NEW YORK – Early cancer detection firm Hawkeye Bio is preparing to take the next steps in the commercialization of its protease-based disease detection platform as a low-cost tool for assessment of individuals with a high risk for lung cancer.
The company, which was founded in 2019, has been working quietly behind the scenes for the last four years to refine and validate its technology and has ongoing collaborations in in China and Europe that CEO and Cofounder André de Fusco said it believes will allow it to establish its kits as in vitro diagnostics and market them as an affordable and easily accessible lung cancer screening option.
Although Hawkeye believes the technology could have potential in other disease areas like neurology, its near-term focus is on establishing this first lung cancer application.
The company recently published the first data describing its technology in detail and calculating the performance of its kits. In the paper, which appeared last month in the Nature journal Communications Medicine, the authors reported that the assay could detect 90 percent of stage I cancers with 82 percent specificity.
Hawkeye's second cofounder and CSO, Paul Dempsey, said that the firm's platform is an evolution of work initiated by Stefan Bossmann at the University of Kansas Medical Center (KUMC).
Bossmann had developed an in vitro platform for protease detection, aiming to be able to achieve high enough sensitivity to detect protease activity in a blood sample.
"There have been pretty nice detectors around for a long time, but Stefan's work as a phytochemist showed that it was possible to build a system that had a background low enough that you could detect the small amount of residual activity that's present in a blood sample," he said.
"We looked at a number of technologies but came back to Stefan, because we thought [his platform] could have the right profile, the right performance, and the right cost target for what we're trying to do," he added.
Hawkeye then refined the platform, most notably incorporating a graphene technology developed by another KUMC researcher. "We spent the time to move his original protease detection technology from an iron oxide backbone onto this graphene, and that was the enabling step," Dempsey said.
Hawkeye is not alone in exploring protease activity for disease detection. The authors of the company's study wrote that these molecules are attractive as activity-based sensors because, as irreversible modifiers, they are highly regulated and have a well-defined role in cancer, with an impact on tumor cell invasion, angiogenesis, epithelial-to-mesenchymal transition, and, ultimately, malignancy.
Other efforts, like that of MIT-spinout Glympse Bio have involved introducing substances into a patient in order to trigger protease activity, with the products (excreted in urine) then indicating this activity.
Glympse has also set its sights on lung cancer, but its most recent work has remained limited to mouse models.
Hawkeye's platform also allows much higher-scale testing at a much lower cost than would be achievable with an in vivo approach, Dempsey argued. "The most expensive part for us is the plastics that we put the enzymes into. The enzymes are coming from the patient. We've put our cost of goods in a position where we can think about testing 19.3 million people, which is the population in this country that should be getting checked for lung cancer every single year."
Hawkeye's lung cancer classifier, which it developed using machine learning to identify targets that differentiate cancer case and control samples, incorporates 18 graphene biosensors, lyophilized into a 384-well format, each of which is tuned to detect one of 18 target proteases. The firm has dubbed the approach Lung Enzyme Activity Profile, or LEAP.
In the firm's study, investigators tested samples from 450 lung cancer cases and at-risk controls, recruited at KUMC, and at two partner sites in Denmark, Marmara University, and Vejle Hospital.
Participants without lung cancer underwent chest CT screening and any necessary follow-up review to confirm negative status. Patients with lung cancer all had their disease pathologically confirmed.
Using an AI algorithm development tool, the group was able to define a cutoff that distinguished cancers from non-cancers with 90 percent overall sensitivity, 90 percent sensitivity for stage I tumors, and 82 percent specificity.
Dempsey said that other early detection and screening assays being developed with genomic and molecular technologies require higher specificity at the expense of sensitivity, most notably, multi-cancer screening tools being advanced by companies like Grail and Exact Sciences.
In the defined setting of high-risk lung cancer screening, the danger of false positives weighs less heavily than the need to capture true positives, something that the current option of low-dose CT screening has struggled to do in the face of low compliance. "We don't want to miss anyone, so we're anxious to exceed the published performance of low CT itself, which we've done in this study, and with the version that we're using in the lab," Dempsey said.
Estimating that the LEAP blood test could achieve 75 percent compliance, compared to less than 4 percent compliance observed currently, the study authors calculated that it could increase lung cancer detection 17-fold.
If the company were to pursue other applications in the future, it could tune assays more toward sensitivity, depending on the clinical need, Dempsey said.
De Fusco said that LEAP is meant to be a distributed test and is currently being manufactured in California. With this business model, the company is relying on its partners to help it generate additional validation data to support regulatory processes for clinical approval of its kits in various areas of the globe.
"We have four or five partners around the world that want to do longer-term prospective studies, and they're also interested in some things that are a little bit different or adjacent to what we've already looked at," de Fusco said. For example, some groups are interested in vapers and younger people. And some are interested in female nonsmokers, especially in Asia.
De Fusco said that the firm's platform uses standard fluorescent technology from Thermo Fisher Scientific, whose VarioSkan Lux plate reader was used in the recent study.
"We can train people on this assay in half a day, or really two hours … so when we go and install this, and this gets adopted in the cancer care cycle, it's not a few million dollars' worth of sequencing equipment. You don't need Ph.D.s."
Although labs could implement the platform in their own laboratory-developed testing, and some unnamed partners already have, de Fusco said that Hawkeye intends to achieve IVD clearance for its kit not just in Asia, but also in Europe and in the US, by submitting it for US Food and Drug Administration review.
The timeline for these efforts could be variable he said, depending on the company's partnerships and the data generated from their ongoing studies.
One boon, de Fusco said, is that the test is manufactured in the US by an undisclosed company that has worked with the FDA before.
"[FDA] is still going to have to be convinced that proteases are indeed reliable, and why we picked these 18, but there is 15 years of background on that, and that's what I would consider a valid request in a constructive conversation," he added. "It's a conversation we're actually looking forward to having with the FDA, I don't know how many people at FDA are so well versed on proteases."
Outside of lung cancer, de Fusco also suggested that Hawkeye's technology could turn out to be additive or complementary to the multiomic sequencing tests now being pursued for multi-cancer early detection.
A previous version of the technology that included 14 cancers had been tested in 15,000 people in China in work led by Bossmann at KUMC that has yet to be published. De Fusco said Hawkeye now aims to expand that to a larger cohort of 50,000 individuals.
Whether combination efforts actually happen may depend on the larger players in the field. "Does [our platform] belong with other tests? We think so," de Fusco said.
"Could we hop on these large studies? … We could, and we should," he added. "But it's a little bit atypical for some of these larger companies to partner with a small player because they've spent many, many years and a lot of money to support a certain approach, and it's tough to tell your shareholders, 'Hey, here's somebody who's got a solution for 15 bucks, and we're gonna bolt it on.'"