Geoffrey R. Oxnard, MD
Preliminary research has shown the potential application of cell-free DNA (cfDNA) in detecting cancer and directing treatment decisions, explained Geoffrey R. Oxnard, MD; however, the approach requires clinical validation before it can be used in practice.
“Today, cfDNA genotyping is used in the advanced setting in patients to figure out the right treatment based on what mutation is in the tumor. We want to turn that into a detection approach where we can find cancer when it's not apparent,” explained Oxnard. “The most exciting application would be as a screening test through free-floating DNA in a patient who is otherwise normal and healthy. That's not the case today, but there are several groups that are working on that.”
Data from The Circulating Cancer Genome Atlas (CCGA) study echo the promise of these strategies. In the trial, early-stage lung cancers were detectable in cfDNA by way of a genome-wide sequencing assay. The trial accrued 12,292 patients, of whom 70% had cancer and 30% did not. Three prototype sequencing assays were used in conjunction with cfDNA, all of which detected ≥38% of stage I to IIIA cancers and ≥89% of stage IIIB to IV cancers.
In an interview with OncLive
, Oxnard, an associate professor of medicine at Harvard Medical School and a thoracic oncologist at Dana-Farber Cancer Institute, discussed the use of cfDNA as a potential method of cancer detection and subsequent assessment of treatment response in patients with solid and hematologic malignancies.
OncLive: What are some ways in which cfDNA is being used to detect cancer?
: CancerSEQ is an assay that uses a next-generation sequencing (NGS) panel to look for key mutations in addition to proteomics. GRAIL is developing a broad genome-wide assay that looks across the entire genome for any abnormal signals; this is done by deleting the white cell signals and leaving the hidden signal that may be related to cancer. [However], there's a long road ahead of us before we will be able to develop a test that can help healthy patients know whether they have cancer or not.
Before that, we are looking for cancer in high-risk populations; that’s generally when we think about minimal residual disease (MRD). In patients who have finished surgery for their lung cancer or colon cancer, we can look in their blood to see if there is remaining evidence of stage IV disease. There's some pretty cool evidence now showing that you can use cfDNA assays in that space to find leftover cancer. [Researchers] have built a bespoke assay of whole-exome sequencing to personalize an assay for a given patient. If you see [a mutation] with high statistical power, you can say that the patient is going to recur.
The CancerSEQ approach uses an NGS panel that is applied to the tumor and is then used to find the genes that are normal and the ones that are mutated. Then, it is applied to the plasma to see if those genes in the tumor are in the plasma. It's complex bioinformatics to use what you know about the tumor that you removed to see if you can find it in the blood. The technical approach is there. The question is, “How can this technological learning become an assay that we use in the clinic?” That's the missing step.
Additionally, who are the right patients for this and what is the performance of an assay in that regard? What sensitivity should we expect? How accurately can I tell a patient they’re going to recur or not? Just like in advanced cancer, sensitivity is going to be average. We're not going to have 100% certainty that a patient is going to recur or not, but we do want to make sure that a test that says, “Cancer is present after surgery,” is reliable. Those are the kinds of assays I hope we can offer our patients sometime soon.
What work still needs to be done in order to get to that point?
Reports so far are largely preliminary validation rather than a rigorous clinical validation in an intended-use population. I treat [patients with] resectable stage II lung cancer, so I want to see how this test behaves in a cohort of 100 patients with resected stage II lung cancer; it's a simple requirement. I don't want an assay that's still in development. I want a rigorous, blinded statistical demonstration of its accuracy before I offer it to patients with stage II lung cancer. That kind of data are missing.