NEW YORK – At the European Society for Medical Oncology Congress on Saturday, liquid biopsy startup Precede Biosciences showcased what it believes is a novel platform that incorporates the activation status of gene promoters and enhancers through the analysis of circulating chromatin, overcoming limitations of both DNA mutations and methylation analyses to detect the presence of cancer, subtype it, and predict its response to treatment.
The Boston-based company believes its technology can be broadly useful as a tool for identifying transcriptionally regulated drug targets, and as a minimally invasive diagnostic or therapeutic indicator. The company debuted its approach, alongside case studies illustrating potential clinical utility, in an oral presentation at ESMO, where Sylvan Baca, a medical oncologist at Dana-Farber Cancer Institute and a Precede cofounder, shared data from a study in which researchers analyzed blood plasma samples from more than 400 individuals representing 15 cancer types. The data were also published simultaneously in Nature Medicine.
Precede CEO Rehan Verjee said that his background in the pharmaceutical industry, coupled with the company's Dana-Farber cofounders' precision medicine expertise, converged in a recognition that both the development and clinical use of precision medicines had been hampered by the limitations of DNA mutations, epigenetics, and transcriptomics in explaining the complex biology of tumor development, evolution, drug response, and drug resistance.
"I'm really proud of some of the things we achieved in getting new drugs across the line, but I was also really impacted by the inability sometimes to make really good decisions in R&D just because we lacked access to important relevant biology," Verjee said. "You treat patients [and] they respond but is that response really correlated with the modulation of the pathways that you think you should be impacting?"
And when studies had been unsuccessful, there often weren't clear answers as to why the therapy failed. "You're scratching your head," he said. "Too often, in my mind, we would be empirically moving on to the next thing as opposed to really understanding why this large $150 million experiment didn't work."
Precede is betting that its technology can overcome these limitations. The company is focused on interrogating disease-defining transcriptional biology via analysis of both gene methylation and histone modifications found within circulating chromatin structures.
Chromatin, according to Verjee, is more dynamic and far more correlated with gene expression than methylation is. "You need to open up the chromatin to let the transcriptional machinery get in, and you need to mark the chromatin with regards to where you want the transcriptional machinery to actually collect," he said.
Explaining the company's approach, Verjee cited early research in recurrent prostate cancer, which showed that a broader view of gene regulation via enhancer regions could allow prediction and functional understanding of both therapy response and resistance in cases where DNA mutations and methylation fell short.
During his presentation at ESMO, Baca said the firm's technology involves an immunoprecipitation-based approach to generate genome-wide profiles of DNA methylation and histone modifications associated with active enhancers and active gene promoters. "If you haven't thought about histones in a while, histones are proteins that DNA is wrapped around, and they have tails that can be marked with specific modifications," he said. "Placing and removing these modifications is a major way that cells regulate gene expression and decide which genes are turned on and turned off."
"That's why we're called Precede, because we're looking at the signal preceding transcription," Verjee said. "We decided to form the company, and here we are, two years later."
In the study presented at ESMO, Baca and colleagues generated 1,268 plasma-based epigenomic profiles from 433 individuals with one of 15 types of advanced cancer. One of the first takeaways from this study, according to Baca, is that individual cancer types appear to feature high promoter signals at the genes that would be expected to be expressed in their type of tumor.
For example, in a prostate adenocarcinoma case, the study showed a high signal at the KLK3 promoter, which encodes prostate-specific antigen (PSA). In contrast, an estrogen receptor-positive breast cancer case showed a high promoter signal at the ESR1 gene. Apart from these individual examples, Baca said that there was a high specificity for these cancer-specific promoter signals across the cohort.
The same patterns emerged looking at therapeutic targets, he added. Based on the level of promoter signal at the ERBB2 locus, which encodes HER2, investigators could distinguish between pathologically confirmed HER2-positive and HER2-negative cancers. He highlighted a colorectal cancer case that hadn't initially been stained for HER2, but whose blood test results showed an "extremely high signal at ERBB2."
"We went back and looked at metastatic biopsy tissue and confirmed that HER2 was indeed expressed in this cancer, as it is in about 2 percent of colorectal cancers," Baca said.
