Considerations before the application of 5-hydroxymethylation levels of long non-coding RNAs for non-invasive cancer diagnosis.

Abstract

Previous studies have suggested that aberrant 5-hydroxymethylcytosines (5hmC) modifications are related to cancer pathobiology. Genome-wide profiling 5hmC in circulating cell-free DNA (cfDNA) using the highly sensitive chemical labeling-based 5hmC-Seal technique has been demonstrated to have the potential to be a robust epigenomic tool for cancer biomarker discovery. Prior studies have mostly focused on cfDNA-derived 5hmC-Seal data summarized in well-annotated genic features (e.g., gene bodies) or unbiased bins. Zhou et al. recently proposed long non-coding RNAs (lncRNAs) as an alternative molecular target for biomarker discovery using publicly available 5hmC-Seal data. Considering its potential clinical impact, we would like to comment on Zhou et al. and advocate more serious consideration of critical issues such as the availability of clinical information and technical variables, especially when performing secondary analysis using publicly available data, with the aim of improving data transparency and translatability.