BALTIMORE – Illumina this week announced the release of a new version of its TruSight Oncology 500 assay, designed for use with circulating tumor DNA in liquid biopsy testing. Two researchers who have had early access to the new panel shared their experience in a sponsored session at the annual meeting of the Association for Molecular Pathology on Thursday.
The TSO 500 ctDNA assay, which Illumina said is optimized for use on the NovaSeq 6000, will become available to regular customers early next year alongside an updated, "high-throughput" version of the tissue-based TSO 500 panel, which is currently optimized for the NextSeq 550 and 550Dx systems.
The TSO 500 ctDNA panel covers all exons of 523 genes in key cancer-related pathways, allowing customers to determine single-nucleotide variants, indels, copy-number variants, and fusions, and to assess microsatellite instability and tumor mutational burden.
It will compete with other blood-based sequencing assays on the market, some similarly covering hundreds of genes, though many of which are operated out of single centralized labs. Guardant Health, for example, has launched a 500-gene assay, which it offers as a send-out research-use-only service, and Roche's Foundation Medicine offers testing services using FoundationOne Liquid, which covers 70 genes. Personal Genome Diagnostics has developed 500-plus-gene assays for liquid biopsies with plans to advance FDA-approved kit versions, though these are not yet available.
Roche and Thermo Fisher Scientific do currently offer NGS kits for ctDNA, but those are limited to about 170, and 60 genes, respectively.
Although Illumina's new TSO 500 ctDNA panel is for research use only, reports from early users at AMP illustrate that the kit has garnered significant interest by clinical research programs and could eventually be embraced by the laboratory-developed test community for use in oncology.
One early adopter is Frederick National Laboratory, which announced its intention to validate the technology in January. FNL, which is managed by private contractor Leidos Biomedical Research, said at the time that its Molecular Characterization Laboratory (MoCHA) was committed to using the assay to support various National Cancer Institute precision oncology trials, assessing liquid biopsies in up to 7,000 subject samples.
At AMP, Biswajit Das, MoCHA's principal scientist, described an initial feasibility study that the lab has conducted as part of its planned validation and provided additional details about how FNL intends to adopt the TSO 500 ctDNA panel.
According to Das, as part of the NCI's Experimental Therapeutics Clinical Trials Network, FNL performs both retrospective NGS analyses, which are not reported to clinicians, and prospective testing that is returned for use in clinical trials like NCI-MATCH. For the former, the lab does exome and RNA sequencing, and for the latter, it has up until now been using the Thermo Fisher Scientific Oncomine pan-cancer assay for cancer tissue sequencing.
The goal of validating the Illumina TSO 500 liquid biopsy panel is primarily to incorporate it into prospective, trial-supportive testing. Das said this will include the existing MATCH trial and IMPACT studies, as well as a set of upcoming new trials called AML-MATCH, COMBO-MATCH, and Immuno-MATCH, which "will be coming up in the next few years."
In feasibility testing, FNL found the assay to have excellent specificity and high sensitivity within the reporting thresholds for SNVs, indels, CNV, and translocations that it was expecting. The panel even allowed for a lower allele frequency threshold than what Illumina recommends for the product, Das said, going down to 0.25 percent mutant AF for SNVs while still maintaining good specificity.
As an example of the sensitivity the group is seeing with the panel, Das shared some data from a comparison, in which six patients were profiled using the TSO 500 kit and BEAMing, a digital PCR technology. Using the Illumina panel, "we could identify T790M mutations [called by PCR] even at a very low allele frequency, as low as .04 percent AF," he said.
Das and his colleagues did observe three false-positive SNVs in their assessment of 119 normal control samples, but no false-positive CNVs, translocations, or indels.
However, these may actually represent clonal hematopoiesis, which can be filtered out by performing parallel sequencing of DNA from white blood cells — something the FNL team has adopted — or increasingly, using bioinformatic tools.
Clinical trials services firm Q2 Solutions also discussed its analytic validation of TSO 500 ctDNA at the AMP meeting. A joint venture between Quintiles and Quest Diagnostics, Q2 has been operating since 2015 and inked an agreement with Illumina in 2016, allowing it to develop companion diagnostics that would run on Illumina's clinical sequencers.
Q2 staff scientist Stephanie Hastings said her firm is taking a phased approach to validation, initially testing the TSO 500 panel for detection of SNVs, indels, and TMB in circulating tumor DNA.
In collaboration with pharmaceutical partner Bristol Myers Squibb, the company performed a series of experiments with the early-access ctDNA panel, using both patient plasma samples and contrived samples to test sensitivity and specificity.
Overall, Hastings and her colleagues analyzed over 400 diseased plasma samples, and 25 self-proclaimed normals.
Only about 80 percent of these samples yielded over 20 ng of DNA, she noted, which became a driving factor in the lab's desire to try to test the TSO panel below its recommended 30 ng input, going down to 10 ng in the validation.
In addition to human samples, the Q2 team tested four types of contrived samples, using materials from NIST, reference samples from both SeraCare and Horizon Discovery, and a newer product developed by Agilent, which Hastings said her team found "particularly valuable" in providing "hundreds of characterized low-frequency mutations in the sub-1 percent allele frequency range."
Q2 wanted to be able to detect mutations across the panel in a range between 0.2 and 0.5 percent AF. "We found that we needed about 1,300x median coverage as well as at least 80 percent of bases covered at 1,000x. And what this allows us is to ensure that we have at least one megabase covered at 1,000x, which is the minimum requirement for TMB calling," Hastings said.
Under these criteria, she added, "over 95 percent of the samples with at least 20 ng of material achieved the [desired] metrics."
Using the Horizon, SeraCare, and Agilent contrived samples, the group performed a variety of reproducibility experiments, both within and across runs, operators, and reagent lots, "showing very good correlation between replicates, even down well below 0.5 percent," Hastings said.
Examining analytical sensitivity using their collection of different commercial reference samples, she and her colleagues calculated that at a LOD of 0.5 percent, they could maintain 99 percent sensitivity. For even lower levels, between 0.5 and 0.25 percent, the team still had "quite good sensitivity" at about 80 percent.
For specificity, the lab again turned to contrived reference samples with verified true negative positions, but also tested self-proclaimed or verified normal samples from human plasma.
In reference materials, Hastings said, one issue is that some positions are not as well characterized. This can make it hard to cement a true-negative rate for an assay because it requires you to have a solid picture of all variants that are present or not present.
Regardless, she said, the team achieved "very, very high specificity" with the TSO 500 assay. "We feel very confident that when we are calling a variant, it is in fact present."
Like FNL's Das, Hastings also raised the issue of clonal hematopoiesis in assessing specificity in clinical "normal" samples. For their validation, the Q2 team analyzed plasma from 25 of these individuals and did call some alterations, which they plan to study further as possible clonal hematopoiesis cases.
In a second validation phase that the company is working on now, Hastings and colleagues are expanding to the detection of copy number alterations and the assessment of microsatellite instability from ctDNA.