For Methylation Sequencing, The Cancer Is In The Details

Typically, CRC is detected through colonoscopy, a powerful method that is nonetheless flawed and often fails to detect early-stage disease [2, 3, 4]. In recent years, assays that screen a patient’s blood for fragments of circulating tumor DNA (ctDNA) have emerged as potentially more sensitive, patient-friendly alternatives [5].   

Fragments of wayward DNA can be shed into circulation from individual tumor cells, bringing with them tell-tale genetic markers that point to their malignant origins. In theory, this should mean that tumors can be found in their most nascent stages simply by screening for ctDNA. However, successfully recognizing ctDNA can be difficult: w ith very few cells to shed DNA, the ctDNA produced by a stage I tumor represents a small fraction of the total cell free DNA (cfDNA) that’s already in circulation. And, an even smaller fraction of ctDNA fragments will possess the critical, tumor-identifying mutations. 

One way to improve sensitivity is to instead search DNA fragments for defining epigenetic markers. Global changes in DNA methylation patterns have become a hallmark of cancer and may take form early in the disease’s life course [6, 7]. This effectively paints the tumor genome with distinctive epigenetic markers that will be carried on subsequent ctDNA fragments. Rather than hoping to find the rare bits of ctDNA carrying tumor mutations, researchers can search for the far more numerous fragments that have tell-tale epigenetic signatures. 

DNA methylation typically describes the formation of 5-methylcytosine (5mC), a reversible epigenetic tag that is often associated with gene repression. 5mC may be converted to a stable intermediate, 5-hydroxymethylcytosine (5hmC), during the demethylation process. While related, 5mC and 5hmC appear to have opposing effects on gene regulation, with 5hmC coalescing around genes that are becoming active. Despite their differences, traditional methylation sequencing technology has failed to differentiate between them, conflating the two into a single modified cytosine (modC) readout [7].  

We believe that this conflation can hide important patterns in the epigenetic landscape, patterns that may be critical to recognizing ctDNA released from early-stage tumors. During transformation, tumor cells must re-wire gene expression, activating specific genes while silencing others. Both the methylation and demethylation of cytosine will be critical to this process. Evidence suggests that levels of 5hmC are likely to begin changing as genes re-awaken before 5mC and overall modC changes are registered [8].  

In our recent study, 5-methylcytosine and 5-hydroxymethylcytosine are synergistic biomarkers for early detection of colorectal cancer, we set out to test whether finer resolution methylation sequencing—using 6-base sequencing technology—could improve early stage CRC detection.  

6-Base Sequencing Improved CRC Detection 

6-base sequencing technology is a recent advancement that allows researchers to resolve the four canonical bases (A, T, G, C) along with both 5mC and 5hmC individually. Such fine detail has the potential to reveal changes overlooked by traditional, modC-based readouts [7]. For example, where a decrease in 5mC is matched by an increase in 5hmC, traditional readouts may report no change in methylation status, thus obscuring nuanced changes in the epigenetic landscape [8]. We hypothesized that these nuanced changes could be valuable for recognizing ctDNA in early stage CRC. 

Using 6-base sequencing we screened liquid biopsy samples for ctDNA [9]. Samples in this study were selected from a cohort deliberately weighted toward early disease: 32 healthy controls, 26 stage I CRC patients, and 11 stage IV CRC patients. 

Though methylation changes are global in scale, not every location in the genome will be affected. To determine which loci to focus on, we narrowed our analysis to regions that are known to change in CRC. Through means described in the paper, we identified 11,686 modC regions spanning a wide range of genomic features that were covered in our 6-base sequencing data. 

We then built generalized linear models (with leave-one-out cross-validation and Lasso regularization) to distinguish stage I CRC from healthy controls. In these models, we checked the performance of ctDNA detection using patterns of: 

• 5mC only 

• 5hmC only 

• modC 

• 5mC + 5hmC together 

The difference was stark. AUC values (representing successful ctDNA detection rates) showed that analyzing 5mC and 5hmC as separate signals in a larger pattern (5mC + 5hmC) far outperformed other models. AUCs were: 

• 5mC = 0.69 

• 5hmC = 0.55  

• modC = 0.66  

• 5mC + 5hmC = 0.95 

 By recognizing 5mC and 5hmC as distinct markers that contribute to a larger methylation landscape, our model achieved 97% specificity and 81% sensitivity in the stage I versus healthy comparison—far exceeding current guidance for blood-based biomarker testing in CRC screening which calls for minimum thresholds of 90% specificity and 74% sensitivity. 

