A novel approach that relies on artificial intelligence to detect cancer in the blood is able to identify seven different cancer types in a single liquid biopsy in most patients tested to date, and also accurately identifies the tissue of origin in up to 75% of cases, a team of US researchers report.
The test is also proving to be exceptionally reliant in healthy controls, misclassifying few healthy patients as having cancer, the same research indicates.
However, these data come from an early proof-of-concept study and need to be validated in a prospective clinical trial before the true clinical utility of the test can be assessed, cautions an expert.
"Existing liquid biopsy tests typically focus on mutations in the DNA sequence within a cancer cell or methylation changes, but unfortunately, not all cancer patients have changes that are detectable using these methods and typically [these biopsies] are very limited," Victor Velculescu, MD, PhD, professor of oncology, Johns Hopkins Kimmel Cancer Center, Baltimore, Maryland, explained to Medscape Medical News during a press briefing.
"In contrast, our approach is very global — it looks broadly across the genome at what we call fragmentation profiles of cell-free DNA (cfDNA), and by analyzing these fragmentation profiles in the bloodstream, we can identify abnormalities genome-wide and avoid many of the pitfalls that have to do with specific mutations or other factors," Velculescu noted.
The research was published online May 29 in Nature and also presented during a press briefing organized by John Hopkins.
During the briefing, researchers explained that the novel test, known as DNA Evaluation of Fragments for Early Interception (DELFI), first underwent a proof-of-concept study. It was trained to detect the presence of cancer in blood samples from 208 patients with breast, colorectal, lung, ovarian, pancreatic, gastric, or bile duct cancers. Participants were from the United States, Denmark, and the Netherlands.
The same test was also used on blood samples from 215 healthy volunteers.
As Velculescu explained, the nuclei of healthy cells package DNA like a "well-organized suitcase" in which different regions of the genome are carefully placed in various compartments.
"In contrast, the nuclei of cancer cells are more like disorganized suitcases with items from across the genome thrown in haphazardly," he noted.
As a cancer genome is disorganized in the way it is packaged, "we found that when cancer cells die, they release their fragmented DNA in a chaotic manner into the bloodstream," Velculescu continued.
By examining cfDNA, DELFI is able to detect abnormalities in the size and amount of DNA in different regions of the genome based on how it is packaged.
Using this novel liquid biopsy approach, the research team was able to detect cancer in 73% of cancer patients overall, and misclassified only four patients out of 215 healthy volunteers as having cancer, giving the test a specificity of 98%.
Importantly, the majority of cancer samples analyzed were early stage cancers that were amenable to surgical resection, "a very important characteristic of this test because if you are just looking at late stage disease, obviously the test would not be nearly as helpful," Velculescu observed.
Because the genome-wide fragmentation patterns also reveal differences associated with specific tissues, fragmentation patterns can also indicate the source of that cancer such as the breast, colon, or lung.
For example, in the current analyses, DELFI was found to be 61% to 75% accurate in identifying the tissue of origin of the cfDNA, the team reports.
"What this means practically is that when you get a test like this in the future, you'll first get a result that says it's a positive test — it's a cancer signal — but then it's very important to know where the cancer is coming from because that will affect how you treat the patient," Velculescu said.
"So one of the benefits of this test is that it also tells you in a high fraction of cases where the tumor appears to be coming from and then that can help focus subsequent diagnostic tests and ultimately interventions," he added.
Velculescu indicated that their international research team is in the process of scaling up their analyses and they hope to study the performance of DELFI in many more thousands of individuals.
However, they are already "very encouraged" by the potential of this new technology because it looks at a completely independent set of DNA characteristics than those that have proven extremely challenging to the research community over the years.
Most immediately, Velculescu envisions DELFI being used in a high-risk patient population such as smokers or patients with a hereditary predisposition to cancer where a test specificity of 98% would still prove to be useful for those at high risk for cancer.
On the other hand, the test should not prove to be that costly, he added, as it is very easy to administer and needs only currently available, inexpensive laboratory methods to deliver results.
"As such, we expect DELFI will be more cost-effective than other cancer tests," Velculescu opined.
Holy Grail of Cancer Research
Asked by Medscape Medical News to comment on the findings, Daniel Hayes, MD, Stuart B. Padnos professor of breast cancer research, University of Michigan Rogel Cancer Center in Ann Arbor, noted that one of the holy grails of cancer research is finding a meaningful and accurate method of screening to detect cancers that might be more effectively treated early rather than late.
"This objective is a huge undertaking, with many obstacles, which is why there are so few really established screening strategies — arguably imaging for breast cancer, PAP smears for cervical cancer, and colonic visualization (for example, colonoscopy) for colon cancer," he elaborated in an email.
On the other hand, the prostate-specific antigen (PSA) story for prostate cancer is a sobering example of screening on steroids, leading to over-diagnosis and unnecessary treatment, as Hayes pointed out.
Nevertheless, a circulating biomarker that is an effective screening strategy would be a very welcome addition to the field of oncology.
"The current study is, as the authors note, a 'proof of concept' investigation that does, apparently, work in a case-control design," as Hayes suggested.
"And without that success, there would be no reason to even consider moving forward, and I applaud Velculescu and colleagues for making this progress," he added.
However, the real proof of true clinical utility will still require a prospective trial involving many patients in which the assay is performed on participants who walk in the door with no prior diagnosis of cancer to see if they can separate out those who have cancer versus those who do not.
"Whether finding a 'positive' test [on this assay] results in treatment that improves overall, or cancer-specific, survival, or at least decreases the odds of long-term distant metastases is even more important," Hayes noted.
"But without [such a trial], the assay cannot be applied to guiding patient care, since it would not have clinical utility, although I am sure we all look anxiously for such data," he concluded.
Velculescu is a founder and shareholder of Delfi Diagnostics and Personal Genome Diagnostics, and a member of their scientific advisory boards. He has also been an advisor to Daiichi Sankyo, Janssen Diagnostics, Ignyta, and Takeda. Hayes owns stock or has other ownership interests in Oncimmune and InBiomotion and has served as a consultant or advisor to Cepheid. He has also received research funding, mostly for his institution, from AstraZeneca, Puma Biotechnology, Pfizer, Eli Lilly, Merrimack Pharmaceuticals, Parexel, and Menarini Silicon Biosystems (Veridex/Johnson & Johnson). He holds a number of patents and receives royalties for some of them.
Nature. Published online May 29, 2019. Abstract
For more from Medscape Oncology, join us on Twitter
Medscape Medical News © 2019 WebMD, LLC
Send comments and news tips to firstname.lastname@example.org.
Cite this: Novel Liquid Biopsy Opens Door to Pan-Cancer Testing? - Medscape - May 31, 2019.