NEW YORK – As the focus of multi-cancer early detection efforts has shifted increasingly toward epigenetic aspects of circulating tumor DNA, a new firm, Serum Detect, believes it has devised a unique and complementary signal in the T Cell receptor repertoire that could notably boost sensitivity of other approaches, especially for early-stage cancers.
The firm's technology originated in work by University of Texas Southwestern researcher Bo Li, now an associate professor of pathology and laboratory medicine at the University of Pennsylvania. Roman Yelensky, the firm's CEO, said that the company was formed a few years ago, licensed in the relevant IP, and began to further develop it for clinical use.
At its core, the company's methodology relies on what it calls circulating TCR repertoire functional units, computationally derived sets of TCRs with similar sequences that may recognize shared antigens. The firm's current assay uses PCR library construction followed by next-generation sequencing.
"The kind of breakthrough that was needed, we think, to see the signal that we do, was the idea that individual TCRs are insufficient for subtle detection applications, such as early-stage cancer," Yelensky said. "So, the algorithmic work that was done over the last several years was around how to group TCRs that are similar by sequence to make a more powerful predictor."
"Other groups have been working on related methods as well, but Bo has really been one of the key leaders in the space," he added.
The company presented a poster on its work at this week's annual meeting of the American Association for Cancer Research in San Diego, which demonstrated that, when combined with ctDNA and cancer-related protein biomarkers, the TCR-based approach could significantly improve sensitivity for the detection of early-stage cancer.
Study investigators analyzed buffy coat samples extracted as part of a standard blood draw from almost 1,000 lung cancer cases and controls. In addition, the cancer group was enriched for early-stage disease with 86 percent of samples from stage I tumors.
The researchers identified 197 TCR repertoire functional units (RFUs) that differed among cancer cases compared to controls, including 110 RFUs that were enriched in cancer patients and 87 that were enriched in non-cancer controls. The predictive power of any individual RFU was small, but using machine learning, the team was able to create an algorithm with much stronger performance, as represented by an area under the receiver operating characteristic curve of 0.7. AUC represents a test's accuracy without a locked sensitivity of specificity, with 1 representing a perfect test and 0 a completely unpredictive one.
Authors also reported that the AUC for stage I cancer appeared to exceed that for stage II-IV tumors in this initial cohort. With specificity locked at 80 percent, the TCR model could detect 50 percent of lung cancers. It was also robust against co-morbid, potentially confounding lung conditions, such as benign nodules and chronic obstructive pulmonary disease.
Yelensky said that Serum Detect believes that its commercial path forward is in pairing its TCR method with DNA-based technologies so that the two, independent signals can complement one another.
"We actually don't think of the other segments as competitors. We think this approach is complementary to those approaches, and we see them as potential partners, where we could complement the signal that they're seeing from those ctDNA analyses," he said.
This makes even more sense, he added, because the firm's assay uses the buffy coat, essentially a byproduct of DNA-based tests that are performed on extracted plasma.
In the firm's AACR poster, study authors estimated the gain in sensitivity for stage I tumors that could be seen if its immune-based approach were added to a ctDNA mutation method, protein biomarker analysis, or both, concluding that it could contribute a 20 percent sensitivity boost at the 90 percent specificity target commonly set for cancer screening tests.
Moving forward, the company is now working to extend that analysis to the methylation and fragmentomic ctDNA signals that have taken the lead over mutation profiling for blood-based cancer screening.
"What our goal is and what we think our data is showing, at least initially, is that the TCR base signal that we've identified can add [value]. We think we can help a lot of tests potentially," Yelensky said.
"That is really the big gap in the field right now," he added. "If you follow what has been published for methylation-based liquid biopsy, for example, both Guardant Health and Freenome have had great data in colorectal cancer, but … they really start to see reduced performance for stage I cancer and pre-cancers."
"What we're showing what the TCR signature suggesting right now is that it's boosting signals substantially in the earliest stage of disease. Plus, there's a suggestion that it doesn't get weaker with earlier stage, unlike ctDNA. If anything, maybe it gets a little bit stronger, which one might expect and hope from an immune-based signal because immune responses tend to be stronger when you don't have advanced cancer," Yelensky said.
"We think we can boost performance exactly where ctDNA underperforms,” he added.
Yelensky said that Serum Detect believes it can demonstrate similar results in other tumor types. The decision to start its work in lung cancer reflects the disease's position as the number one cancer killer. "Also, because it's an immune approach and lung cancer is known as a 'hot tumor,' we thought if it was going to work anywhere, it should work there."
In terms of commercial next steps, Serum Detect plans to seek out a partner or multiple partners in the ctDNA-based cancer detection space. "I think we are very open at this point to any kind of any combination," Yelensky said.
The firm also has what he called a secondary interest in developing a combined ctDNA and TCR test in house for use in patients with pulmonary nodules.
Yelensky said that the company is hoping that it will have an easy time moving forward with pilot studies pairing its TCR functional unit method with ctDNA tests because prior studies of existing mutation, methylation, and fragmentomic approaches may already have banked buffy coat samples that match the plasma samples used.
"It's further validation for the TCR score, but at the same time it also answers this important question of, does the TCR score add signal to whatever the ctDNA application may be," he said.
Serum Detect hasn't yet disclosed any of its fundraising results. Yelensky said the firm has been "focusing on the science" and is funded sufficiently to now shift toward partnership conversations and hopefully collaborative studies. After that, it will be looking to raise more in the future for expansion and in-house development efforts.