NEW YORK (GenomeWeb) – Startup Bluestar Genomics, a spinout of the Stephen Quake lab at Stanford University, has uncovered a link between an epigenomic signal in cell-free DNA (cfDNA) and pancreatic cancer in patients.
Based on its findings, Bluestar is developing a non-invasive liquid biopsy assay that tracks epigenetic modifications linked to gene regulation and pancreatic cancer pathogenesis in circulating cfDNA in patient blood samples.
With a five-year survival rate of less than 10 percent, pancreatic cancer is one of the biggest causes of cancer death in the US. It is challenging to diagnose, but early detection can help save patient lives and improve long-term treatments.
Bluestar was cofounded in early 2017 by Quake, a bioengineering professor at Stanford and scientific advisor to the company; CEO Patrick Arensdorf; and advisor Mattias Westman, all of whom previously worked together at ImmuMetrix, which was sold to CareDx in 2014.
In a study published in Cell Research in August 2017, Quake and his team explored the diagnostic potential of monitoring 5-hydroxymethylcytosine (5hmC) levels in cfDNA using a sensitive chemical labeling-based low-input sequencing method.
Retrospectively sequencing blood samples from a variety of different cancers, the researchers discovered that cfDNA displayed distinct features depending on the cancer type. They found that lung cancer leads to a loss of 5hmC in cfDNA, while pancreatic cancer led to upregulated and downregulated 5hmC genes compared to healthy patients.
Based on the results of that study, Bluestar focused on developing a tool to detect early-stage pancreatic cancer in patients based on these epigenomic signatures.
In the newest study, which was posted on the BioRxiv preprint server last month, Quake and Bluestar researchers further investigated the use of 5hmC in circulating cfDNA to specifically detect pancreatic ductal adenocarcinoma (PDAC). The team first collected and isolated plasma samples from a cohort of 51 pancreatic cancer patients and 41 non-cancer controls, then enriched for and sequenced regions of interest.
To do this, they use a method called "click chemistry" to modify hydroxymethyl groups on cytosine by attaching biotin tags. The researchers then enrich the biotinylated DNA fragments by using streptavidin-coated magnetic beads, which enables an effective "pull-down" assay to separate DNA molecules that contain 5hmC from those that don't contain the biomarker.
The team then sequences the fragments using Illumina's NextSeq550 instrument, generating the data that can be used to derive an epigenetic signature.
"We essentially … look at small amounts of cell-free DNA from plasma in patients, and then compare the results it to patients with no cancer," said Bluestar Chief Scientific Office Samuel Levy, a senior author on the study.
Using this method, they identified regions or peaks of dense "read accumulation that mark[ed] the location of a [5hmC] residue in a CpG content."
After performing a set of regression models on the sequenced data, Quake and his colleagues found that PDAC patients possessed thousands of genes with different epigenomic signatures — including areas of enrichment and absence of 5hmC — compared to non-diseased individuals. By filtering the genes with the most differentially hydroxymethylated states, the team found genes that were previously linked to pancreas development or pancreatic cancer.
Levy noted that when the researchers partitioned the data to compare earlier stages versus later stages, they saw that the assay detected the signals "very well" in all stages. The team validated the method on external cohorts from previous studies that contained pancreatic cancer and healthy samples, producing an area under the curve of "74 to 97 percent." The researchers believe that sub-partitioning PDAC and non-cancer individuals into different categories will improve detection and classification of the disease.
"[Further] studies would have to be done that look at high-risk individuals for cancer, and essentially try to [identify biology] that occur that put patients at risk," Levy said. "For the study to be notable, we will work with a network that collects [prospective] samples over time."
Because the study is currently on BioRxiv server, the Bluestar researchers welcome additional manuscript feedback.
According to Arensdorf, Bluestar has currently licensed two patents from Quake's lab awarded in October 2017 "involving several 5hmC" liquid biopsy applications, "and specific technology elements that make it broadly applicable" to additional cancers. He highlighted that Bluestar has since improved the technology and is working with collaborators to apply epigenetic signal detection to a broad range of cancer applications.
Bluestar currently has two offices: one in San Diego that serves as the location for a wet lab and sequencing, and one in San Francisco that serves as the location for biostatistics, bioinformatics, clinical science, and marketing. While Arensdorf declined to comment on the commercial timeline for clinical implementation of the 5hmC assay, he noted that Bluestar aims to eventually establish a CLIA lab to further develop the tool.
"We believe there is a great need for biomarkers in the oncology space, especially in early-stage cancers, and there is plenty of room to grow there," Arensdorf said.
Several groups offer liquid biopsy tests to detect pancreatic cancer in patients. According to its website, Invitae currently offers a hereditary NGS pancreatic cancer panel that covers 20 genes related to the disease. Firms such as Genentech and Exosome Diagnostics (part of Bio-Techne) are also currently developing assays that analyze specific genes potentially linked to pancreatic cancer.
Arensdorf said that Bluestar's 5hmC method differs from other assays because it examines an additional source of information regarding the "underlying biological function [of], not simply the mutational spectrum" found in cell-free DNA in liquid biopsies.
Because Bluestar emerged from stealth mode about two months ago, Arensdorf declined to comment on the firm's current funding besides that it is "comfortably funded for the programs we plan to do."
In addition to reproducing the study's results in larger cohorts, Bluestar will see if the 5hmC method could be applied to other types of cancer — such as lung hepatocellular carcinoma — as well as for monitoring different tumor stages in certain cancers.