NEW YORK – Following a strategic pivot from infectious disease to oncology, with an associated leadership turnover earlier this year, multiplex PCR firm ChromaCode is beginning to build data for its first cancer product, a panel test for actionable mutations in late-stage non-small cell lung cancer, presenting validation results at the Association for Molecular Pathology Meeting this week.
The company also highlighted future applications, including liquid biopsy-based minimal residual disease testing and cancer monitoring, and a new non-cancer application, kidney transplant rejection testing.
While larger health systems continue to embrace in-house NGS for biomarker testing in solid tumors, smaller community oncology programs — lacking the upfront investment funds, technical expertise, and high-sample volumes necessary to make in-house NGS practically and economically feasible — are limited to sending out to centralized commercial labs.
The issue that emerges with send-out testing is that the turnaround time for receiving results can stretch into multiple weeks, and even months, meaning patients harboring actionable genetic alterations have to wait to act on them much longer than they would with in-house testing.
ChromaCode Chief Business Officer Padma Sundar said in an interview that recent estimates indicate that about 70 percent of the NSCLC genomics market is served by central labs. With its system, ChromaCode aims to provide a platform that will allow smaller hospitals and health systems to bring testing in-house, ensuring more patients get timely results.
According to ChromaCode CEO Mark McDonough, benefits to its approach also include its simplicity. The NSCLC assay doesn't require external bioinformatics, a specialty that smaller community medical groups are unlikely to have access to.
ChromaCode's core technology, called high-definition PCR (HDPCR), uses machine learning algorithms and novel chemistries to increase multiplexing capabilities of standard lab instruments.
Although more easily implementable NGS systems, like Personal Genome Diagnostics' Elio, have begun to enter the market, they are still cost-prohibitive compared to HDPCR. The Qiagen QIAcuity instrument, on which ChromaCode's NSCLC test is designed to run, can be purchased for as little as $65,000, McDonough said.
At the AMP corporate workshop, Bharat Thyagarajan, director of the division of molecular pathology and genomics at the University of Minnesota, shared data from his and his colleagues' validation study of the NSCLC assay, which launched in July.
"The real advantage of this method was the shorter turnaround time," said Thyagarajan, with a total assay turnaround time of around four hours in his team's hands. "That's compared to when we used targeted NGS panels at our institution, which typically took between 36 to 48 hours to get the answers," he said.
ChromaCode's assay requires at least 15 nanograms of DNA and 5 ng of RNA and covers 15 alterations that represent more than 50 percent of all known lung cancer mutations. Although the assay's ability to detect variants isn't 100 percent across all these alterations — struggling, relatively, with fusions and insertions compared to SNVs — Thyagarajan said that based on prevalence data from the COSMIC database, he and his team estimated that for every 1,000 individuals tested, HDPCR would miss only three people with alterations in the targeted regions.
In their validation, the University of Minnesota team started with a 100-sample analysis designed to optimize their software parameters for calling positive versus negative. The cohort was split between 71 known-positive specimens and 29 negative controls.
After this, the team analyzed another 20 known-positive specimens with five negative controls, comparing HDPCR results to NGS calls using a UM in-house LDT that includes 127 genes plus a supplemental fusion test.
Overall, investigators calculated sensitivity of 92 percent and specificity over 99 percent for the HDPCR assay.
ChromaCode also highlighted results at the AMP session from a study it conducted comparing the failure rate of its assay with Illumina's TSO 500 NGS assay performed in an unnamed CLIA lab.
Sundar said in an interview that published data suggests that as many as 22 percent of NSCLC patients are unable to receive a test result because of issues with sample quantity or quality that result from scant samples, often from core needle biopsies that are widely used in this tumor type.
In its study, ChromaCode tested a combination of biological samples and reference samples with varying DNA content. The company found that below 20 nanograms, a majority of samples failed to yield a high enough read depth to pass quality control for the TSO 500 assay. In contrast, even at 10 ng input, HDPCR only had a 10 percent quality not sufficient (QNS) rate.
ChromaCode Director of Genomics Jeff Gole added that for the nearly 90 percent of samples meeting QC requirements for HDPCR in the study, detection accuracy was 100 percent.
McDonough said that ChromaCode currently has several early adopters who have brought the NSCLC assay in-house, and Sundar noted that the company is working to develop health economic data to support payor reimbursement.
"We have developed an economic model, which shows that if you use this assay in the first line, you are finding something at least 50 percent of the time, and it is more economical to reimburse this assay, even at $1,000, and then reflex negatives [cases] to NGS" versus using NGS out of the gate, she said.
One of the firm's first early-access partners, Florida testing firm Protean Biodiagnostics, is preparing to take that data to its local Medicare contractor to try to make a case for reimbursement, Sundar said.
During the same workshop, Jerrod Schwartz, VP of advanced technology at ChromaCode, presented data from experiments using the HDPCR platform to develop patient-specific, tumor-informed PCR assays for tracking minimal residual disease in blood samples, a growing competitive clinical testing market.
