NEW YORK (GenomeWeb) – Adding to efforts to design standards and reference materials for assessing clinical next-generation sequencing-based tests, researchers from Partners HealthCare have designed a plasmid-based multiplexed control for a targeted hypertrophic cardiomyopathy panel.
The group reported the results of a proof of principle study testing the control in the Journal of Molecular Diagnostics this month. Senior author Birgit Funke, director of clinical research and development at the Partners HealthCare Center for Personalized Genetic Medicine, told GenomeWeb that the group collaborated with SeraCare Life Sciences as a "knowledge provider" and that SeraCare would commercialize the control.
Funke said her lab was interested in designing a control for NGS-based inherited disease panels because molecular diagnostics increasingly relies on NGS. Although labs that develop such tests do validate their assays, she said, validation is typically methods-based, attesting to a lab's ability to call certain types of variants, SNVs, or indels, for instance. "But there's also a separate need for disease-specific validation," she said. Current methods for disease-specific validation rely on analyzing existing patient samples, and verifying that the assay can detect the disease-causing variant in those samples. But because sequencing-based diagnostics is a relatively new field for most disorders, it can be difficult for labs to acquire enough patient samples, Funke said. "We are contacted frequently by labs around world who have trouble getting disease-specific samples," she said. Particularly for heterogeneous disorders that can have many different causative mutations, it can be cost prohibitive to try and acquire samples with each mutation.
That "led to the realization that we need new types of reference materials," Funke said. In its collaboration with SeraCare, the Partners HealthCare group focused on hypertrophic cardiomyopathy because it is a disease in which the lab has extensive diagnostic experience. There are also currently no reference materials available for the disease, and it has a "tricky variant spectrum," Funke said, with one of the disease-associated genes being rich in indels. The disease itself is also very prevalent and has severe health outcomes so it is "important for labs to have the right tools."
The researchers made use of a technique called the control plasmid spiked-in genome that had been previously developed by a team at the Frederick National Laboratory for Cancer Research. They designed control plasmids, each containing around 1,000 bases of sequence, including one mutation of interest, which are then spiked into genomic DNA.
The Partners HealthCare team made slight modifications to the protocol, choosing to leave out a 6-base pair barcode that was included in the original design. Funke said they decided not to include the barcode because they thought it might cause alignment problems.
In order to create the control for hypertrophic cardiomyopathy, the researchers selected 10 pathogenic and likely pathogenic variants, including SNVs, insertions, and deletions, from five different genes. The variants were chosen based on clinical sequencing data that had been collected at Partners' Laboratory for Molecular Medicine.
The team tested two different plasmid designs: In one design, they constructed 10 different 1-kb plasmids, each of which included a single variant. In the second design, they created one 10-kb plasmid that included all 10 variants.
The researchers spiked the 1-kb plasmids into wild-type genomic DNA at three allele frequencies typically seen for heterozygous germline variants: 45 percent, 50 percent, and 55 percent. They also spiked the 10-kb plasmid into genomic DNA at 50 percent.
Then they sequenced the plasmid-spiked DNA samples in quadruplicate, along with the standard reference DNA, using the lab's 62-gene hypertrophic cardiomyopathy panel.
The researchers detected all variants at the expected allele frequencies and also confirmed that the 10-kb plasmid had similar performance for the 10 variants as each of the 1-kb plasmids.
One variant, a deletion in the MYBPC3 gene, is known to be difficult to detect via NGS, and as expected, it was detected at a lower-than-expected allele frequency.
The researchers compared the plasmid controls with data from patient samples, and found the two to be comparable. "It looked really encouraging," Funke said. "The results were essentially indistinguishable from real patient data." She noted that the deletion in the MYBPC3 gene, which was difficult to detect, "underscored the need for labs to test this up front." Such validation is particularly important for labs that may not have expertise in a given disease area and wouldn't necessarily know to look for a certain hard-to-detect variant. In such cases, an NGS panel may come back negative, but may have jut missed the variant, she said.
Funke said that her group planned to develop plasmid-based controls for additional diseases, likely those in which the lab has expertise, such as other types of cardiomyopathies.
Other groups have also set out to develop standards for assessing NGS assays. The National Institute of Standards and Technology's Genome in a Bottle consortium has designed reference genomes that researchers can use to validate their methods by seeing how well their specific technology matches with the "truth set." However, the GIAB reference material represents healthy genomes, so would not assess an assay's ability to detect a specific disease-causing variant.
Also, a group from the Garvan Institute of Medical Research in Sydney, Australia developed a technique that uses synthetic spike-ins called sequins to quantify an assay's ability to detect specific mutations. The sequin method can be used to represent disease-related variants, although it differs from the plasmid-based method in that it is more general, while the plasmid method focuses on specific disease-causing variants.
Funke said that her team's work would add to the "repertoire of tools available to test development and clinical operational quality management," but laboratories would still have to perform other types of validation work, including methods validation and testing of actual patient samples. In addition, she said, it remains to be seen how this type of validation would fit into regulatory guidelines around clinical NGS tests.