Abstract
The early detection of cancer can reduce cancer-related mortality. There is no clinically useful noninvasive biomarker for early detection of breast cancer. The aim of this study was to develop accurate and precise early detection biomarkers and a dynamic monitoring system following treatment. We analyzed a genome-wide methylation array in Taiwanese and The Cancer Genome Atlas (TCGA) breast cancer (BC) patients. Most breast cancer-specific circulating methylated CCDC181, GCM2 and ITPRIPL1 biomarkers were found in the plasma. An automatic analysis process of methylated ccfDNA was established. A combined analysis of CCDC181, GCM2 and ITPRIPL1 (CGIm) was performed in R using Recursive Partitioning and Regression Trees to establish a new prediction model. Combined analysis of CCDC181, GCM2 and ITPRIPL1 (CGIm) was found to have a sensitivity level of 97% and an area under the curve (AUC) of 0.955 in the training set, and a sensitivity level of 100% and an AUC of 0.961 in the test set. The circulating methylated CCDC181, GCM2 and ITPRIPL1 was also significantly decreased after surgery (all p < 0.001). The aberrant methylation patterns of the CCDC181, GCM2 and ITPRIPL1 genes means that they are potential biomarkers for the detection of early BC and can be combined with breast imaging data to achieve higher accuracy, sensitivity and specificity, facilitating breast cancer detection. They may also be applied to monitor the surgical treatment response.