"I think we can all appreciate some of the challenges with building PCR panels in oncology," Schwartz said in his presentation. "We all understand that cancer can be extremely heterogeneous and dynamic, while PCR panels are inherently sort of static in their design."
Traditional methods to develop panels can take months, he added, because of the need to iteratively order and test primers and probes to optimize your assay. ChromaCode, he said, has developed a new novel chemistry to "abstract away the need for probes during screening."
The firm has also moved to sequencing as an upfront tool to filter through hundreds of thousands of candidate primers for different targets of interest, which has allowed the company to build out high-plex assays in a matter of weeks, a necessity considering the practical needs on MRD testing, where patients need tests available within weeks of their curative-intent surgery.
Schwartz said that another challenge for PCR in liquid biopsy is performance. Circulating tumor DNA molecules are present at minute fractions amidst a background of normal DNA. As such, a test's detection power for a fixed sample input depends on the number of markers that you can include in an assay.
"As a hypothetical example," Schwartz said, "if you're looking at tumor fraction of one in 100,000, and your assay can only look at four markers, your detection power is just barely above zero. But if you can increase that by a factor of 25 or more, your detection power goes up quite a bit."
The challenge, he said, is how to achieve this with digital PCR where you are limited in terms of the number of color channels on an instrument.
ChromaCode's solution is to implement the methodologies for amplitude modulation and multispectral encoding that it has been innovating. On a four-color channel instrument using two intensity levels, HDPCR can encode up to 80 targets in a single reaction, and using newer instruments that have six color channels, it's possible to get up to over 700 targets encoded in a single reaction, Schwartz said.
The company has thus far been able to build out this chemistry and have it working well on all four leading digital PCR platforms — Thermo Fisher Scientific's Absolute Q, Bio-Rad's QX600, the Roche Digital LightCycler, and Qiagen's QIAcuity.
In dilution experiments with a 32-plex prototype assay, Schwartz said the team is seeing a limit of detection below 0.03 percent.
"Interestingly, even at that low value of 0.03 percent spike in, we're still tenfold above the background noise in terms of the average copy number per target across all four instruments, which suggests there is even more room to push the LOD down further," he added.
Furthermore, given that these PCR instruments leverage multiple wells, Schwartz said it is possible to split a sample across several wells, pushing the number of assayable targets into the hundreds for a four-color system and into the thousands using a six-color-channel instrument.
Schwartz said ChromaCode has started to build out a prototype workflow that starts with targeted deep sequencing of tumor tissue samples and matched normal samples to identify and prioritize variants for building tumor-informed bespoke panels.
"Using either existing primers that we've already built out or new designs [using the firm's probe-less chemistry], we can build tumor-informed digital PCR panels and then deploy them really quickly," he said.
The company is still early in its effort to validate this application, and Schwartz only highlighted data from three late-stage colorectal cancer patients where the ChromaCode results matched up well with sequencing. Future evidence, if the company hopes to market tests for minimal residual disease, will require validation in samples from early-stage patients in the adjuvant setting.
Finally, Gole shared emerging data for a new non-oncology application of HDPCR — in kidney transplant rejection testing.
Like cancer MRD, this is an area that has emerged as a commercial niche with the advent of cell-free DNA technologies, and it is one where McDonough said that the company is confident in the potential of its technology to be "incredibly disruptive."
CfDNA-based rejection testing is already reimbursed by Medicare when used to avoid biopsies, but Gole stressed that this application requires a tight, one- to two-day turnaround time, as well as assays that quantitatively measure donor DNA fractions significantly under 1 percent.
ChromaCode is hoping to break into the market with a decentralized PCR testing kit, differentiated from the lab-based and largely NGS-driven options currently sold.
Gole cited a minimum two- to five-day turnaround time for centralized organ rejection testing, which he said makes it challenging for transplant physicians to make a clinically relevant decision on whether to biopsy or not and can hinder reimbursement.
As an alternative, the company has developed a 48-SNP, eight-well assay that Gole said covers more than 99.9 percent of the population, including all major ethnicities, and can return a test result within 10 hours.
"The first few steps are identical to NGS. You have to draw the blood. You have to separate out the plasma and extract DNA. But for us, once you have that, the PCR is very streamlined. There's an initial pre-amplification followed by digital PCR and automated analysis on our software," Gole said. "If you compare this NGS … it's challenging to get a result within less than 24 hours, even if it is done in-house," he added.
As part of its early development, ChromaCode has already performed initial analytical validation using cell lines with known genotypes for every allele in the assay — simulating donor fractions ranging from zero to 3 percent and confirming by NGS.
Gole reported that for every allele that the team has tested thus far, donor DNA was detected at the lowest possible donor fraction of 0.25 percent.
"We're excited about these results, and we're hoping to start some alpha trials with this assay some time in 2024," Gole said.