mesothelioma.