NEW YORK – With the validation and commercialization of its blood-based colorectal cancer screening tests underway, multiomics firm Freenome is pushing its core platform into other areas of the cancer care continuum.
At the annual meeting of the American Association for Cancer Research in April, the company presented results from a study demonstrating that its ctDNA sequencing and informatics pipeline could identify predictive signatures of immunotherapy response that appear consistently across melanoma, lung, and kidney cancers and have the potential for use in patient stratification and monitoring.
The new data adds to the firm's earlier forays into advanced early and advanced cancer biomarker discovery, including a study presented at AACR in 2020 exploring whether pre-surgery cell-free DNA and circulating proteins could predict progression in non-small cell lung cancer patients after surgery, and an abstract at last year's American Society of Clinical Oncology annual meeting searching for biomarkers associated with response to the cancer immunotherapy nivolumab (Bristol-Myers Squibb's Opdivo).
Jimmy Lin, Freenome's chief scientific officer, said in an interview this week that while the company's first emphasis remains early cancer detection and screening, "there has been a lot of interest from pharma and biotech in using our overall multiomics platform for other purposes."
The recent AACR study was exploratory and performed in a small cohort: 21 patients with kidney cancer, 14 with melanoma, and 91 with lung cancer. As such, Lin said it should be viewed as a proof of concept that will require further validation in larger cohorts.
In the meantime, he said the data mainly paint a picture of how Freenome's cell-free DNA sequencing and artificial intelligence platform can glean new biomarkers and drive the development of novel clinical diagnostics in areas of great interest like immuno-oncology.
"What we showed is that we are able to start with whole-genome sequencing of cell-free DNA and actually make estimations of gene activation [to] then look at potential signatures," he said.
This methodology expands on technology Freenome has been developing for some time, with company researchers reporting in 2019 on a strategy to model gene expression based on cell-free DNA fragmentation patterns — gleaning the transcriptional state of the cells that are releasing their DNA into the bloodstream at any one time.
At that time, the investigators were interested in the potential to use this to aid early cancer detection. But one of the benefits of the company's platform, Lin said this week, is that it can analyze not only cancer signals but the full landscape of cell-free DNA, both tumor- and non-tumor-derived.
Although the currently available clinical tests for immunotherapy response prediction focus on signals associated with a tumor itself, the precision oncology research community has long hypothesized that because immunotherapy response depends on a complex interaction between a tumor and the immune system, combined approaches that measure both tumor-associated and innate immune factors would be needed to predict patient response with the highest possible accuracy.
Immune checkpoint inhibition has improved clinical outcomes for multiple cancers, including the three tumor types Freenome focused on, but despite this success, significant numbers of patients fail to respond robustly or at all. Meanwhile, although current biomarkers like PD-L1 expression and tumor mutational burden are predictive enough to meet the threshold for clinical adoption, they only explain some of this divergence in response.
"Because we're using whole-genome sequencing of cell-free DNA, we can actually do almost an estimated sort of expression analysis … by looking at what parts of genes are actually in the blood and which ones are protected and thus are not being expressed. Then we can look at these … potential immune or stromal signals that are related to immune checkpoint inhibition response," Lin said, describing Freenome's methodology.
In the study, the company identified 13 transcription factors and 269 genes, which they linked to ICI treatment resistance and response that were present across the 126-patient cohort. These signals indicated a potential pathway of resistance through JAK/STAT signaling and a possible signature of response linked to epithelial-mesenchymal transition.
"This was an exploratory study obviously, so the next step is to replicate that in a larger cohort and to do a prospective study to be able to validate these results," Lin said.
Realistically, developing such diagnostics would probably have to wait for the completion of Freenome's efforts in early colorectal cancer detection, where the company is racing several competitors, including Guardant Health and Exact Sciences.
According to Lin, the firm's prospective PREEMPT trial, which is intended to support a bid for US Food and Drug Administration approval, is going well. "We're still recruiting, but internally of course, we are also scaling up our efforts to be ready for not only reading out [the] trial but for commercial rollout of the test," he said.
The company shared some data from its AI-EMERGE trial earlier this year at the ASCO-GI meeting, showing that its blood-based test could detect not only early-stage colon tumors but also the high-risk precancerous lesions that are found and removed during colonoscopy.
Investigators analyzed blood samples from colonoscopy-confirmed advanced adenomas and negative controls, applying its analysis pipeline to derive a signal to differentiate the two. The resulting predictor showed 41 percent sensitivity and 90 percent specificity, a performance comparable to that of existing stool-based tests, the authors wrote.
Freenome CEO Gabriel Ott has said previously that the company expects to meet its goal of completing the PREEMPT trial by the end of 2021, and Lin said that with that pathway solidified, there is new energy to move on multiple new fronts.
"We built our company as platform-based, which means we can apply [our technology] to early cancer screening, but we can also apply it to many other indications, whether it's as we've shown now in terms of response prediction, [or in other ways] throughout the entire patient journey … so we're very excited [to continue to show] that the platform is flexible and robust and can be applied to multiple different problems beyond CRC early detection," he said.