Skip to main content
Premium Trial:

Request an Annual Quote

Lung Cancer Study Provides Clinical Validation for Digital PCR-Based Liquid Biopsy


NEW YORK (GenomeWeb) – A new study suggests genotyping cell-free DNA (cfDNA) in the blood is a feasible method for prospectively picking up common, treatment-informative mutations in the EGFR and KRAS genes in individuals with advanced lung cancer.

As they reported in JAMA Oncology, researchers from the Dana-Farber Cancer Institute and other centers in Boston used a Bio-Rad droplet digital PCR system to genotype cfDNA in blood samples from nearly 200 individuals with advanced non-small cell lung cancer.

Compared with direct biopsy genotypes, the ddPCR-based liquid biopsy approach picked up the EGFR or KRAS mutations of interest with between 64 percent and 82 percent sensitivity — a range that appears to reflect differences in the proportion of tumor DNA shed into the blood. The team noted that the presence of metastatic tumors in the bone, liver, or at multiple sites bumped up the ability to detect tumor mutations when present.

"The sensitivity [of the assay] is entirely related to DNA shed, we believe," senior author Geoffrey Oxnard, a physician with Dana-Farber's Low Center for Thoracic Oncology, told GenomeWeb. "The most important covariant in sensitivity was the number of metastatic sites — as you have more disease and more [tumor] burden, you have more shed and, therefore, the assay is more valuable."

Consequently, Oxnard noted that the assay has the greatest value in metastatic NSCLC and may be more difficult to use for screening or very early-stage lung cancer.

In a related editorial in JAMA Oncology, Frederick National Laboratory for Cancer Research investigator Mickey Williams and Barbara Conley, with the National Cancer Institute, called the clinical validation study "a step in the right direction," but warned that the performance characteristics described for the assay will likely make it better suited for tracking treatment than finding mutations to guide treatment at the outset.

"Owing to low sensitivity and high [positive predictive value] and specificity, this approach is probably best suited for detection of resistance mutations and for serial plasma testing to assess treatment response, and should not replace tumor biopsy assessment for initial treatment decision making."

To explore this further, Dana-Farber Cancer Center plans to launch a ddPCR-based cfDNA assay for EGFR and KRAS mutations for guiding treatment in newly diagnosed NSCLC and NSCLC with acquired resistance to EGFR kinase inhibitors — the indications included in the current validation study. Oxnard said the assay has already been validated in a CLIA lab and is awaiting final administrative approval.

"In establishing that assay in the pathology lab, we are launching a prospective trial to initiate treatment based on a blood test without the tumor result, to show that the results you get when you treat based on a blood test are just as good as the results when you're treating based on a tumor test," he said.

The study represents a shift towards using ddPCR for plasma genotyping to prospectively guide treatment decisions for NSCLC patients, said George Karlin-Neumann, director of scientific affairs at Bio-Rad's Digital Biology Center.

"What really distinguishes this from a lot of other studies … is that this is really a prospective, real-world assessment of clinical performance of droplet digital PCR for following patient disease status in plasma cfDNA," he said.

He noted that the ddPCR-based liquid biopsy approach appears to be particularly amenable to cancer types such as NSCLC or melanoma that are known for having a relatively small number of causative mutations in a large proportion of tumors.

Oxnard and his colleagues from the Dana-Farber Cancer Institute have been pursuing clinical applications of ddPCR-based plasma genotyping in lung cancer for some time. In a paper published in Clinical Cancer Research in 2014, for example, they used Bio-Rad's QX100 ddPCR system to search for the treatment resistance-related T790M change in the EGFR as a means of tracking erlotinib chemotherapy response and predicting treatment resistance and relapse.

For their latest analysis, the investigators relied on a Bio-Rad ddPCR-based liquid biopsy targeting three EGFR mutations and one KRAS glitch in a prospective, clinical setting.

"This is a prospective clinical validation proving that the assay, in the clinical setting, can behave as we hoped it would behave in that prior paper," Oxnard said.

In particular, the team targeted an exon 19 deletion and an exon 21 L858R substitution in EGFR, which are used to determine eligibility for treatment with erlotinib, an EGFR kinase inhibitor; an EGFR substitution called T790M that's been implicated in erlotinib resistance; and a range of substitutions at KRAS codon 12, known as G12x mutations.

