NEW YORK – Multi-cancer screening startup Harbinger Health said this week that it is maintaining an aggressive timeline to a commercial launch, with recent data presentations offering the first insight into the sensitivity and reproducibility of the firm's test.
When it first announced its intentions in the space, Harbinger had yet to publish on its approach or share any performance data, describing its methodology as methylation-based but providing few details. But in recent months, the company has begun to release at professional meetings results from pilot studies, which thus far, indicate that its test performance is competitive with other players in the emerging multi-cancer early detection (MCED) space, namely Grail, which launched its Galleri assay as a laboratory-developed test in 2021.
Harbinger CEO Stephen Hahn said this week that the company currently expects to bring its own LDT to market in early 2025. Like Grail, the firm intends to seek US Food and Drug Administration approval but believes that by offering this test to patients in the interim it can collect important real-world data, particularly in diverse populations, that will produce an even better measure of test performance to provide to regulators.
According to Hahn, Harbinger's 10,000-person prospective study with the Sarah Cannon Research Institute is proceeding apace, and the company has been "working hard to make sure that we have participation from diverse patient groups because we realize that cancers may be different in different ethnic and racial groups." It expects the trial to be fully enrolled by the end of this year or early in 2024.
In the meantime, a series of meeting presentations have offered the field a first glimpse into Harbinger's unique methylation-based methodology and its early case-control performance data.
At the annual meeting of the American Society of Clinical Oncology in June, the company reported on a study with the University of California, Davis and the Dana-Farber Cancer Institute that used the Harbinger assay to test samples from 1,046 subjects — 621 patients with newly diagnosed, treatment-naive cancer across 15 tumor types and 425 individuals with no history, diagnosis, or symptoms of cancer.
The researchers reported an overall cancer detection sensitivity of 82 percent at 95 percent specificity. And, crucially, sensitivity did not drop as precipitously for early-stage tumors as it has for some competing MCED approaches: For stage I cancers, the sensitivity was 74 percent, while for stage II cases it was 84 percent.
Like its competitors in the field, Harbinger has also trained its assay to predict where in the body a cancer has occurred, something physicians and other opinion leaders have stressed will be crucial if blood-based multi-cancer screening is to be employed on a population scale.
But unlike some approaches being pursued by Grail and others, which involved training assays using broad, largely biology-agnostic surveys of epigenetic changes and other aspects of cell-free DNA, Harbinger's platform is built upon specific methylation changes that are associated with the earliest aspects of cancer formation.
"We're very much focused on a couple of issues, and these data support that focus," Hahn said. "The first of those is making sure that you have robust sensitivity for detecting early-stage cancer."
Stage III and stage IV cancers aren't particularly hard to detect, he added, and they don't provide the same possibility of intervention as incipient cancers. As such, early-stage detection is really where there is an unmet need.
Hahn described the underlying biological basis of Harbinger's platform as a signature that "appears to occur very early in the natural history of a cancer," when there are only a few cancer cells.
Although Harbinger hasn't stated anything regarding a price target for its planned LDT, as a targeted methylation panel, the company's test likely costs less to perform than assays that require more extensive DNA sequencing.
Beyond the ASCO performance data, two newer presentations at the American Association for Clinical Chemistry meeting last month added evidence for the test's robustness and for its ability to be used longitudinally to monitor cancer progression.
"Cancer screening isn't only one point in time. For example, if a woman were to go for a mammogram, there might be something that they see, but [they] say, 'Come back in six months, and we'll take another a look at it.' I think the data we showed supports that we are going to be able to do that [in the blood]," Hahn said.
The community should expect to see full publications from some of these early studies within the next six to 12 months, he added.
Although Harbinger's early-stage sensitivity numbers in the ASCO cohort were notable, Hahn said that the company isn't resting on its laurels and is working hard to see if it can improve even further.
"Of course, the data are going to be what the data are, and we have to follow the truth in all of our clinical studies. But to the extent that we can optimize what we're doing, and we think we can, then [we want to do that]," he said.
Improvements could also come from sources beyond the epigenome, according to Hahn. Although he and his team have a "strong belief that methylation and methylation patterns are fundamental to disease," it's "completely rational and, I think, probably accurate to say that orthogonal approaches will improve the performance," he said.