Why use circulating tumor methylated DNA (ctmDNA) to detect disease?

Circulating tumor DNA (ctDNA) isn’t abundant enough for subtle signals from circulating DNA

History and context of cell-free DNA and its role in cancer

In 1949 Mandel and Matais discovered cell-free DNA in plasma in a French publication and in 1965 (seventeen years later) it was hypothesized that cell-free DNA in the bloodstream is implicated in metastatic cancer.  The analysis of cell-free DNA has been popularized by what is termed Non-Invasive Prenatal Testing (NIPT).

As an aside, this more correctly should be termed minimally invasive prenatal screening or MIPS, as you need to draw blood, and it is a screen for a condition rather than a test; a test is used for making a treatment decision or to confirm a screen, but we digress.

With four major firms providing this non-invasive screening, it is expected that by 2019 the US test volume will increase to 2 million tests per year. There are approximately 4 million live births in the US annually; by now you probably know someone first-hand who has had one of these screening tests as part of their prenatal examinations.

In the summer of 2015 a report was published in JAMA with an analysis of 125,426 expectant mothers who had the screening test, and 39 of those mothers had multiple chromosomal abnormalities reported by the test. As the MIPS assay, depending on the provider, looks at whole-genome or targeted chromosome relative copy number of fetal DNA to maternal DNA, if the expectant mother has cancer shedding abnormal chromosomes into the bloodstream the MIPS will detect the aberrant signal.

Of those 39 mothers with abnormal screening test results, 10 were diagnosed with cancer. Naturally there are severe limitations to the genetic detection approach, as the MIPS assay was designed to detect aneuploidy in chromosomes 13, 18 and 21 (trisomy in chromosome 13 is called Patau Syndrome, in 18 called Edwards Syndrome, and 21 is Down Syndrome). Nonetheless the promise of detecting cancer early by cell-free DNA was demonstrated.

The promise of ‘liquid biopsy’ and its benefits to guide therapy, monitor relapse, and screen for disease

For patients undergoing therapy, the benefits of a blood draw as an alternative to invasive biopsy is clear, and for ongoing monitoring of disease can complement current imaging diagnostic tools. Liquid biopsy is a broad term that encompasses many types of analytes. In the case for cancer liquid biopsy typically means analysis of cell-free nucleic acids (both DNA and RNA), circulating tumor cells (while few in number, CTCs are showing increasing utility as an alternative to circulating tumor DNA analysis), and exosomes and their contents.

The first circulating tumor DNA liquid biopsy approved by the FDA is the Roche cobas EGFR Mutation Test v2 as a companion diagnostic for the NSCLC therapy Tarceva®. Based upon the cobas 4800 real-time PCR system, it is expected that the convenience of a blood draw liquid biopsy over invasive biopsy in NSCLC will help drive EGFR testing for targeted therapy.

In the summer of 2016 a group at Johns Hopkins University working with collaborators in Melbourne Australia published a study in Science Translational Medicine following up a group of 230 Stage II colorectal cancer patients for a full four years (from 2011 to 2015) with regular samples of blood drawn. The Hopkins group not only used a sensitive ctDNA assay but also complemented their approach using protein-based biomarkers. Of the 230 patients, their assay detected ctDNA in 20 patients, of which 6 received additional chemotherapy. 3 of those 6 did experience cancer recurrence over the four-year timeframe.

However another 14 patients experienced cancer recurrence, without any detection of their ctDNA.

For a screening test for early detection of cancer, this is such a ‘holy grail’-type goal that a company called GRAIL has not only embarked on broad and deep sequencing (508 genes at 66,000x-fold sequencing coverage depth), but also enrolling 15,000 in a clinical trial called The Circulating Cell-free Genome Atlas Study (CCGA) (ClinicalTrials.gov Identifier: NCT02889978)  and another one called The STRIVE Study: Breast Cancer Screening Cohort enrolling 120,000 (ClinicalTrials.gov Identifier: NCT03085888). In order to fund these large and expensive prospective studies, GRAIL has raised $1.5B to-date in three rounds of funding.

As these studies take time for enrollment and then monitoring the health of patients over the course of their cancer care, and then doing the analysis and submitting for regulatory approval, it could well be six or ten years or longer before an approved screening test is available for a disease like lung cancer, where currently there is no screening test available and the penalty for a false-positive result from such a screening test is high. While screening for lung cancer, the confirmatory diagnostic test would be a needle biopsy, which carries its own risks not to mention costs and effort.

Why circulating tumor methylated DNA is an attractive analyte

GRAIL for their screen is looking at 508 cancer-related genes frequently mutated across a number of cancer types. Other companies offering tests for therapy choice such as Guardant Health and Foundation Medicine offer tests with genes that number in the several dozen. These assays are sensitive, down to one part in one thousand (0.1% allele fraction), and examine multiple mutation types including single nucleotide variants (SNVs), insertion-deletion mutations (indels), copy-number alterations (CNAs), and structural variants (SVs).