Conflated Signals Obscure Biology  

One potential reason for why 5hmC + 5mC models outperformed traditional methylation sequencing is that demethylation processes begin in the disease’s early stages but may not complete until later stages. Our results support this possibility: Among regions that showed increased 5hmC in stage I plasma (relative to controls), 68.7% also showed decreased 5mC in stage IV plasma—a striking enrichment that’s suggestive of active demethylation dynamics.  

This suggests that early disease is accompanied by dynamic regulatory changes that include the oft overlooked formation of 5hmC. As the cancer progresses, demethylation becomes more widespread and is marked by a more dramatic loss of both 5hmC and 5mC in those same regions. When these signals are conflated into a single modC readout, such early-stage changes may be obscured. Put simply, by flattening methylation dynamics to a single modC readout, researchers risk overlooking markers that signal early-stage disease. 

Towards Sensitive Early Detection 

Our results are exciting on their own, but all the more so with the recent presentation of corroborating data from AstraZeneca at the Festival of Genomics and Biodata meeting in London, UK [10]. As in our study, the value of disambiguating modC was evident: in using 6-base sequencing, the team observed a more nuanced epigenetic landscape that changed over the course of CRC progression. That nuance, the team found, could help improve early stage detection. 

Together with our results, these two studies suggest that the epigenetic patterns carried on cfDNA can be a valuable source of clinical information. In failing to distinguish between 5mC and 5hmC, traditional methylation sequencing technology overlooks the subtle epigenetic patterns that arise in stage I CRC. As a result, the ability to recognize rare bits of ctDNA shed from nascent CRC tumors will likely be diminished using these technologies. With the advent of 6-base sequencing, however, researchers gain the ability to observe nuanced changes in the epigenetic landscape and potentially develop better, more sensitive tools for CRC screening.  

References 

1. “SEER*Explorer Application.” Cancer.gov, 2022, seer.cancer.gov/statistics-network/explorer/application.html. Accessed 3 Mar. 2026. 
2. Robertson, Douglas J, et al. “Colorectal Cancers Soon after Colonoscopy: A Pooled Multicohort Analysis.” Gut, vol. 63, no. 6, 21 June 2013, pp. 949–956, https://doi.org/10.1136/gutjnl-2012-303796. 
3. Li, Ying, et al. “Missed Colorectal Cancer Diagnosis by Screening Colonoscopy Based on the PLCO Cancer Screening Trial.” International Journal of Colorectal Disease, vol. 40, no. 1, 3 Oct. 2025, p. 206, https://doi.org/10.1007/s00384-025-04952-4. 
4. National Cancer Institute. “Colorectal Cancer Screening (PDQ®)–Health Professional Version - National Cancer Institute.” Www.cancer.gov, 4 June 2021, www.cancer.gov/types/colorectal/hp/colorectal-screening-pdq. Accessed 3 Mar. 2026.
5. Hany Emile, Sameh, et al. “An Umbrella Review of Systematic Reviews on the Diagnostic and Prognostic Utility of Circulating Tumor DNA and MicroRNA in Colorectal Cancer.” Cancer Treatment Reviews, vol. 141, Dec. 2025, https://doi.org/10.1016/j.ctrv.2025.103040. 
6. Nishiyama, Atsuya, and Makoto Nakanishi. “Navigating the DNA Methylation Landscape of Cancer.” Trends in Genetics: TIG, vol. 37, no. 11, 1 Nov. 2021, pp. 1012–1027, https://doi.org/10.1016/j.tig.2021.05.002. 
7. Crawford, Robert, et al. “5mC and 5hmC Methylation Sequencing: The Power of 6-Base Sequencing in a Multiomic Era.” Epigenomics, vol. 18, no. 1, 12 Nov. 2025, pp. 101–115, https://doi.org/10.1080/17501911.2025.2586452. 
8. Guerin, Lindsey N., et al. “Temporally Discordant Chromatin Accessibility and DNA Demethylation Define Short- and Long-Term Enhancer Regulation during Cell Fate Specification.” Cell Reports, vol. 44, no. 5, May 2025, p. 115680, https://doi.org/10.1016/j.celrep.2025.115680.
9. Puddu, Fabio, et al. “5-Methylcytosine and 5-Hydroxymethylcytosine Are Synergistic Biomarkers for Early Detection of Colorectal Cancer.” Communications Medicine, vol. 6, no. 15, 20 Jan. 2026, https://doi.org/10.1038/s43856-025-01278-8. 
10. biomodal. Biomodal to Showcase Clinical Research for Earlier Detection of Colorectal Cancer at the Festival of Genomics and Biodata London 2026 | Biomodal. 27 Jan. 2026, biomodal.com/news/biomodal-to-showcase-clinical-research-for-earlier-detection-of-colorectal-cancer-at-the-festival-of-genomics-and-biodata-london-2026/. Accessed 3 Mar. 2026.