Establishing Clinically Significant Prostate Cancer Diagnosis Models Based on Serum Methylation Levels of Four Genes.

Background: Prostate cancer is the most common malignant tumor of the male reproductive system. Improving the ability of noninvasive diagnosis of prostate cancer is an urgent clinical problem to be solved and has broad research prospects. Our explores the application value of serum DNA methylation indicators and models in the diagnosis of clinically significant prostate cancer (csPCa), including its diagnostic efficiency and application in unnecessary biopsy.

Methods: This study retrospectively enrolled 223 Chinese patients and recorded patients' clinical information. Five serum gene promoter methylation levels were detected before operation and then established various logistic regression models based on serum methylation levels and clinical information. The diagnostic efficacy of each model for csPCa was analyzed and compared.

Results: In our research, the area under the receiver operating curve of serum methylation regression model (abbreviated as GRP) was significantly higher than that of traditional models based on prostate specific antigen model (p < 0.05). When the sensitivity of the model was 95.0% or greater, and the rate of unnecessary biopsy was increased by 33.1%, the rate of missed diagnosis was only 5.8% (3/56). The net clinical benefit of GRP model was also much higher than that of the other two models in the decision curve analysis.

Conclusions: The diagnostic capacity of the serum methylation regression model is superior to the traditional model. Compared with traditional model, it can reduce the number of unnecessary prostate biopsy by 16.5% to 21.6%, and the rate of missed diagnosis is not significantly improved.