Identification of novel tumor-associated antigens and evaluation of a panel of autoantibodies in detecting oral cancer

Abstract

Background: We aimed to identify tumor-associated antigen (TAA) biomarkers through bioinformatic analysis and experimental verification, and to evaluate a panel of autoantibodies against tumor-associated antigens (TAAbs) for the detection of oral cancer (OC).

Methods: GEO and TCGA databases were used to screen significantly up-regulated genes related to OC, and protein-protein interaction (PPI) analysis and Cystoscope software were used to identify key genes. Enzyme-linked immunosorbent assay (ELISA) was used to detect the expression levels of autoantibodies in 173 OC patients and 173 normal controls, and binary logistic regression analysis was used to build a diagnostic model.

Results: Using bioinformatics, we identified 10 key genes (AURKA, AURKB, CXCL8, CXCL10, COL1A1, FN1, FOXM1, MMP9, SPP1 and UBE2C) that were highly expressed in OC. Three autoantibodies (anti-AURKA, anti-CXCL10, anti-FOXM1) were proven to have diagnostic value for OC in the verification set and the validation set. The combined assessment of these three autoantibodies improved the diagnostic value for OC, with an area under the curve (AUC), sensitivity and specificity of 0.741(95%CI:0.690-0.793),58.4% and 80.4%, respectively. In addition, the combination of these three autoantibodies also had high diagnostic value for oral squamous cell carcinoma (OSCC), with an AUC, sensitivity and specificity of 0.731(95%CI:0.674,0.786), 53.8% and 82.1%, respectively.

Conclusions: Our study revealed that AURKA, CXCL10 and FOXM1 may be potential biomarkers and the panel of three autoantibodies (anti-AURKA, anti-CXCL10 and anti-FOXM1) had good diagnostic value for OC.

Keywords: Autoantibody; Bioinformatics; Early detection of Cancer; Logistic model; Oral cancer.