After officially launching the “Identification and Validation of ctDNA Quality Control Materials” Initiative last fall, a team at the Foundation of the National Institutes of Health (FNIH) has identified 14 clinically actional variants that it will incorporate as part of quality control materials to submit to the US Food and Drug Administration for eventual widespread use in circulating tumor DNA (ctDNA) testing.
Having raised $1.5 million in funding since announcing the project in 2017 during a meeting co-sponsored by the FDA and the American Association for Cancer Research (AACR), the group anticipates completing the first phase of the project within the next few months.
Having well-defined cell line-derived reference standards that accurately match clinical patient samples and contain relevant variants may allow researchers to validate new circulating tumor cell or circulating tumor DNA assays.
Dana Connors, a senior project manager at FNIH’s Biomarker Consortium, noted that there has not been much discussion in either the research or clinical markets as to what the standards should be for liquid biopsy test validation and quality control.
“Our concern is that a patient could go to Lab 1, get a liquid biopsy test, and then visit another lab and receive completely different results because Lab 2 is applying a test in a different way,” Connors explained. “If the testing sites aren’t applying the same tools, examining the same analytes, or searching for the same variants, the patient will receive wildly different results.”
FNIH began working with four molecular characterization labs that used reference material donated by commercial partners. The partners included labs led by Chris Karlovich and Mickey Williams at Frederick National Laboratory for Cancer Research’s Molecular Characterization Laboratory, Kenneth Cole at the National Institute of Standards and Technology, and Cloud Paweletz at Harvard University, as well as biopharmaceutical firm AstraZeneca.
Williams originally debuted the project during a 2017 webcast sponsored by the FDA and AACR, and the group has since raised $1.5 million for the project, in addition to “in-kind support” from public and private partners.
“In order to get this problem under control and help clinical practice and patient benefits, the goal is to make commercially prepared controls and partner with [reference material] firms,” project consultant Robert McCormack, who previously served as Head of Biomarker Strategy at Johnson & Johnson when the initiative began, explained. “[We then] validate the controls in independent labs so that any lab any lab around the world can see the result, understand what their performance should be, and if they conform to it, then they’re on good ground for reporting the result.
In Phase I, led by Williams, the team acquired the sample material and standardized how the material could be extracted in a centralized lab. When measuring ctDNA, the researchers searched for common variants in all genomic classes, including single nucleotide polymorphisms, copy number variants (CNVs), fusions, and indels.
Williams noted that the labs used five methods, including two droplet digital PCR (ddPCR) and three next-generation sequencing (NGS) assays representing both amplicon-based and hybrid capture-based platforms, to search for potential mutation targets and identify any potential biases among the different assays.
Running the tests on Illumina sequencing instruments, the labs identified about 20 common variants among the four classes that most commercial tests use — including BRAF, EGFR, and KRAS — and sent the information to the centralized lab, where National Cancer Institute statisticians aggregated the data. From there, the group ran a performance evaluation on the mutations and eventually chose 14 that were clinically actionable and covered several different cancer types.
Connors noted that the group members have begun to discuss the results and review information with manufacturing companies that provided the material, as well as companies providing additional funding.
With liquid biopsy genetic tests being pushed closer toward the clinical space in the last few years, major commercial groups are developing or offer their own reference materials. Many of these firms are involved in Phase I of the project, including Thermo Fisher Scientific, Horizon Discovery, and SeraCare Life Sciences. Other private sector partners in the project include Roche subsidiary Genentech, Janssen R&D, Merck Sharp & Dohme, and Pfizer.
“Once the performance evaluation is complete, we will publish a manuscript that will, from an anonymous perspective, show how we found the results,” Connors said. The team expects to complete Phase I within the next few months after submitting a manuscript for publication.
“[The study] was pre-competitive, so we made the manufacturers understand that we wanted to work together and as collaboratively as possible to figure out what [variants] worked and don’t work together,” Williams said. “While we requested that each manufacturer had the minimum number of variants that we agreed upon, they could also build the list into a larger panel if it was in their best commercial interest.”
In Phase II of the project, the researchers will begin discussions with the FDA to see if a proposed two-part study design and statistical analysis methods are an acceptable approach for functional characterizations of the quality control materials.
Karlovich, who is leading the second phase, explained that the first part of the analytical validation study will help the team determine if the quality control materials perform “similarly to a clinical specimen near the limit of detection.”
“We will use an NGS assay that uses hybrid capture for target enrichment,” Karlovich explained. “We will test several common variants in both the quality control materials as well as in clinical samples [that] we have procured and do the testing at and below the chosen assay’s limit of detection, which is near 0.1 percent variant allele frequency.”
In the second part of the study, the researchers will look at the commutability of the same variants used in the first part, but in the quantitative range of the three assays platforms used in Phase I.
“Using ddPCR, hybrid capture NGS, and amplification NGS, we will test common variants in the quality control materials and clinical samples in a range from 0.1 percent to five percent variant allele frequency,” Karlovich explained.
Connors highlighted that the meeting with the FDA will establish the groundwork for reference material manufacturers to partner with the organization. Once completed, the project will allow the firms to seek “qualification of their reference materials as validated quality control materials.”
However, Connors emphasized that the FNIH project team will not seek qualification of the materials on the manufacturers’ behalf in the second phase of the project.
In Phase III, the team — led by Paweletz — will observe how the quality control material performs against clinical samples. By launching a clinical validation study, the researchers will collaborate with 10 external labs to run the quality control materials on their NGS and ddPCR assays, along with clinical samples Paweletz’ team previously obtained.
After outside labs provide feedback on their results, the team will work with NCI statisticians to compare and demonstrate how well the materials performed across the different labs.
“We try to be cognizant of [the fact] that companies that have specific assays are not put at a disadvantage, so we [will] take these into account as we design [Phase III] with the vendors in mind,” Paweletz said. “Having said that, the project will mimic some of the design elements of Phase I and Phase II.”
While the FNIH has communicated with BloodPac and other government-based consortiums looking at reference materials, including Cancer-ID, Friends of Cancer Research, and the UK’s National Institute for Biological Standards, Connors highlighted that his team plans to prevent redundancies and not duplicate data as part of the project.