Background: Epigenetic variants carried by circulating tumor DNA can be used as biomarkers for early detection of hepatocellular carcinoma (HCC) by noninvasive liquid biopsy. However, traditional methylation analysis method, bisulfite sequencing, with disadvantages of severe DNA damage, is limited in application of low-amount cfDNA analysis.Results: Through mild enzyme-mediated conversion, enzymatic methyl sequencing (EM-seq) is ideal for precise determination of cell-free DNA methylation and provides an opportunity for HCC early detection. EM-seq of methylation control DNA showed that enzymatic conversion of unmethylated C to U was more efficient than bisulfite conversion. Moreover, a relatively large proportion of incomplete converted EM-seq reads contains more than 3 unconverted CH site (CH = CC, CT or CA), which can be removed by filtering to improve accuracy of methylation detection by EM-seq. A cohort of 241 HCC, 76 liver disease, and 279 normal plasma samples were analyzed for methylation value on 1595 CpGs using EM-seq and targeted capture. Model training identified 283 CpGs with significant differences in methylation levels between HCC and non-HCC samples. A HCC screening model based on these markers can efficiently distinguish HCC sample from non-HCC samples, with area under the curve of 0.957 (sensitivity = 90%, specificity = 97%) in the test set, performing well in different stages as well as in serum α-fetoprotein/protein induced by vitamin K absence-II negative samples.Conclusion: Filtering of reads with ≥ 3 CHs derived from incomplete conversion can significantly reduce the noise of EM-seq detection. Based on targeted EM-seq analysis of plasma cell-free DNA, our HCC screening model can efficiently distinguish HCC patients from non-HCC individuals with high sensitivity and specificity.
Background: Whole-genome methylation sequencing of cfDNA is not cost-effective for tumor detection. Here, we introduce reduced representative methylome profiling (RRMP), which employs restriction enzyme for depletion of AT-rich sequence to achieve enrichment and deep sequencing of CG-rich sequences. Methods: We first verified the ability of RRMP to enrich CG-rich sequences using tumor cell genomic DNA and analyzed differential methylation regions between tumor cells and normal whole blood cells. We then analyzed cfDNA from 29 breast cancer patients and 27 non-breast cancer individuals to detect breast cancer by building machine learning models. Results: RRMP captured 81.9% CpG islands and 75.2% gene promoters when sequenced to 10 billion base pairs, with an enrichment efficiency being comparable to RRBS. RRMP allowed us to assess DNA methylation changes between tumor cells and whole blood cells. Applying our approach to cfDNA from 29 breast cancer patients and 27 non-breast cancer individuals, we developed machine learning models that could discriminate between breast cancer and non-breast cancer controls (AUC = 0.85), suggesting possibilities for truly non-invasive cancer detection. Conclusions: We developed a new method to achieve reduced representative methylome profiling of cell-free DNA for tumor detection.
Tumor-specific DNA methylation carried by circulating tumor DNA (ctDNA) is a promising marker for early cancer screening. This study endeavors to create, verify, and deploy a liquid biopsy method based on ctDNA methylation for the early identification of lung cancer (LC). Firstly, we established an LC-specific methylation profile by analyzing the differential expression of DNA methylation between 69 pairs of cancerous and paracancerous healthy tissues. A total of 2,680 differentially methylated positions (DMPs) between lung cancer tissues and paracancerous healthy tissues were identified. Enrichment analysis unveiled associations of these DMPs with various biological processes. The heatmap representation of LC-specific DMPs underscores distinct methylation patterns observed between lung cancer and paracancerous tissues. Subsequently, 127 plasma samples from stage I lung cancer patients and unaffected individuals were split into training and test groups to establish a model incorporating 22 of these LC-specific DMPs. This model yielded an 88.57% sensitivity and 78.43% specificity in the training data, with an area under the curve (AUC) of 0.8913. The corresponding performance in the test set was 60% sensitive and 76.92% specific, with an AUC of 0.7025. These outcomes indicate the promise of ctDNA methylation markers in enhancing the potential for early lung cancer detection.