Performance evaluation of stool DNA methylation tests in colorectal cancer screening: a systematic review and meta-analysis.

Abstract

There is not sufficient evidence about whether stool DNA methylation tests allow prioritizing patients to colonoscopy. Due to the COVID-19 pandemic, there will be a wait-list for rescheduling colonoscopies once the mitigation is lifted. The aim of this meta-analysis was to evaluate the accuracy of stool DNA methylation tests in detecting colorectal cancer. The PubMed, Cochrane Library and MEDLINE via Ovid were searched. Studies reporting the accuracy (Sackett phase 2 or 3) of stool DNA methylation tests to detect sporadic colorectal cancer were included. The DerSimonian-Laird method with random-effects model was utilized for meta-analysis. Forty-six studies totaling 16149 patients were included in the meta-analysis. The pooled sensitivity and specificity of all single genes and combinations was 62.7% (57.7%, 67.4%) and 91% (89.5%, 92.2%), respectively. Combinations of genes provided higher sensitivity compared to single genes (80.8% [75.1%, 85.4%] vs. 57.8% [52.3%, 63.1%]) with no significant decrease in specificity (87.8% [84.1%, 90.7%] vs. 92.1% [90.4%, 93.5%]). The most accurate single gene was found to be SDC2 with a sensitivity of 83.1% (72.6%, 90.2%) and a specificity of 91.2% (88.6%, 93.2%). Stool DNA methylation tests have high specificity (92%) with relatively lower sensitivity (81%). Combining genes increases sensitivity compared to single gene tests. The single most accurate gene is SDC2, which should be considered for further research.