Because KRAS and EGFR mutations are typically mutually exclusive, the presence of oncogenic changes in KRAS suggests EGFR kinase inhibitors will not be effective in a given NSCLC case.

"If you have an EGFR mutation, you could start treatment with a drug like erlotinib. If you have an EGFR T790M mutation, a resistance mutation, you could start treatment with a drug like osimertinib [a third-generation EGFR kinase inhibitor marketed as Tagrisso by AstraZeneca]," Oxnard said. "If you have a KRAS mutation in your plasma, you can initiate chemo and you don't need to purse further genotyping. So each of these has clinical meaning."

He and his colleagues applied this approach to search for EGFR and KRAS mutations in blood samples from 68 men and 112 women with advanced NSCLC who ranged in age from 37 to 93 years old. Two-thirds of the cases represented newly diagnosed advanced NSCLCs, the researchers noted, while the remaining 60 advanced NSCLCs had acquired resistance to EGFR kinase-inhibitor treatment.

It took anywhere from one to seven business days to turn around test results from the ddPCR-based assay of cfDNA in patient blood samples (two to three days, on average, for the patients with acquired resistance or new diagnoses, respectively). In contrast, tissue biopsy genotyping ranged from one to 54 days in the new NSCLC cases and from one to 146 days in samples from advanced NSCLC patients with acquired resistance.

By direct tumor biopsy genotyping, the team knew that at least 80 of these individuals had EGFR-sensitizing mutations (exon 19 deletion or L858R), 35 individuals had the resistance-related T790M change in EGFR, and more than two-dozen individuals had G12x mutations in KRAS.

Using ddPCR on cfDNA, the researchers correctly identified 82 percent of the EGFR exon 19 mutations, 77 percent of EGFR T790 acquired resistance mutations, 74 percent of the EGFR L858R mutations, and 64 percent of the KRAS mutations.

In individuals with positive ddPCR-based tests for exon 19 deletion in EGFR, the EGFR L858R substitution, or G12x substitutions in KRAS, researchers saw corresponding mutations in all of the directly genotyped biopsy samples, representing a positive-predictive-value of 100 percent.

The assay's PPV dipped to 79 percent when they searched for the EGFR T790M change associated with acquired resistance to EGFR inhibitors — potentially due to either false-negative results in tissue genotyping for this mutation, false-positive T790M mutations in blood, or a combination of the two.

For a subset of 50 NSCLC patients who had detectable EGFR or KRAS mutations in the plasma genotyping assay, the researchers used ddPCR to quantify such mutations in plasma samples after treatment, offering a blood-based look at changes in cfDNA levels and mutation patterns.

They are now set to do more extensive plasma response analyses using cfDNA to explore clues that targeted mutations in cfDNA can provide to treatment response, outcomes, and relapse.

Oxnard acknowledged that the mutations interrogated in cfDNA for the current study represent the "low-hanging fruit" as far as recurrent mutations in NSCLC go. He and his co-authors also cautioned that the ddPCR-based plasma genotyping approach falls short for finding copy number changes or structural rearrangements in circulating tumor DNA.

In another paper published in Clinical Cancer Research last year, Oxnard and his team described a targeted next-generation sequencing assay that they developed to detect actionable mutations and rearrangements in cfDNA from four-dozen advanced NSCLC patients.

In general, the researchers hope to establish a set of complementary liquid biopsy assays to rapidly test for a range of clinically relevant mutations in a platform-agnostic manner.

Bio-Rad's Karlin-Neumann noted that ddPCR appears to be finding favor in the clinical research setting, as the approach is part of the design for several trials on

He said the company is working on updates to its systems that are expected to ease this transition into a clinical setting, including strategies to develop an all-in-one system that may replace individual thermocycler, droplet generator, and droplet reader instruments. 

"There is a recognition that as you move toward a clinical setting, it's desirable to have the closest thing possible to 'sample in and answer out' in the form of a report," Karlin-Neumann said. "That encompasses anything from sample processing on the upstream end of it … [to] taking the platform detection approach and compressing it into a single instrument."