Detecting pancreatic cancer using epigenetic biomarkers

pancreatic cancer diagnosis

Date: 21st October 2020

Cancer diagnosis at an earlier stage is crucial and is associated with improved outcomes and increased survival chances.  Pancreatic cancer has one of the lowest fiver-year survival estimates, at ~7%, as it is often presented late due to the disease frequently having few symptoms.   Now, scientists present non-invasive detection of pancreatic ductal adenocarcinoma (PDAC) by 5-hydroxymethylcytosine changes in circulating cell free DNA obtained from a simple blood draw.

Pancreatic ductal adenocarcinoma (PDAC) constitutes more than 90% of all pancreatic cancer cases, and detection of the disease is currently limited to two biomarkers assayed from an invasive procedure – extracting cystic fluid. Now, scientists from Bluestar Genomics, a start-up company developing innovative, data-driven, epigenomic approaches to comprehensive disease analysis and diagnostics, have developed a platform to detect cancer at early stages, and present their work in Nature Communications journal, addressing the unmet need to provide a screening method for PDAC in larger populations.

Previous work by Stephen Quake (Bluestar Genomics was founded out of the Quake lab), had shown that the molecular signatures in circulating cell free DNA (cfDNA) based on cytosine 5-hydroxymethylation (5hmC) had the potential to define the tissue origin and stage of a tumour, in a variety of disease types.  5hmC is an important mammalian DNA epigenetic modification that has been linked to gene regulation and cancer pathogenesis.

Now the team wanted to test the platform on a case–control study aimed at investigating whether DNA 5hmC signatures were present in the blood of PDAC patients compared to a cohort of non-cancer individuals, and whether these signatures could discriminate between cancer and non-cancer patients.

Plasma specimens from 307 subjects were collected from multiple US locations, 64 with PDAC and the rest from non-cancer patients.  To start, the team first established the 5hmC enriched loci in the non-cancer group, this was then compared to the 5hmC peaks in PDAC patients, revealing significant differences.  The analysis suggested that the differential 5hmC enrichment observed over promoters were driven by transcription activation, and that the platform could be successfully used to identify elements of epigenetic dysregulation in cancer cells.

The team then went on to identify disease specific genes from the plasma samples, they found under stringent conditions that 577 genes were upregulated and 217 were downregulated. The genes with increased 5hmC density in PDAC were those related to pancreas development and cancer, whilst the downregulated set in the PDAC cfDNA cohort were enriched for gene sets that were downregulated in KRAS mutant cells (KRAS mutations are particularly associated with pancreatic cancer) as well as immune response and whole blood genes. These results suggested that 5hmC profiling could capture PDAC tumour relevant biological signals in plasma.

This PDAC signature was then used to determine whether this set could partition PDAC from non-cancer samples on previously published datasets.  Indeed, this was the case, the 794 genes separated the non-cancer samples from PDAC samples with high accuracy.

Finally, the team went on to build a predictive model using this comprehensive gene set, the trained model was tested on two independent validation sets of patient samples. The diagnostic ability of model to discriminate the true state of subjects was highly accurate. Furthermore, when comparing the 5hmC predictive model against another clinically relevant pancreatic cancer biomarker, CA-19, 5hmC signals significantly improved detection for both sensitivity and specificity, particularly for early stage PDAC.

Conclusions and future applications

The team here show that hydroxymethylation-based biomarkers in cfDNA may facilitate the development of molecular diagnostic tests to detect pancreatic cancer, not only late but also at early stages. Early stage detection with this method is enhanced over other techniques due to the ability to detect a large range of signals in the plasma, not only from the cancer cells themselves but also from the immune cells, which all contribute to cfDNA.

From a wider perspective, 5hmC also has the potential to detect other types of cancer from plasma samples, such as lung, hepatocellulcar carcinoma, colon and gastric cancer.  The development of a simple, diagnostic test for cancer would be a great asset in the clinic.  Certainly, Bluestar will continue its development work required to commercialise a 5hmC-based test in the next few years.

The idea of diagnosing cancer from biomarkers in the blood is not a new one, and others are also developing assays to detect disease in this way.  Recent we reported the development of an ultra-sensitive biosensor which detects nucleic acids using crumpled graphene and is able to detect cancer markers in patient blood or serum.  The hope there is to develop the test as a hand held device with different cartridges able to detect different target molecules associate with disease.

However, the real strength of 5hmC lies within its ability to escape the limitations of sequence alone. The technique allows the interpretation of DNA hydroxymethylation patterns and other epigenomic characteristic that can provide new disease and clinical information, such as disease prognosis forecasting or therapeutic response monitoring, importantly it has the potential to drive new solutions for our healthcare needs.


For more information please see Bluestar Genomics on our synthetic biology map

Guler, G. D., Y. Ning, C.-J. Ku, T. Phillips, E. McCarthy, C. K. Ellison, A. Bergamaschi, F. Collin, P. Lloyd, A. Scott, M. Antoine, W. Wang, K. Chau, A. Ashworth, S. R. Quake and S. Levy (2020). “Detection of early stage pancreatic cancer using 5-hydroxymethylcytosine signatures in circulating cell free DNA.” Nature Communications 11(1): 5270.