cfDNA Assays Demonstrate High Specificity for Early Stage Lung Cancer Detection


Initial results from the Circulating Cell-Free Genome Atlas study strongly suggest that cell-free DNA tests can be used with a high degree of specificity to detect signs of early stage lung cancer.

Geoffrey R. Oxnard, MD

Initial results from the Circulating Cell-Free Genome Atlas (CCGA) study strongly suggest that cell-free DNA (cfDNA) tests can be used with a high degree of specificity to detect signs of early stage lung cancer, investigators reported at the 2018 ASCO Annual Meeting.1

Three prototype sequencing assays were used: paired cfDNA and white blood cell (WBC) targeted sequencing for single nucleotide variants/indels; paired cfDNA and WBC whole genome sequencing (WGS) for copy number variation; and cfDNA whole genome bisulfite sequencing (WGBS) for methylation.

The “exhaustive” screening applied through parallel use of these assays enabled investigators to screen out WBC nontumor mutations (“noise”)—which have the potential to produce false-positives—and achieve 98% specificity for tumor detection, according to lead author Geoffrey R. Oxnard, MD, of Dana-Farber Cancer Institute.

The WGBS assay detected 41% of stage I-IIIA cancers and 89% of stage IIIB-IV cancers in a population of 127 patients with lung cancer. Investigators found that the signal for cancer increased with higher stages of tumor advancement. In the same population, the WGS and targeted assays were similarly effective, detecting 38% and 51% of early stage cancers and 87% and 89% of late-stage cancers, respectively.

The data represent a subgroup analysis from the CCGA study, and are promising evidence that cfDNA assays have the potential to be developed into effective diagnostic tools for cancer screening, Oxnard said.

“Two years ago, it was a pipe dream,” he added. “It’s a huge [advance] and it means this is going to be a reality going forward.”

Additional results of cfDNA testing for breast, gastrointestinal, gynecologic, blood, and other cancers will also be presented at the ASCO meeting.

Investigators have enrolled 12,292 patients in the study, 70% with cancer and 30% without. Two initial cohorts were created—a training set and a test set. The training set encompassed 1733 clinically evaluable samples, including 127 positive for lung cancer. The test set encompassed 980 clinically evaluable samples—47 positive for lung cancer.

A comparable control cohort was important for accurate analysis of the false positive rate, Oxnard said, and that was achieved in this study. Whereas in the training cohort, the detection rate was 41% for early stage lung cancer (95% CI, 29-54), it was 50% (95% CI, 29-71) in the test group using WGBS. For advanced cancer detection, the respective numbers were 89% (95% CI, 77-96) versus 91% (95% CI, 71-99).

Although lung cancer screening using low-dose computed tomography (LDCT) is shown to improve survival and is recommended by the United States Preventive Services Task Force for those aged 55 to 80 years who have been heavy smokers, it is not widely adopted. A recent analysis of LDCT use at 1800 US lung cancer screening sites found that just 1.9% of more than 7 million current and former heavy smokers received this screening in 2016. 2

“Lung cancer screening rates are much lower than screening rates for breast and colorectal cancers, which is unfortunate,” lead study author Danh Pham, MD, a medical oncologist at the James Graham Brown Cancer Center, University of Louisville, Kentucky, said in a May release. “It is unclear if the screening deficit is due to low provider referral or perhaps patient psychological barriers from fear of diagnosis. Lung cancer is unique in that there may be stigma associated with screening, as some smokers think that if cancer is detected, it would confirm they’ve made a bad lifestyle choice.”

Results from the study by Pham et al, which are also being presented at this meeting, found that even in the South, which has the highest number of accredited screening sites (663) and numbers of smokers eligible for screening (>3 million), the screening rate was only 1.6%, the second lowest in the country. By comparison, the West screens patients at a rate of 1% annually and has the lowest number of screening sites (232).

Such findings justify efforts to search for screening methods such as the prototype assays employed by Oxnard and colleagues, which potentially could enable patients to obtain a blood draw and subsequent noninvasive screening for lung cancer during a routine visit to a doctor’s office, experts said at the ASCO presentation this week.

“This is an important first step to detect lung cancer at earlier, more curable stages,” said ASCO expert David L. Graham, MD, medical director at Levine Cancer Institute in Charlotte, North Carolina.

Richard L. Schilsky, MD, chief medical officer for ASCO, sought clarification from Oxnard about whether the cfDNA multiple assays could in fact succeed in diagnosing lung cancer in patients who had no prior diagnosis—unlike those in the CCGA study. Oxnard indicated there is a possibility of that.

Concern was also expressed about potential significance of the WBC “noise” that was being filtered out by the prototype assays used in CCGA.

“The white blood cells are rich with mutations, which can pollute the DNA and make you think that there’s cancer present in the cfDNA, which in fact is shed from the white blood cells,” Oxnard said. “You screen out this interference from the white blood cells and other biologic noise, and you’re left with the final features—mutations, copy number variations, and metylation signatures, which then go into the final assay being done.”

Schilsky asked whether any evidence of mutations can rightly be discarded during this process. “Is it really noise or is there some clinical or biological consequence?”

“It’s true,” Oxnard responded. “These are real mutations, and they can have very real health implications. They don’t mean cancer, to our knowledge, but that needs more follow-up as well.”

He added that, ultimately, this is a question that needs to be answered during the process of sifting through the evidence to come up with a viable diagnostic tool. “It needs to be accounted for in order to achieve a cancer detection assay. That is the most important part of this.”


  1. Oxnard GR, Maddala T, Hubbell E, et al. Genome-wide sequencing for early stage lung cancer detection from plasma cell-free DNA (cfDNA): The Circulating Cancer Genome Atlas (CCGA) study. J Clin Oncol. 2018;36(suppl; abstr LBA8501).
  2. Vast majority of heavy smokers not screened for lung cancer despite USPSTF recommendations [press release]. Alexandria, VA: ASCO; May 16, 2018.

<<< 2018 ASCO Annual Meeting

Related Videos
Stephen V. Liu, MD
S. Vincent Rajkumar, MD
A panel of 5 experts on lung cancer
A panel of 5 experts on lung cancer
Millie Das, MD
A panel of 5 experts on lung cancer
A panel of 5 experts on lung cancer
Angela Jia, MD, PhD, of University Hospitals
Robert Wang, MD, of Fox Chase Cancer Center
D. Ross Camidge, MD, PhD