NEW YORK – Research from Strata Oncology on a large cohort of cancer patients has shown that the firm's amplification-based sequencing assay can handle samples that don't meet the usual requirements for comprehensive genomic profiling.
The company also believes that assays like StrataNGS that only need a little tumor tissue could help overcome roadblocks to broad and equitable implementation of precision oncology.
Published in JCO Precision Oncology last month, the company's study analyzed more than 32,000 consecutive tumor tissue samples received for StrataNGS testing as part of the Strata Trial.
StrataNGS is a 429-gene assay performed on RNA and DNA from the same sample. It reports out single nucleotide variants, indels, copy number alteration, microsatellite instability status, gene fusions, and tumor mutational burden and was validated in another recent study in the Journal of Molecular Diagnostics.
"When we came together to start Strata, we were really focused on a core belief that every patient with advanced cancer deserves to know if their tumor has an actionable alteration," said Scott Tomlins, Strata's chief medical officer.
At that time, neither Medicare nor private insurers covered tests like Strata's. This has now changed, but even so, "there are still many patients that are eligible that aren't getting it today," he said.
The company's strategy has been to partner with healthcare systems and pharma firms to provide comprehensive genomic profiling, or CGP, "at scale," Tomlins said. "It doesn't work to try to just go convince every doctor that they should do this [and] you also have to efficiently utilize the data that you're getting back, learn from that data, and bring novel precision medicine trials [that you can deliver] locally."
This was the genesis of the Strata trial: an observational study of the company's commercially tested patient cohort, tracking the impact of testing on treatment selection, clinical trial enrollment, and outcomes.
"It wasn't prespecified, but this question of sample availability and the characteristics of the sample — whether that is part of the reason that people aren't getting CGP — is actually challenging to establish, so we thought that we could use the Strata trial to try," Tomlins said.
In their JCO study, the researchers analyzed the records of about 32,000 samples received consecutively for clinical sequencing and reporting during the trial.
For 94 percent of the 31,165 tested specimens, Strata was able to successfully generate sequence data and report results. This was despite the fact that about 11 percent of submitted samples had tumor content of less than 20 percent.
Meanwhile, 58 percent of samples were classified as "small," having less than 25 mm2 of tumor surface area, which the authors cited as the minimum requirement for some of the leading commercial hybrid capture-based CGP tests, Strata's competitors, in the growing precision oncology market.
Clinical researchers have identified multiple factors that reduce advanced cancer patients' access to genomic testing, including physician skepticism, financial or payor challenges, and disparities in healthcare, in addition to turnaround time and sample adequacy issues. But according to Tomlins, tissue availability is one of the largest obstacles to widespread use of CGP.
"When you talk about how well your test works, it matters what samples were sent to you. We set the limit of 2 mm2 … but we received a lot that were smaller than that, which reflects the fact that many samples that are available in real-world healthcare systems are small and challenging to test," he said.
"A laboratory may say its tests work 99 percent of the time, but you don't know what was sent to them. You also don't know what wasn't sent to them," Tomlins added. "In my experience, clinicians quickly realized that if they sent five samples to a laboratory and they get five not-sufficient results back, they will stop sending those kinds of samples, so it may look like the laboratory has a very high success rate, but it's just they're not getting the difficult samples anymore."
According to Tomlins, Strata's ability to glean results from tiny samples is reflective of both its test technology and its streamlining of bioinformatic and other processes. As described in the JMD paper, the StrataNGS assay uses multiplexed PCR-based library preparation and Ion Torrent sequencing, a choice that he said reflects his and Strata CEO Dan Rhodes' prior experience with the platform.
"We were comfortable with this technology and knew its ability to generate results from the most challenging samples. But then we also really had to optimize every single step … from receiving the sample, evaluating it under the microscope, isolating DNA and RNA, and preparing the library, using a fairly complex, multi-panel approach that we designed to really be able to get the absolute most impactful information out of even the most challenging samples."
The company also developed and optimized every step of the bioinformatics process in house, he added. Testing samples below the company's stated minimum tumor content criteria requires robust processes and "back-end review," he said, to separate signal from noise and determine if the sample did not work.
Mark Burkard, a medical oncologist in the University of Wisconsin health system, said in an email that while the issue of sample availability varies by cancer type and where a biopsy was taken, it is well recognized.
He said that UW Health currently has a goal of routine comprehensive NGS testing for all advanced cancer patients with solid tumors, a program that grew as physicians saw its positive impact in the fraction of patients who benefitted from targeted and investigational therapies.
"I would estimate that a quarter to a third of samples may be inadequate based on the requirement for standard NGS tests," Burkard wrote. That doesn't mean a patient cannot have their tumor sequenced though, he added. "If we determine they are inadequate upfront, we notify the physician, repeat the biopsy, or get a liquid biopsy test."
But even though these fallbacks are in place, the fact of a small or otherwise challenging sample is still an issue, Burkard said.
A number of patients travel long distances for their care and are reached through outreach sites, he said. Most have archived diagnostic biopsies available, which were often obtained with procedures such as lung biopsy or bronchoscopy that carry risks from sedation or pneumothorax. "If we can use their initial biopsy, it saves them the difficulty of time, travel, and risk of procedure," he said.
Although he could not cite specific cases, Burkard said he and his colleagues have certainly seen a number of patients with adequate biopsies for StrataNGS that they believe would have been inadequate for other technologies.
The authors of the JCO study cautioned that they didn't make any direct comparisons with other available CGP tests, so any implication that Strata's lower sample requirements lead to improved results is hypothetical.
In practice, while other tissue NGS tests list higher tumor requirements, companies have strategies they can employ to avoid test failures if customers submit a sub-optimal sample. For example, Caris Life Sciences lists a tumor content requirement of 25 mm2 surface area and at least 20 percent tumor nuclei. But it says that, when necessary, it can microdissect specimens with less than 20 percent cancer cells to enable higher-quality testing.
Sequencing firms can also do what Strata did in the trial — attempt to sequence suboptimal samples and use bioinformatic methods to determine if the results are reliable.
"I think a lot of times, knowing what somebody says their minimum requirements are … doesn't really mean anything until you see what their experience is," Tomlins said. "So with this study, I think we wanted to sort of say, 'here's our experience,' and I think that establishes a framework for other laboratories."
"I think the key thing that we saw is that if you have requirements of 25 mm2, that the majority of samples in the real world don't meet that requirement," he said, "which makes sense when you see clinical trials where you have to account for every sample sent in and [success rates] look so dismal, with NGS failing 30 percent or 50 percent of the time."