This story has been updated from a version posted Feb. 9, to include additional expert commentary.
NEW YORK – Canadian liquid biopsy startup mDetect is launching a clinical trial of its DNA methylation pattern-based breast cancer test, following a C$900,000 (US$669,510) grant from the Canadian Institute of Health Research.
The mDetect breast cancer test is a targeted sequencing assay of 90 breast cancer-related methylated circulating tumor DNA (ctDNA) targets. It is currently being tested for clinical utility among metastatic breast cancer patients undergoing treatment as a way to monitor their responses to therapy in real time.
"There is an urgent clinical need, especially in the metastatic setting," said Irsa Shoukat, mDetect's cofounder and development officer.
The response rate to therapy in metastatic cancer settings varies depending on cancer subtype and choice of first-line therapy. In the case of triple-negative breast cancer (TNBC), in which mDetect's first tested its assay, some estimates put the response rate between 25 and 30 percent, where the remaining 70 percent of patients will have waited three months for a CT scan to know whether their therapy was effective or not.
The breast cancer test is the first of what the Kingston, Ontario-based company hopes to be several cancer offerings.
The startup, which incorporated in March of last year, chose to focus on methylation patterns as a biomarker largely because they provide a broader diagnostic target, as they occur frequently and consistently within cancer subtypes. Furthermore, some research suggests that methylation signatures may prove better targets for cancers regardless of their mutational profiles, such as TNBC, which has few consistent mutations present.
The mDetect method was originally published in Nature Precision Oncology in 2021. That proof-of-principle study evaluated the test's performance only in TNBC, but Shoukat says that the company has since evaluated it in other breast cancers, all of which can be included in a single assay.
The assay was designed by mining the TCGA database for breast tumor-specific hypermethylated target regions found in at least half of the patient population for a given subtype and that are not methylated in normal tissues or blood samples.
Shoukat said that this strategy enabled mDetect to focus on a limited number of targets, which reduces overall costs relative to whole-genome approaches.
Ben Berman, a professor of computational biology specializing in DNA methylation at the Hebrew University of Jerusalem, noted the likely low cost of mDetect's test as an advantage, while commenting that since the company's 2021 study, more whole-genome maps of both cancer and normal cell types have emerged, providing mDetect and others with a greater amount of data to incorporate into commercial platforms.
"The problem with TCGA is that it was based on the 450k array, which was very limited in terms of content," he said.
The mDetect workflow involves isolating ctDNA from plasma or serum samples followed by multiplexed PCR to amplify targeted regions of methylation, which are then sequenced to quantitate the methylation status, Shoukat said. Although mDetect is currently sequencing samples via the Ion Torrent, she said that the test is agnostic to the sequencing platform used.
Quentin Gouil, a senior research officer who studies epigenetics at Australia's Walter and Eliza Hall Institute of Medical Research, commented that the mDetect method appeared robust at assessing a "good number of regions" and sensitive to low tumor loads, and that the assay provides a conveniently simple readout of the number of methylated reads.
He cautioned, however, that mDetect's core methodology, PCR of bisulfite-converted DNA, is "always a pain," and prone to artifacts and variable coverage.
Shoukat explained that in developing the company's single-tube combined breast cancer assay, mDetect found ways to improve coverage and make it more uniform, through methods to balance and normalize target regions.
The company is currently wrapping up the technical validation of the combined breast cancer test and expects to publish the results soon.
The upcoming trial of the assay's clinical utility is expected to launch in April and aims to recruit up to 150 participants, whose responses to treatment will be monitored for up to three years. The trial will take place at two sites, in Kingston and Ottawa.
Shoukat said that the firm's commercialization plans for the breast cancer test are currently "fluid," but that mDetect is leaning toward pursuing a path toward a lab-developed test. Nonetheless, the company is designing its trial such that it will fulfill US Food and Drug Administration requirements for either 510(k) or PMA application.
"Our trial is designed in such a way that we're not excluding things that the FDA is going to be looking for, for clearance," Shoukat said.
Anticipating the launch of an LDT, mDetect is currently looking into partnerships with CLIA-certified labs in both Canada and the US.
