NEW YORK – Johns Hopkins University spinout Delfi Diagnostics has staked its claim in the growing push to harness genomic and epigenomic technologies for non-invasive early cancer detection and screening.
Announcing a $100 million financing on Tuesday, the company emerged from what has been a quiet period since its founding in 2019, disclosing plans to use the funding to support clinical validation of its technology through prospective studies.
Like Illumina's Grail, the most prominent player in this emerging market, Delfi's platform is epigenomic. The company's technology detects the presence of cancer in an individual's blood by analyzing molecular signals generated from various biological machinery that modify and/or process DNA molecules.
But unlike Grail's approach, which is based on detection of cancer-specific methylation patterns, Delfi has developed a way to glean this information from DNA fragmentation patterns, which reflect how nucleosomes are positioned throughout a cell's DNA architecture.
"We've been in the liquid biopsy space for a long time ... and there have been important discoveries along the way. But really, it wasn't until the Delfi methodology was developed that we began to be really excited about this being useful for screening and early detection," Victor Velculescu, the company's founder and CEO, said in an interview.
"I think the reason for that is twofold," he added. "One, it's an approach that has high performance … and then the other is that [it's] very simple to do and very cost effective, so one can easily envisage this being widely distributed."
"In this day and age, when we're thinking about public health and the cost of our health care system and so forth, it's actually crucial to think about early detection approaches that are cost effective," Velculescu said.
At its core, Delfi's cancer detection is based on the principle that abnormal mitosis, or cell division, is a fundamental hallmark of cancer.
"We are looking at naturally occurring fragmentation of cell-free DNA across the genome and observing patterns that come from that, and it turns out that those patterns are very different in individuals with cancer versus healthy individual," Velculescu said.
According to Nic Dracopoli, Delfi's chief scientific officer, this is possible because of the way DNA is organized in chromosomes into two basic forms: open chromatin and condensed chromatin.
"[T]hese are very much involved in cell-specific patterns of gene regulation … so each cell type has a very distinct pattern," he explained, as do cancer cells compared to normal cells.
When cancer cells die, either under attack by the immune system or through apoptosis or necrosis, their DNA enters circulation and is digested by endogenous nucleases. Based on the pattern of open versus closed chromatin the result will be a different pattern of resulting DNA fragments.
Because of this biology, DNA fragmentation patterns can indicate not only the presence of cancer, but also link back to a specific cell or tissue of origin, something crucial for multi-cancer screening, where a positive result without indication of where to look for the cancer could be difficult, if not impossible, to follow up clinically.
The team has so far found that there is minimal crossover between cancer-associated fragment signals and similar patterns that might be present in other diseases or generated as people age, Dracopoli said.
Delfi and its founders are not the only group to have hit on the potential of fragmentomic analysis for early cancer detection.
In a 2016 study, researchers at the University of Washington, led by Jay Shendure, reported that they could infer where in the body cell-free DNA molecules originated based on fragmentation patterns, with early evidence that the same could be true for distinguishing cancer from non-cancer.
More recently, Stanford University researchers have shown that DNA fragmentation patterns can also be used to predict tissue- and tumor-specific gene expression without the need to sequence expression products or characterize specific gene mutations.
According to Velculescu, what's unique about Defi compared to other specific methods is that it incorporates fragmentation patterns from across the genome while requiring relatively low sequencing coverage.
"This means that the test can be made inexpensive and easy to do and so forth … while still having a profile that's connected to biology," he said.
As various blood-based early detection methods have begun to advance toward the clinic, researchers have explored whether different biomarker sources or signals — mutations, methylation, fragmentation, etc. — might be synergistic with one another, offering even higher sensitivity in combination than they do alone.
Velculescu said that so far the added value of combining fragmentomics with mutations has been marginal. And the downside of moving to a more complex multi-analyte combination would be losing the simplicity and low cost of Delfi's method.
Dracopoli added that fragmentation patterns themselves are in some ways a surrogate of some of the other events that could provide a signal of an unknown cancer.
"If you are measuring a mutation or a copy number or a gene amplification or an epigenetic change [all] of those … also affect fragmentation."
Over the past 18 months, Delfi has been building a retrospective test performance dataset that now includes thousands of individuals.
Velculescu declined to detail any updated sensitivity data for the method from this new work, but he said the results he and his colleagues have been generating are in the same range as what they described in their initial 2019 Nature publication describing the method.
In that study, the researchers analyzed DNA fragmentation profiles in a much smaller set of samples from 236 patients with breast, colorectal, lung, ovarian, pancreatic, gastric, or bile duct cancer and 245 healthy individuals.
They could detect cancer with sensitivity ranging from 57 percent to more than 99 percent depending on the tumor type, while maintaining 98 percent specificity. In 75 percent of cases, fragmentation profiles accurately reflected the tissue or origin, or location of the tumor.
"The objective initially was to extend the studies that were done at Hopkins in an independent effort … and now that we've gone on to that stage, I think we're excited to move into a larger and more public phase of the company," Velculescu said.
The company has a few immediate goals. One is to build out its team in order to expand R&D efforts and prepare to develop the clinical evidence necessary to ensure adoption.
According to Dracopoli, the funds will be especially important for large prospective clinical studies, for which the team is currently working on protocols and study designs.
As various other epigenomic methods have staked their own claims in the cancer early detection/screening space, firms have divided across one major line: either focusing, at least at first, on a specific cancer type, or dedicating their efforts to multi-cancer screening.
Although Delfi views its fragmentomic method as applicable across cancers, as evidenced in the 2019 study, the company initially plans to create tests designed to detect specific cancers where early detection has already been demonstrated to have a clinical benefit. After that, it will expand to a model for multi-cancer screening in a general population.
"We want to be able to detect the cancers that are known to be important for early detection … which are some of the most common, so lung cancer, breast cancer, colon cancer, for example," Dracopoli said.
"Obviously, the regulatory and development issues around a pan-cancer test are much more complicated, so in a sense … we're sort of trying to have our cake and eat it, if you will, in terms of doing both," he added.
"But because we are measuring these consequences of abnormal mitosis the test is very much able to detect all solid tumors," he stressed. "Even much more than methylation, [this] is very much a measure of a very core characteristic of cancer: the unregulated mitosis that leads to the inability of the tumor cell to package its DNA appropriately."