Guardant Health shared new data this week from its efforts to validate blood-based tests for early cancer detection and monitoring, reporting on a study of samples from colorectal cancer patients in which it applied a combination of somatic and epigenetic analyses to classify samples as either positive or negative for the presence of a tumor.
The firm is currently a few months into offering its first commercial, research-use-only assay for early cancer detection, called LUNAR. Right now that test combines only mutational and methylation analyses. But what Guardant calls fragmentomics is also emerging as an important focus of internal research, according to the company's presentation yesterday at the annual meeting of the American Association for Cancer Research here.
The timing is significant considering Guardant's announcement last week that it is
acquiring University of Washington spinout Bellwether Bio.
Bellwether
was formed to advance a method of blood-based cancer detection developed in the lab of UW professor Jay Shendure. That company's focus has been on detecting cancer via patterns of circulating DNA fragments, which reflect the unique nucleosome positioning within these molecules as they once existed in their cell of origin. Investigators led by Shendure and his former graduate student Matthew Snyder first published on their discovery regarding nucleosome positioning
in 2016.
Guardant Health President and Chief Operating Officer AmirAli Talasaz said in an interview that he and others at Guardant have known Shendure for years, and that the company had been talking with Bellwether about its work for some time before making the decision to acquire the firm.
The potential added value of layering new sources of cancer signal — like nucleosome positioning or fragmentomics — over more commonly interrogated somatic mutations, is "amazing," Talasaz said. "Scientifically, [there is] an incredible amount of signal in tumor cells, so if you have the right analytical tools you can improve sensitivity and specificity significantly," something the company believes it has begun to demonstrate with the new data.
"It's clear that [Bellwether] is one of the pioneers, and in terms of this fragmentomics approach, we think they are the pioneer, with the paper they published in 2016 being a landmark," Talasaz said. "It really made sense to us to join forces, and was a good opportunity for them because they are truly passionate to get this technology out in the clinic," he added.
In the meantime, Guardant has begun to provide evidence of what the incorporation of fragmentomcis and nucleosome positioning with Bellwether may add to its early detection and monitoring assays.
In a presentation on Sunday, Talasaz described a study of a combined approach — including circulating tumor DNA sequencing, methylation analysis, and fragmentomics — in a small case control cohort of early- to late-stage colorectal cancer patients and age-matched controls.
Overall, the company's assay was able to produce 90 percent sensitivity at a set specificity of 89 percent. Setting specificity cutoffs higher, at 94 percent and 98 percent, sensitivity was 88 percent and 79 percent, respectively.
Importantly, investigators also broke down the data by cancer stage. At 89 percent specificity, sensitivity was 100 percent for stage IV samples, 95 percent for stage III, 90 percent for stage II and 84 percent at stage I.
At 94 percent specificity, sensitivity dropped to 76 percent for stage I tumors. And at a 98 percent specificity cutoff sensitivity was 64 percent.
Interestingly, the company also calculated what assay sensitivity would have been at these various cutoffs if it was limited only to ctDNA sequence, with detection levels dropping nearly 50 percent compared to what the combined assay could achieve in early stage patients.
Among Guardant's competitors in the cancer early detection race, several have also presented data in colorectal cancer, in some cases also combining both mutation detection and methylation.
In a study presented at the American College of Gastroenterology Annual Meeting last year,
Freenome investigators shared data on about 800 CRC samples most of which represented early-stage disease (up to stage II).
As in Guardant's study, sensitivity was lower for the earliest-stage cancers — dropping below 80 percent — but stayed close to the 82 percent level for stage II and III patients and was close to 100 percent for late-stage cancers.
Grail previously reported that its method could pick up 69 percent of stage I and II CRC cases, increasing to 85 percent detection of stage III and IV tumors.
Talasaz said that Guardant believes that the layering of multiple signals, in its case somatic mutations, methylation, and now fragmentomics, additively benefits both sensitivity and specificity. At least thus far, the signals don't seem to be redundant, he added.
The firm is not alone in recognizing the need for, and value of, multi-modal strategies for this difficult project of detecting incipient cancers. Among those promoting combined approaches,
investigators from Johns Hopkins have shown, for example, that protein markers may also offer added value.
As part of his presentation of the Guardant CRC results, Talasaz shared a slide that plotted each sample in the study by epigenomic tumor fraction against somatic tumor fraction, demonstrating a large group of well-defined cancer positives with both epigenomic and mutation signals of cancer. However, there were also defined groups in which there was either only an epigenetic signal or only a somatic signal. Interestingly, all of the assay false-positives appear to have been in either of these single-positive subsets, though the same was true for a significant number of early-stage cancers.
According to Talasaz, the sensitivity and specificity numbers in the current study — near 90 percent sensitivity even in stage I cancers at a 90 percent specificity cutoff — is something Guardant believes would be competitive considering currently available clinical tools.
For other screening applications in other cancers, where there isn’t the same clinical paradigm that CRC offers (with colonoscopy as a readily available diagnostic follow up) the firm might need to improve performance further.
As with other studies of this type, questions remain around the status of individuals who were false-positives according to the Guardant assay. If followed for longer time periods these presumably cancer-free individuals that nonetheless have a somatic or epigenetic cancer signal at some point might come to show signs of a tumor.
Talasaz said that Guardant is investigating this and will hopefully have more information to share in the future.