In the meantime, mDetect also has tests for prostate, ovarian, pancreatic, and lung cancers in the pipeline, as well as for the rarer uveal melanoma.
These tests, said Shoukat, have undergone technical validation, and the company is beginning to conduct tests using preserved patient samples.
In all cases, Shoukat said, "our main setting is metastatic patients, but we are going to be exploring different settings, [such as] minimal residual disease."
The field of cancer detection and monitoring via methylation profiles is small but growing, with entrants at all ends of the company-size spectrum, from mDetect, with two full-time employees, to Grail, with approximately 1,300. Competitors are also targeting different methylation sequencing applications, ranging from therapy response and disease monitoring to early detection and minimal residual disease (MRD) monitoring.
Israeli startup JaxBio Technologies, for example, is building microarray-based diagnostics that include DNA methylation detection, and used to inform cancer type, subtype, stage, and response to therapy. The company's methylation sequencing panel involves chemoenzymatically labeling DNA methylation patterns coupled to computational deep learning tools to achieve single-molecule epigenetic profiling.
Arrays and PCR-based tests both offer targeted means for rapidly assaying defined biomarkers, but differ in throughput and cost. Arrays offer a much higher throughput than PCR-based tools, while PCR is potentially cheaper per marker. JaxBio has estimated a cost of approximately $30 per analysis, while it remains too early to speculate on that for mDetect.
Hebrew University's Berman noted, however, that methylation-biased PCR techniques faced some scrutiny in the field.
"A methylation-biased PCR approach may not be as quantitative as methylation-neutral approaches such as methylation-neutral PCR, or other methylation-neutral approaches such as hybrid capture, cfMethyl-Seq, or whole-genome sequencing," he said. "In methylation-neutral approaches, the fraction of reads in each methylation state is a direct sampling of the fraction of cells the DNA was derived from. In a methylation-biased approach, this quantitative relationship is lost and you get more of an all-or-none signal."
UCLA spinout Early Diagnostics, which pioneered cfMethyl-Seq, is working toward offering it as an LDT for low-cost genome-wide cell-free DNA methylation profiling. This library preparation method involves blocking both ends of all cfDNA fragments in a sample to prevent enzymatic digestion and avoid the occurrence of multiple fragments with the same start and end sites. Sequence analysis is then conducted via a machine learning bioinformatics platform.
The high sensitivity and specificity achieved this way show potential for early cancer detection and MRD, while the diversity of methylated sequences it captures also provides information on tissue-of-origin.
Jasmine Zhou, founder and CEO of Early Diagnostics, commented that it might be challenging for small, targeted panels to detect early-stage cancers.
"For detecting early-stage cancers, very few tumor DNA fragments are available in a blood tube," she said. "A small, targeted panel is set to capture only a tiny proportion of those fragments, likely incurring false negatives."
Zhou also suggested that methylome assays like cfMethyl-Seq may provide greater flexibility in designing pan-cancer detection assays, moving toward a more "all-in-one" cancer diagnostic.
Gouil of Australia's Hall Institute concurred with the last point, saying that the targeted design of mDetect meant that "you definitely miss out on a lot of potentially relevant information."
In response to this point, Shoukat commented that methylation patterns have proven to be very robust and durable, with minimal changes to the profiles established in the primary cancer seen in subsequent metastasis.
"This provides reassurance that our methylation signatures persist in all tumor populations, allowing for consistent detection of all cancers," she said.
Gouil noted that the targeted assay design makes it efficient and effective in achieving its stated goal of monitoring disease progression.
"It's good for a breast cancer test," he said, "much better than single-locus tests."
Shoukat also pointed to mDetect's lightweight targeted assay design and lack of reliance on AI-assisted analysis as advantages in marketing the company's assays as economical and simple diagnostic options.
Methylation-based assays used for early disease screening, Shoukat said, tend to have relatively large data requirements in order to generate strong conclusions. In contrast, monitoring established cancer cases can deliver immediate value to those patients.
mDetect hopes to raise a Series A round in the coming year, which will help advance the company's clinical trials, business development plans, and the implementation of the mDetect assay.