NEW YORK –Cancer minimal residual disease testing firm Haystack Oncology is on track to launch a tumor-informed, patient-specific assay platform next year, supported by an infusion of $56 million that the company announced this week.
Founded in early 2021, Haystack is advancing an MRD methodology developed by Johns Hopkins researchers, led by cancer genomics pioneer Bert Vogelstein. The company's personalized assay approach follows the model of several other companies that have launched MRD testing services in recent years, including Natera, Inivata (now part of NeoGenomics), and, most recently, Personalis.
But Haystack believes its technology can offer significant improvements in sensitivity and accuracy compared to existing assays.
"There's been a lot of thinking and a lot of development on the technology side in order to ensure a level of rigor going into the testing that we're bringing forward," said CEO Dan Edelstein, adding that the decision to form the company was a "natural evolution" of technology development and clinical need.
Haystack is keeping details of its methods to itself, but Edelstein highlighted error correction as a key aspect of the company's approach and said that its claimed sensitivity is enabled by methods that expand upon Vogelstein and colleagues' Safe-SeqS.
"We have a number of workflow steps that are focused on ensuring that we analyze the maximum number of DNA molecules that are derived from a patient's specimen … as well as error-corrected sequencing and other techniques that reduce the background noise in the system and really pull up the signal," he said.
On its website, Haystack claims its sequencing has an error rate that is more than 100 times lower than other MRD assays on the market, which purportedly allows it to set a lower cutoff for MRD detection than competitors. This, in turn, leads to a lower rate of false negatives.
The company hasn't yet published a validation study, but it describes its "Haystack Duo" technology as achieving sensitivity down to 1 part per million or 0.0001 percent tumor fraction with 100 percent specificity.
Reference to minute limits of detection such as these have sometimes been criticized in the research community because at frequencies this low, probability would imply that a standard blood sample would be unlikely to contain even one copy of the mutant DNA molecule in question, rendering such high sensitivity clinically irrelevant.
But Edelstein said that targeting multiple mutations nullifies this issue to some extent. Although a standard sample may fail to capture a single ultra-low frequency mutation, it would be much more likely to contain one of a group of 40 or 50 target DNA fragments.
"If we track X number of variants in a patient, we have X number of shots on goal to detect a circulating tumor DNA molecule. There is stochastics involved with low level samples, but what we've taken into account here is tracking an appropriate number of variants with an appropriate level of sensitivity and specificity in order to deliver [optimal sensitivity and specificity]," he added.
Existing commercial MRD platforms demonstrate a diversity of approaches to assay size. Natera, which was first to the clinical market, uses up to 16 targets gleaned from patients' tumor tissue sequencing data.
NeoGenomics' Radar assay targets up to 48 tumor tissue-derived alterations. Invitae's Personalized Cancer Monitoring test, developed originally by ArcherDx, aims for at least 18 and up to 50 variants. Personalis, which launched its Next Personal assay last year, aims to include up to 1,000 targets by starting with whole-genome rather than exome-only tumor sequencing.
How these different strategies affect clinical utility is yet to be determined, but some early research in lung cancer found that tracking larger numbers of variants improved MRD sensitivity, but at the cost of a drop in specificity.
Edelstein said that Haystack's assays will aim for between 30 and 50 alterations, depending on how many targets are identified in patients' tissue exome sequencing.
"It's a little bit dependent on what makes sense in that particular tumor type, but in colorectal cancer, we've seen data from the DYNAMIC study where, using a very similar test method [and] following just one to two mutations … generated very reasonable, very good clinical outcome data. So now, if we can multiply that by a factor of X, we can push that even higher," Edelstein said.
In DYNAMIC, a study of ctDNA testing in stage II colon cancer, results of which were published in the New England Journal of Medicine this year, Vogelstein and colleagues employed their Safe-seqS error-corrected sequencing method, which makes up part of the backbone of the testing Haystack is now commercializing.
With a goal of launching clinical testing by the second half of 2023, potentially in colorectal cancer and/or other tumor type, Edelstein said Haystack is working on CLIA validation of its test, as well as conducting a few other projects in "research-use-only mode."
"We're aiming to have more clinical evidence published throughout next year across different tumor types," he said.
Toward distinguishing itself from clinical competitors, Edelstein said Haystack is focusing on areas where other MRD tests have not sought to —or potentially not been able to — demonstrate value. "When we apply the system and the test modality to patient groups or clinical decision points where thus far the clinical data has been somewhat lacking ... we hope to kind of pave a path into tumor types and [indications] that have historically been more challenging to detect circulating tumor DNA," he said.
Natera's Signatera test and Guardant Health's Guardant Reveal assay (which uses a non-patient-specific fixed panel approach) are both covered by Medicare under contractor Palmetto GBA's foundational local coverage determination. NeoGenomics has been attempting to join them but was recently asked by the MolDx program, which conducts technical assessments of assays seeking coverage under the LCD, to provide additional data.
Edelstein said Haystack is exploring applying to MolDx as well as other potential Medicare coverage pathways.