While 508 genes sounds like a lot of genes, the task in measuring 0.1% allele fraction from a single 10 mL blood draw is a significant challenge, due to the numbers of molecules available. A typical yield of plasma from 10 mL whole blood is approximately 4 mL, and there is in a typical healthy individual 5-15 ng of cfDNA per mL. Thus in a nominal draw there may be 40 ng of cfDNA available to analyze.

In that 40 ng, if there are 3.3 picograms per genome equivalent (GE) there are roughly 12,100 GE’s in that sample. Therefore if you are looking for 0.1% of the GE’s, that’s only 12 copies available for analysis. The assay to detect only 12 copies of a single piece of DNA (out of 12,000) means you must be very efficient at turning those 12 molecules into a result; alas if your assay is only 25% efficient you may not see anything at all.

Methylated DNA, on the other hand, has literally millions of analytes available for analysis. It is estimated that roughly 1.5% of the human genome is methylated – upwards of 45 million bases. And DNA methylation has been studied in detail for a few decades, and the role of 5’-methyl Cytosine in gene expression has been firmly established, in addition to other types of methylation such as 5’-hydroxymethyl Cytosine.

The addition of a -CH3 group occurs at a specific dinucleotide in the promoter region or "CpG island" or "CpG shore" near a gene, in what is known as a CpG site. (The ‘p’ stands for an intervening phosphate group on the DNA backbone; this nomenclature is to differentiate a CpG on the same strand of DNA from a CG pair of bases on opposite strands.)

The difficulties of measuring methylation

Do you remember those twelve target copies available in a 10 mL blood draw? To study methylation it is the same number of target molecules, except for this time you are going to denature the strands of DNA and change the unmethylated Cytosines with a harsh chemical treatment, called bisulfite. Now your challenge is made even more difficult, as bisulfite can reduce by half the amount of target molecules present.

The 12 copies of affected genomic equivalents now becomes only 6. Therefore diagnostic developers have not had assays on the market that looked at measuring methylation in a massively parallel fashion; they had to resort to methods such as real-time PCR which are highly sensitive but also very limited in scope, in terms of the number of specific methylation markers that could be examined.

Singlera Genomics was able to overcome such difficulties through use of a proprietary biochemistry that can simultaneously look at 20,000 methylation haplotype markers, and use this welter of information from such a miniscule amount of target material in order to detect cancer up to four years prior to conventional diagnosis in asymptomatic individuals.

What is a haplotype, and why is it important?

In classical genetics, a haplotype is where DNA polymorphisms (called Single Nucleotide Polymorphisms or SNPs) can be assigned contiguously along the same strand of DNA. This information becomes important in population genetics when you can assign SNPs that are in Linkage Disequilibrium, in that these are SNPs that travel together in blocks because they are physically associated next to each other on the same strand of DNA.

Singlera’s analysis looks at the individual CpG methylation status, yet these CpG sites in upstream regions of a gene tend to cluster closely together. These CpG sites can occur in gene promoters, a group of CpG’s termed “CpG Islands”, and CpGs up to 2kb away arranged in groups called "CpG Island Shores". When Singlera’s assay looks at 20,000 CpG sites, we retain the stranded methylation status information. That is, we know which methylated base (and at what percent it is methylated as these sites are not all unoccupied by -CH3 or occupied).

Haplotypes as it relates to methylation increases the power of detecting the unique signature of methylated DNA coming from cancer cells; these form a unique identifier for cancer that increases discrimination between DNA from cancer and healthy cells. And with 20,000 unique methylation haplotypes to choose from to form your classifier, a sensitive and specific method has been developed in order to do what was once thought impossible: the capability for early detection of cancer in asymptomatic individuals.

Results that speak for themselves

Last week Singlera published a press release that read “Previous studies have reported that biomarker signals from early cancers are difficult to detect. By utilizing Singlera's proprietary PanSeer assay for targeted DNA methylation haplotype next-generation sequencing, the research team was able to demonstrate early detection of colorectal, esophageal, liver, lung, and stomach cancer four years prior to disease diagnosis.  These five cancer types represent 43% of cancer mortality in the US and 74% of cancer mortality in China.”

Understand that in this study, the 120,000 healthy volunteers were asymptomatic, and we were able to look at those individuals with cancer and analyze their blood for a methylation haplotype signal from cancer four years before their first diagnosis.

Does analyzing methylation haplotypes from cell-free DNA sound intriguing to you? Would you like to access this remarkable technology? Please contact us with an inquiry. If you'd like to download the posters, you can access them here with registration.