SAN FRANCISCO (GenomeWeb) – Despite the proliferation of next-generation sequencing panels to evaluate somatic mutations for cancer patients, there has been little evaluation of the methods and design of gene panels.
In a study published this week in the Journal of Molecular Diagnostics, researchers from the BC Cancer Agency in Vancouver, British Columbia, evaluated both hybrid capture and amplicon-based targeting for the design of a 32-gene panel for lymphoid cancers, and found that the hybrid capture technology performed significantly better.
The group now plans to test the 32-gene panel in clinical trials to establish its clinical utility for prognosis, subtyping, and selecting the appropriate therapy for lymphoma patients.
Christian Steidl, who heads the lymphoid cancer department at the BC Cancer Agency, said that the goal of the study was to "establish the best platform and to pick smartly the genes that should be represented" in order to develop a clinically useful test.
Although amplicon sequencing has the advantages of a simpler workflow and the ability to work with lower input volumes, Steidl said that a key consideration was which pipeline would work better on "routinely acquired formalin-fixed paraffin-embedded tissue."
In a commentary accompanying the study, researchers from Stanford University and the University of Pennsylvania wrote that "there is a dearth of studies that provide a scientifically rigorous first-hand experience of difficulties, failures, and successes during the iterative design and implementation of a targeted NGS panel specifically tailored for lymphomas." In addition, they added that although differences between hybrid capture and amplicon sequencing have been known, "head-to-head comparisons … are lacking."
The researchers evaluated both Agilent's SureSelect hybrid capture method and Illumina's TruSeq amplicon technology on FFPE lymphoma samples. Sequencing was done on the Illumina MiSeq.
Overall, Steidl said that the hybrid capture had higher sensitivity in detecting variants than the amplicon pipeline. He said that the hybrid capture technology could deal with the FFPE tissue better. The fixing process for FFPE "introduces errors and noise; it physically modifies the DNA and introduces artifacts," he said. "Amplicon-based techniques amplify that noise, while with capture sequencing you have the opportunity to filter out the noise," Steidl said.
In the study, the researchers first tested a 20-gene panel on both fresh frozen and FFPE samples from eight lymphoma patients. They compared both the hybrid capture and amplicon-based techniques and validated the mutations called with Sanger sequencing. Of 16 variants that were validated by Sanger, all were detected in both fresh frozen and FFPE tissue by the hybrid capture method, but seven were missed by amplicon sequencing. The missed mutations occurred in regions that were difficult to amplify due to high GC content or other regions.
Steidl added that amplicons tend to be better for hotspot mutations rather than looking at large exonic or whole-gene regions due to the allele dropout that was observed. "We have to create many amplicons to capture the entire space, so there's a higher likelihood for dropouts, either because the PCR doesn't work in a specific region or because primer design is systematically difficult" because of things like high GC content.
The team compared the two technologies across samples from an additional 41 lymphoma patients and found that across nearly every metric, the hybrid capture technology performed better. Coverage was more uniform and there was less dropout.
The researchers next designed an assay based on the hybrid capture method to focus specifically on the most common lymphocytic malignanices: chronic lymphocytic leukemia, diffuse large B-cell lymphoma, and follicular lymphoma. They identified 32 genes that fell into three categories: genes that were correlated with outcomes, mutations associated with a known drug target, and mutations that can distinguish between disease subtypes. The panel covers the exons of the 32 genes, spanning 263 kilobases of sequence.
Although there are some commercially available panels for leukemias and lymphomas, like Foundation Medicine's FoundationOne Heme , Steidl said that "it's important to build panels together with experts in the field," and "based on the clinical needs of the centers" themselves. He added that although Foundation Medicine's panel is good for the field, it is important to also "maintain a diversity of assays."
Once the researchers designed the panel, they compared the mutations called with whole-genome sequencing of 66 samples and Sanger sequencing of 588 variants. Overall, sensitivity was 93 percent and specificity was 100 percent. In addition, the team analyzed 219 cases and validated that the mutation patterns and frequencies in those cases were consistent with what has been previously reported in the literature.
When looking at samples where whole-genome sequencing called a mutation that the panel missed, the researchers found that often the panel had identified the variant, but it was below the established 10 percent cutoff or had too few supporting reads. The researchers noted that these cases were typically associated with poor quality of the FFPE sample.
Steidl said that the next step is to demonstrate the clinical utility of the panel to determine whether it leads to better patient care. He said that the group is beginning to set up clinical trials with the Canadian Cancer Trials Group and eventually hopes to offer the test routinely to all patients in the Canadian healthcare system.