NEW YORK – Dartmouth researchers have launched precision medicine startup Episteme Prognostics in the hopes of commercializing an epigenetics-based technology to predict whether pancreatic cancer patients will respond to chemotherapy — and, down the line, whether they might be eligible for new targeted therapies.
Episteme Prognostics, launched as an LLC in 2019 and converted to an incorporated company this February, is still in its infancy. However, with validation studies ahead for its foundational product, dubbed ATAC-array, and seed funding rounds still ongoing, cofounders Steven Leach and Surajit Dhara of the Dartmouth-Hitchcock Norris Cancer Center have ironed out a clear plan for what they hope to accomplish in the coming years.
First, they hope to secure government and private investor funding for their test's additional validation studies. Second, they plan on obtaining 510(k) premarket clearance or de novo market clearance from the US Food and Drug Administration for the test as a tool for determining which pancreatic cancer patients, based on their recurrence risk, would benefit from gemcitabine and nab-paclitaxel chemotherapy. Finally, down the line, they will conduct clinical studies to evaluate the test as a companion diagnostic for personalizing epigenetic therapies for pancreatic cancer patients.
The small company, currently run by a team of six experts who are contributing their time on a part-time basis, is focused on improving treatment options for pancreatic cancer patients. Dhara and Leach, who serve as Episteme's president and scientific advisory board chair, respectively, hope to bring personalized therapies to patients with pancreatic ductal adenocarcinoma, a notoriously aggressive tumor type that withstood advances in targeted treatment.
"One of the many reasons that pancreatic cancer has represented such a challenge is that, unlike many tumor types ranging from C-KIT-mutated gastrointestinal stromal tumors, to BRAF-mutated melanoma, to EGFR-mutated lung cancer, pancreatic cancer doesn't have a dominant molecular subtype for which there is a targeted molecular therapy," Leach said.
As a result, pancreatic cancer patients' treatment options are, for the most part, limited to one of two chemotherapy regimens, after which disease recurrence rates are high. "We've treated everybody as if they have the same disease, and while their tumors may look similar under a microscope, we know that their molecular profiles — and, now, their epigenetic profiles — are dramatically different."
Focus on epigenetics
Episteme's ATAC-array identifies an adverse pattern of tumor-intrinsic chromatin accessibility, an epigenetic feature that plays a role in gene expression mechanisms like DNA methylation, and in turn, affects tumor growth and suppression. Accordingly, Dhara, Leach, and colleagues hypothesized that different patterns of chromatin accessibility might play a role in whether patients experience disease recurrence following chemotherapy.
The science behind the test, along with an initial validation study, was described in a recent paper published in Nature Communications. Using resected tumor samples from a cohort of 54 previously untreated pancreatic cancer patients at Memorial Sloan Kettering, Leach, Dhara and colleagues at MSK and Dartmouth first analyzed genome-wide chromatin accessibility patterns with a sequencing technology called ATAC-seq. This step allowed them to home in on more than 1,000 regions that were differentially represented between patients who, after surgical resection and adjuvant chemotherapy, lived longer than one year without recurrence and those whose cancer returned within a year.
Within these chromatin regions, the researchers also zeroed in on two transcription factors — ZKSCAN1 and HNF1b — for which nuclear localization patterns also correlated with disease-free survival times.
Following these experiments, they recognized the potential for both the chromatin accessibility patterns and the transcription factors' nuclear localization patterns to predict patients' outcomes. They also recognized, however, that the sequencing technology they had used to identify these potential biomarkers wasn't quite scalable.
"Even while sequencing has gotten more affordable, accessible, and rapid, the computational biology and bioinformatics expertise that it takes to make sense of [ATAC-seq] data still makes it relatively inaccessible to use in a clinical workflow," Leach said. "If I were forced to bet one way or the other on whether ATAC-seq data is going to ever … impact clinical decision-making, I would suggest it likely would not."
