No abstract

No abstract

No abstract

Currently, no clinically relevant non-invasive biomarkers are available for screening of multiple cancer types. In this study, we developed a serum diagnostic signature based on 5-methylcytosine (m5C)-related miRNAs (m5C-miRNAs) for multiple-cancer detection. Serum miRNA expression data and the corresponding clinical information of patients were collected from the Gene Expression Omnibus database. Serum samples were then randomly assigned to the training or validation cohort at a 1:1 ratio. Using the identified m5C-miRNAs, an m5C-miRNA signature for cancer detection was established using a support vector machine algorithm. The constructed m5C-miRNA signature displayed excellent accuracy, and its areas under the curve were 0.977, 0.934, and 0.965 in the training cohort, validation cohort, and combined training and validation cohort, respectively. Moreover, the diagnostic capability of the m5C-miRNA signature was unaffected by patient age or sex or the presence of noncancerous disease. The m5C-miRNA signature also displayed satisfactory performance for distinguishing tumor types. Importantly, in the detection of early-stage cancers, the diagnostic performance of the m5C-miRNA signature was obviously superior to that of conventional tumor biomarkers. In summary, this work revealed the value of serum m5C-miRNAs in cancer detection and provided a new strategy for developing non-invasive and cost effective tools for large-scale cancer screening.

No abstract