With the recent expansions of HER2-directed drugs into tumor types other than breast cancer, renewed attention is being paid to questions of biomarker sensitivity and inter-assay concordance. A long-recognized problem is inter-lab variability in tissue-based HER2 testing, which involves subjective, hands-on analysis by pathologists and highlights the potential value of a blood-based, algorithmic alternative.
In another example, Baca highlighted the fact that small cell neuroendocrine cancers in this study cohort showed high promoter signals at the DLL3 gene. "DLL3 is interesting because it's a cell surface protein that's targeted by several antibody-based therapies in development," Baca said. "This provides an example of how you could think about using an assay like this [in drug trials] to select patients who might be more likely to respond because they express the target."
According to Baca, the study also demonstrated that enhancers provide additional information beyond what can be gleaned from promoter signals. As sites where transcription factors bind to regulate gene expression, enhancer regions are often highly specific to a given cell type or cell state, he said.
Moreover, they are much more ubiquitous than promoters, with about a million enhancers scattered across the genome compared to roughly 20,000 promoters. Looking at patients' plasma samples, the Dana-Farber team measured elevated enhancer signals at cancer type-specific regulatory elements that had been defined in a separate tumor tissue study. The same was not true for control samples.
According to Baca, this suggests a potential for Precede's platform to be applied to both cancer detection and tumor type differentiation, an approach already being pursued for pan-cancer early detection and screening by a number of commercial firms.
Finally, the study showed that enhancer activity can also be useful for identifying transcriptionally regulated drug targets, as well as in the context of emerging resistance.
Baca used the example of the androgen receptor (AR) gene. "It's normally turned off in adult prostate tissues, but when prostate cancer becomes resistant to androgen inhibition therapy, it turns it back on … boosting transcription of the gene [and] allowing prostate cancer to survive in low-testosterone conditions," he said. As such, enhancer activity can be used to differentiate AR-driven resistance from other resistance mechanisms in prostate cancer patients who progress on androgen deprivation therapy.
Neuroendocrine transformation is another mechanism that can evolve in a cancer's progression toward resistance to existing therapies. In their study, the Dana-Farber team was able to use enhancer signals from multiple sites to create a classifier, which, according to Baca, could accurately distinguish neuroendocrine prostate cancer, small cell lung cancer, and Merkel cell carcinoma from non-neuroendocrine cancers. The authors reported an area under the receiver operating curve of 0.93 for separating neuroendocrine cancers from non-neuroendocrine tumors.
Baca suggested that this type of approach could also be used to follow treatment resistance in patients over time by correlating clinical factors with epigenomic changes. He highlighted the case of a patient with EGFR-mutant lung adenocarcinoma who progressed on AstraZeneca's EGFR inhibitor Tagrisso (osimertinib). "We profiled plasma from this patient before therapy and after the development of resistance, and we looked at enhancer signals at adenocarcinoma regulatory elements and neuroendocrine regulatory elements," Baca said. "After small-cell transformation, there [was] a loss of signal from the adenocarcinoma enhancers and a gain of enhancer signal at neuroendocrine regulatory elements, consistent with small-cell transformation."
Verjee said a number of pharma companies have already partnered with Precede to use its platform in drug development, but he declined to provide further details.
In emerging from stealth mode earlier this month, the company hopes to open up its platform for broader use. "Rather than us just talking to people, we'd love for people to come and talk to us," he said.
In drug development, the firm is open to exploring use of its platform in any context where cell-free DNA is abundant, including cancer and autoimmune disease.
On the clinical diagnostics side, Verjee said that Precede has specific programs moving forward in breast and lung cancer, as well as in pan-cancer applications. The new ESMO data lays out some use cases that may feature in the firm's future efforts. "Our focus is on areas where drugs are already used today, where they require an invasive test to detect the presence or absence of some transcriptionally regulated phenotype, and we can create the minimally invasive alternative to that," he said.
In cancer, in particular, the infeasibility of tissue-based biomarker testing can present a big barrier to delivering precision medicine. "The more [patients] progress, the less willing or able [they] are to get biopsies, and the more difficult it is to get an impression of the disease from a single biopsy," Verjee said.
Beyond that, he noted there is also a need for minimally invasive testing in the context of antibody-drug conjugates, for example. "There are so many targets at the moment that people are going for across so many cancers, and there's no way to test for those in a minimally invasive way." Estimating that there are at least 60 new ADCs in development as of last year, Verjee said that there "might not be a better time" for the company's entry.