Skip to main content
Premium Trial:

Request an Annual Quote

UCSD Methylation Haplotype Method Tracks cfDNA Origin; Singlera to Commercialize


NEW YORK (GenomeWeb) – Investigators at the University of California, San Diego have developed a new method for determining the origin of circulating DNA fragments, which they hope to develop as a strategy for more sensitive detection of cancer, and potentially also for blood-based diagnosis of other diseases.

The team published a study describing the approach in Nature Genetics this Monday. The method relies on identifying methylation haplotypes — instances of co-methylation across a number of different CPG sites — that are specific to a particular tissue or cell type.

By looking for these patterns, the investigators showed that they could accurately predict cancer status and also identify a tumor's location from circulating DNA in patients with lung cancer or colorectal cancer.

Senior author Kun Zhang, a bioengineering professor at UCSD's Jacobs School of Engineering, said in an interview that he and his colleagues plan to further develop the method as a tool for early cancer detection through commercial genetics startup Singlera Genomics, which Zhang cofounded with Johns Hopkins' Yuan Gao and others in 2014.

Although a number of liquid biopsy tests are now clinically available, current methods speak only to the presence or absence of a tumor or its particular molecular changes. They don't indicate where it is located in the body.

But over the last year or so, several new advances have been made in inferring or tracking the tissue of origin of circulating DNA fragments.

For example, a team from Jay Shendure's lab at the University of Washington launched a new company last month called Bellwether Bio, which is working to commercialize a method for inferring the origin of circulating cell-free DNA based on patterns of nucleosome spacing.

The UW researchers published a proof of concept in Cell last year, in which they found different fragmentation/nucleosome fingerprints in the cell-free DNA of five cancer patients, which they could link to the patients' particular type of cancer.

Other groups have also looked at methylation. A team from Hebrew University-Hadassah Medical School published its own results in Proceedings of the National Academy of Sciences last year, and another group led by Dennis Lo and colleagues at the Chinese University of Hong Kong shared its approach in PNAS in 2015.

Still others are hoping that they can detect other molecules like RNA in circulation that can be linked to specific tissues in the body as a way of diagnosing and monitoring disease or dysfunction.

According to Zhang, all of these developments represent a huge step forward, offering the possibility for more sensitive and more precise early cancer detection than current liquid biopsy methods can manage, as well as new opportunities for cfDNA-based tests in other, non-cancer diseases.

"Current [liquid biopsy] assays are mainly applicable to people who have already been diagnosed with cancer, so you already know where the cancer comes from … But to go to the holy grail of early detection you need to know tissue origin," Zhang said.

"With just a probability of cancer what do you do with that?" he added. "But with an origin you can do imaging … maybe a needle biopsy. So in early detection tissue origin is equally important as whether you have the evidence of cancer."

Zhang said he and his colleagues believe that their newly published haplotype method overcomes some of the limitations of previously published approaches to determining tissue of origin in circulating DNA that use individual sites of epigenetic alteration.

"When you are doing methylation analyses there are all these technical errors — sequencing errors — so detecting low-levels of signal is difficult," he said.

The concept of methylation haplotypes is akin to the theory of linkage disequilibrium, which describes the co-segregation of adjacent genetic variants on human chromosomes, Zhang and his coauthors wrote.

Most research so far — such as the study by researchers at Hebrew University last year — has used relatively limited data from methylation microarray studies. Zhang and his team decided to try to collect a much more comprehensive set of haplotype patterns from public repositories of bisulfite sequencing data as well as new sequencing datasets they created themselves.

In their study this week, Zhang and his colleagues began by defining 147,888 blocks of methylation haplotypes using an analysis of 61 whole-genome bisulfite sequencing data sets, and validation with 101 reduced-representation bisulfite sequencing data sets and 637 methylation array data sets.

Using a metric called methylation haplotype load, they performed tissue-specific methylation analysis at the block level, picking out subsets of informative blocks that should be able to deconvolute heterogeneous DNA samples.

The team then screened blood samples from a set of 75 healthy individuals, 29 patients with lung cancer, and 30 with colorectal cancer looking to see if they could use these methylation haplotype blocks to distinguish cancer patients from normal healthy controls, and to identify different types of cancer based on markers of cellular origin in the body.

Tweaking their cutoff points and prediction methods, the team was able to reach 83 percent accuracy in identifying the correct tissue of origin in the colorectal cancer samples, and 92 percent in the lung cancer samples.

Misclassified samples were mainly due to the inclusion of samples with heterogeneous clinical status, the authors wrote. Four of five colorectal cancer samples samples were from patients with metastatic CRC, whereas the fifth was a tubular adenoma. Similarly, one lung cancer sample came from a patient with cryptococcal pulmonary infection who later developed lung cancer.

Interestingly, the team found that coupling both cancer-specific CPG haplotypes and non-cancer tissue-specific methylation patterns together offers an even more sensitive method for identifying circulating DNA from a cancer and determining its source.

Yuval Dor, who led the team from Hebrew University that also identified groups of co-methylated sites with the ability to track tissue of origin last year, said in an email this week that Zhang and colleagues approach reflects and expands upon his team's findings.

It is also a "significant advance" over what Lo and colleagues demonstrated in their analysis using bisulfite sequencing data the year before, Dor added.

Though it remains to be seen how well methylation haplotype-based analyses (or other methods being investigated) hold up in additional samples and diseases, Dor said, "the clinical potential is enormous, for detection of cancer and much beyond."

Zhang said that based on the results reported by Shendure and colleagues at UW, the nucleosome approach that that group is commercializing could potentially be complementary to methylation haplotyping.

"It's two different aspects of epigenetics — in one you are looking at which part of the chromatin is open and closed and in the other you are looking a stretch of DNA that has methyl groups attached," so there could be independent and added value from each type of analysis.

Zhang's startup Singlera said last summer that it had formed an alliance with Fudan University's Taizhou Health Science Institute to analyze samples from a large cohort of Chinese volunteers to identify biomarkers of early-stage cancer.

He explained this week that that cohort will be used to support further development of the haplotype method described in Nature Genetics.

"That cohort has been going on for about 9 years with about 200,000 normal [individuals], some of whom develop cancer, with [longitudinal] samples both before and after they are diagnosed."

Though he believes that the study results also bode well for using tissue-specific haplotypes to diagnose or track other types of diseases, Singlera's near-term focus is on the early cancer detection.