NEW YORK (GenomeWeb) – Researchers from the Dana-Farber Cancer Institute and Resolution Biosciences have identified more actionable gene fusion mutations with higher allele frequencies in lung cancer patients using Resolution Bioscience's ctDx-Lung liquid biopsy test than using Guardant Health's Guardant360 assay.
One of the major issues that researchers have encountered regarding liquid biopsy assays is the lack of evaluation and validation between commercial testing. Recent attempts to compare cfDNA assays have not converged on clinically actionable mutations or applied a tumor sample to verify plasma results. In addition, researchers have found that certain mutation types like fusions are more challenging to detect than other types with cfDNA assays.
In a study presented in a poster at the American Association for Cancer Research conference in Atlanta, the Dana-Farber and Resolution researchers retrospectively compared the ability of the ctDx-Lung assay with Guardant360 to detect fusion-positive NSCLCs from blood samples using solid tumor as a reference standard.
"A few years ago, Guardant presented data from about 8,000 NSCLC patients, but their fusion detection rate didn't match what you'd expect based on [The Cancer Genome Atlas] or the literature," Resolution CEO Mark Li said. "So we thought that their fusion detection wasn't great."
Li said that Resolution began working with Dana-Farber researcher and principal investigator Geoffrey Oxnard on the study in November 2018, when he had identified a patient in a clinical trial with a confirmed NSCLC RET gene fusion using fluorescence in situ hybridization data. When Oxnard couldn't find the mutation using Guardant360 in a clinical trial, he reached out to Resolution in order to see if their technology could identify the gene fusion.
"We actually saw one of the highest numbers of unique reads over that breakpoint we'd ever seen, with one of the highest allele frequencies," Li said. "Geoff therefore wanted to see if there's a technical difference as to why the two assays [performed] differently in fusion detection."
In the study, the researchers pooled a group of 169 patients with advanced NSCLC from a Dana-Farber database who had previously undergone clinical testing using the Guardant360 assay. Of these patients, 28 had tumor testing positive for ALK, ROS1, or RET rearrangements, and 16 of those had leftover plasma for next-generation sequencing.
Oxnard and his team initially collected plasma from each cancer patient previously tested at an earlier date with Guardant360, performing NGS on the sample using the ctDx-Lung cancer assay. The researchers then uploaded the results to Resolution's cloud-based informatics platform and generated a detailed diagnostic report.
Comparing the results, the team saw that ctDx-Lung identified 13 mutations and tended to report higher allele frequency percentages in fusions than Guardant360. In contrast, they noted that Guardant360 only detected seven cases of gene fusions.
"In every single case that Guardant and [Resolution] both detected the same fusion, we saw it at a higher frequency than they did," Li said.
According to Li, the ctDx-Lung assay requires about four days to produce diagnostic results.
"We attach these little 40 base probes and perform primer extension, where we attach the probes, extend them, and recreate the other half of the molecule," Li explained.
Li also noted that the ctDx assay identifies the breakpoint — where the genes fuse together — and is gene fusion partners agnostic, allowing for novel partner discovery. Because the probes are only 40 base pairs in length, the firm can create custom assays that are straightforward, optimized for fusion detection, and do not require changes in the wet lab protocol.
Li argued that large capture probes or amplicons will not "stick" or capture to the gene fusion fragment as well as shorter probes, reducing detection or the detected allele frequency. However, he acknowledged that users need to be careful with potentially detecting off-target reads when using shorter probes such as the ctDx platform. Li noted that due to its proprietary chemistry, the ctDx platform has upwards of 99 percent on-target rates, allowing sequencing requirements and therefore costs.
Overall, the researchers wrote that they believe that benchmarking plasma genotyping assays should focus on actionable mutations, in addition to using a tumor as a reference standard to establish clinical specificities.
In addition, the study authors noted that bioinformatics differences may affect variant calling. The team found increased reporting of unusual breakpoints by ctDx-Lung compared to the Guardant360. At the same time, repeating an analysis with Guardant using the latest data analysis pipeline also produced two additional fusion calls.
In response to the study, a Guardant spokesperson said in an email that the firm has "repeatedly demonstrated best-in-class liquid biopsy performance as evidenced through over 100 peer-review publications."
The spokesperson pointed to a recent validation study published in October 2018 in JAMA Oncology, where researchers compared tissue and liquid-based testing using the Guardant assay. They saw that plasma sequencing increased targetable mutation detection by about 36 percent, in contrast to using tissue biopsy testing alone.
The Guardant spokesperson argued that "properly designed performance comparisons must include a statistically appropriate number of samples, measure both sensitivity and specificity, [and] are ideally prospective and comprise an unbiased sample set blinded to the participants."
The AACR poster states that the study was blinded to both Resolution and the DFCI.
Li said that Resolution Bio aims to have its ctDx cell-free DNA (cfDNA) platform power most liquid biopsy tests in order to increase access to personalized cancer care for patients.
In addition to lung cancer, Li explained that Res Bio has used the ctDx platform in trials to look at prostate cancer, ovarian cancer, breast cancer, cholangio-sarcomas, and bladder cancers. For bladder cancer, the firm has found that the assay works on urine samples and tissue samples as well.
"One of the differences for us is that our probes are 40 bases long," Li said. "Because we work with drug companies, we don't make the 80 to 90 pan-cancer gene panel, and instead make custom panels and have over 60 custom panels created."
According to Li, Resolution is currently working on a companion diagnostic version of the ctDx cfDNA platform that will be a globally distributed, in vitro diagnostic ready kit. The firm soon plans to submit a pre-market approval application to the US Food and Drug Administration.
"I think [the study] confirms that our strategy of capture is working in this challenging cell-free DNA space for more challenging mutations than SNPs and indels," Li said.