Researchers at Stanford University have developed a new method of detecting cancer from circulating tumor DNA (ctDNA) that combines DNA fragmentation analysis with deep sequencing to infer RNA expression profiles of tumor cells.

The ability to predict RNA expression from ctDNA is a first for blood-based cancer detection, and has potential for diagnosis, prognosis, and therapy guidance, the researchers said.

Blood tests that can detect tumors promise to usher in a whole new era of cancer screening that is fast, cheap and noninvasive. Many tests available today are designed to identify fragments of DNA that contain hallmark mutations of specific cancer types, like BRCA1/2 genes in patients with ovarian cancer, for example. While the tests show much promise in early detection, limitations in sensitivity remain.

Critics are quick to point out that the tests may not detect circulating DNA in small quantities, which is often the case in early-stage cancers. Furthermore, detecting mutations can be hit or miss since not all patients have the classic genetic signatures characteristics of certain cancers. Additionally, mutations can be shared among various cancers which makes it difficult to determine the exact tissue origin of the tumor.

But now, an alternative approach is gaining momentum. Many researchers have discovered that analyzing the pattern of DNA fragment sizes in the blood allows them to detect cancers and their location. Most recently, in a study published last week in Nature Biotechnology, the Stanford researchers showed that cell-free DNA (cfDNA) fragment size was sufficient to predict RNA expression indicative of cancer.

The new fragmentomic liquid biopsy method, called EPIC-seq (epigenetic expression inference from cell-free DNA sequencing), relies on promoter fragmentation entropy to predict tissue RNA expression from cfDNA, and was able to detect subtypes of lung and blood cancers with high sensitivity.

“Most of the liquid biopsy methods have focused on detecting genetic mutations or chromosomal abnormalities in the bloodstream,” said Ash Alizadeh, co-senior author and professor of medicine at Stanford University. But according to him, “those can only get you so far.”

Previous research has shown that fragment sizes of free floating DNA in the blood carry information about their tissue of origin. Each tissue has a signature pattern of nucleosome positioning along the length of the chromosome that is directly correlated to DNA fragment size that is shed from the cell during cell death. According to Alizadeh, previous studies have shown that the lengths of these DNA fragments can be used to infer the tissue from which the fragments came.

In their study, Alizadeh and his team wanted to see if the fragments could tell them anything about gene expression. Such information could be used to identify cancer or other abnormalities in gene activity in diseased cells. Determining RNA transcription in the tumor is something that has so far been impossible without direct sampling.

The researchers first described EPIC-seq at the American Association for Cancer Research’s annual meeting in 2020, noting how it can be used to analyze the patterns of DNA fragments to determine which genes are being expressed. Those that are actively being transcribed tend to have an array of fragments that are diverse in length.

In the new study, the team designed an EPIC-seq panel to enable the detection and subtyping of two common cancers, non-small cell lung cancer (NSCLC) and diffuse large B cell lymphoma. They profiled more than 300 samples from patients and controls. Using machine learning techniques, they were able to achieve an area under the receiver operator characteristic curve (AUC) of roughly 0.9 for each type of cancer. “We were struck by how remarkably it correlated,” said Alizadeh.

The researchers termed this approach “promoter fragmentation entropy (PFE)” and announced it as a new metric for inferring RNA expression levels in the cell through circulating DNA.

“Getting RNA expression levels without looking at the RNA is impressive” and a milestone, said Ryan Morin, a genomic cancer scientist at Simon Fraser University in British Columbia who was not involved with the study. “It’s hard to get quality RNA from the tumor but PFE means this is no longer an issue.”

In an accompanying commentary in Nature Biotechnology, Peiyong Jiang and Dennis Lo, cancer researchers at the Chinese University of Hong Kong, said it “adds to our armamentarium,” and that it will “augment current efforts in developing noninvasive biomarkers for cancer detection and monitoring.”

Morin is excited about the prospect of using it for cancer monitoring in patients who are receiving treatment. He said the detection of newly emerging mutations in cancer genes could allow oncologists to adjust precision medicine treatments to fit the evolving tumor progression without having to re-biopsy tissue, which can be a potentially dangerous procedure. “I see this as a technology that can be used to fill in the gaps of cancer care where we can’t get a good tumor sample, when it is not safe to biopsy, or we have to re-biopsy,” he said.

In addition to detecting the presence of cancer, the researchers also showed it can distinguish between cancer subtypes. They focused on 69 genes that were differentially expressed between NSCLC and squamous cell carcinoma.

Some caveats of the approach are that the sensitivity decreases with early-stage cancers. In addition, its ability to infer RNA transcription patterns is still unknown and therefore its ability to deduce cancer type in patients with unknown diagnoses is currently unclear. Nevertheless, the approach sparks hope in many researchers, and not just in the cancer field.

Golnaz Vahedi, professor of genetics at Perelman School of Medicine at University of Pennsylvania, is excited about using this approach in other disease areas, as well, such as autoimmune disease. “This study shows that epigenetics and chromatin is a reflection of disease-specific events in cells,” she said, adding that she thinks the technique could have far-ranging applications and “very elegantly captures disease signatures from the circulating DNA.”

The study’s authors are also interested in exploring other applications. Alizadeh thinks the approach could be used to monitor for toxic side effects of cancer treatment like toxicities in the heart and other organs. He hopes the approach can see damage to organs before symptoms emerge. The researchers also hope to expand the method to other cancers like breast and colon and test whether it can detect cancer evolution over time, such as the development of new mutations and drug resistance.

“The key idea is that we’ve shown we can predict RNA from DNA. And it’s kind of magic that it can work. We’re pretty excited to put it to use, not just in cancer but also in other areas,” said Alizadeh

This article originally appeared on GenomeWeb. Click here for more information.