5-hydroxymethylcytosine as a liquid biopsy biomarker in colorectal cancer.

291 Background: 5-hydroxymethylcytosine (5hmC) is an epigenetic modification which regulates gene expression and is associated with active transcription. The optimization of 5hmC sequencing in cell-free DNA (cfDNA) could therefore enable assessment of gene activity through a liquid biopsy. We aimed to investigate the 5hmC landscape of colorectal cancer (CRC) in plasma-derived cfDNA to evaluate the potential of 5hmC modifications as a liquid biopsy-based biomarker of CRC. Methods: Genome-wide 5hmC modifications were analyzed with a low-input whole-genome 5hmC sequencing method based on selective chemical labeling in cfDNA from 72 CRC patients and 70 healthy samples. Differential 5hmC analysis between CRC and healthy samples was conducted using DESeq2 (padj < 0.05 and log2FC <= -1 and >=1). Low-pass whole genome sequencing (LP-WGS) was performed on all cfDNA samples. Tumor fraction was estimated using ichorCNA to classify samples into high ct-fraction (>=0.1) and low ct-fraction (<0.1) groups. Nucleosome profiling was performed on LP-WGS to reveal chromatin accessibility patterns. Elastic Net Regression (ENR) was applied to identify changes in 5hmC levels for classifying samples as cancerous or healthy. Results: Differential cfDNA 5hmC levels of 139 genes distinguished high ct-fraction CRC (n=16) and healthy samples (n=70). Although the remaining CRC samples (n=56) had similar ct-fractions as the healthy samples, as estimated by ichorCNA, we found a differentially hydroxymethylated gene signature could discriminate between these low ct-fraction CRC and healthy samples. Nucleosome profiling analysis with LP-WGS in cfDNA was used to identify active transcriptional drivers of the genes with increased 5hmC levels. The analysis revealed increased chromatin accessibility near the binding sites for the transcription factor caudal-related homeobox transcription factor 2 ( CDX2 ), a master transcriptional regulator of intestinal cell fate and cancer, in CRC patients relative to healthy samples. Corroborating this, CDX2 and its downstream targets had increased 5hmC levels in cfDNA in CRC relative to healthy samples, consistent with colorectal-derived tumor DNA in CRC plasma samples. Conclusions: Genes with differential 5hmC levels in cfDNA can discriminate CRC and healthy samples. Our 5hmC signature identified CRC samples that did not harbor copy number gains or losses; which is typically used to estimate ct-fraction and routinely included in targeted ctDNA sequencing assays. Integration of 5hmC and LP-WGS can be used to identify gene activation in disease. In the context of CRC, CDX2 displayed increased chromosome accessibility in the promoters of its target genes which in turn displayed increased 5hmC levels. Further studies are aimed at optimizing and validating 5hmC-based biomarkers throughout patient treatment.