This concern was what ended up propelling the researchers toward their ATAC-array technology. To simplify the process such that it could be implemented in routine clinical practice, Leach, Dhara, and colleagues tried isolating just the accessible regions from the differential chromatin signature, arraying them on glass slides, and then hybridizing them with fluorescent-labeled ATAC libraries. The signals from the accessible region-only array approach, unlike the whole-genome approach required in the sequencing process, could be easily read and interpreted by pathologists.
"The idea was, if we know specific regions and have already discovered the signature, and if we print those signature DNA regions on our glass slide covered by appropriate probes, and hybridize the ATAC libraries with the probes, we can get an idea of the accessibility profile of the libraries," Dhara explained. "That was the novelty of the technology."
In other words, Leach and Dhara took a complex, expensive, and resource-intensive sequencing technology and boiled it down to a scalable microarray-based technology that, in their view, is comparable to commercial prognostic tests deemed cost effective, such as Agendia's MammaPrint breast cancer recurrence test.
Analyzing only the accessible chromatin regions, the researchers found, produced results that correlated closely with the ATAC-seq results and could predict for patients' outcomes to a similar extent. When they added in immunohistochemical determination of HNF1b nuclear positivity as a combined prognostic marker with the chromatin accessibility signatures, they were able to stratify patients into clear prognostic groups based on disease-free survival outcomes.
The prognostic groups included patients with "good" ATAC-array scores — that is, above the median score of 0.6 — and HNF1b localized to the nucleus; good ATAC-array scores and HNF1b not localized to the nucleus; poor ATAC-array scores and HNF1b localized to the nucleus; and poor ATAC-array scores and HNF1b not localized to the nucleus. Evaluating outcomes for 39 patients corresponding to these groups, the researchers found median disease-free survival times of 1,343 days, 940 days, 559 days, and 183 days, respectively.
Combining the ATAC-array score and HNF1b nuclear localization, the researchers concluded in their Nature Communications paper, offers a new "chromatin organization-based prognostic paradigm for precision oncology." The two combined technologies now comprise a product called PancEpistemeDx, which, contingent on further validation studies in the metastatic disease setting, could go on to be sold as a kit. If validated, pathologists could use the kit to determine patients' ATAC-array scores in fewer than 24 hours, Dhara said, adding that Episteme has not yet set a price for the test kit, but he expects the cost to be significantly less than that of a sequencing test.
Validation, commercialization plans
Episteme still has several steps to work through before it can begin selling the test as a kit, however. Although the recent paper in Nature Communications detailed results from an initial validation cohort, additional validation studies will be crucial for pursuing FDA clearance — something Leach and Dhara, along with the four other members of the Episteme team — see as necessary to market their test.
Through these additional validation studies, the researchers intend to prove that the test works on fine-needle aspiration biopsies and in metastatic lesions, as opposed to resected tissue specimens, which they used in their initial studies. Leach and Dhara emphasized that they did carefully separate epithelial cells and run their test on those in the initial discovery and validation studies, and that they replicated the test in very small numbers of cells. But they still must show that they can perform the test on scant tumor samples such as fine-needle biopsies, particularly since many pancreatic cancer patients, including those with metastatic disease, aren't candidates for surgical resection.
Dhara explained why this validation step is key. "That is the real value of this technology," he said. "We are hoping that we should know the [test] results at the time of first disease assessment, before decision-making for first-line therapy." The researchers are confident that the test will perform well in these further validation studies but ultimately, the FDA will have to evaluate the data.
Companion diagnostic potential
According to Dhara, Episteme's ATAC-array test is the "first ever microarray that can read chromatin accessibility." Down the line, it could contribute to one of the first precision medicine approaches to epigenetic therapy, too.
In addition to marketing the test for predicting chemotherapy response, Leach and Dhara see tremendous potential when it comes to using their test to identify patients who might instead be eligible for clinical trials — particularly those evaluating epigenetic therapies.
There are nine FDA-approved epigenetic therapies, Dhara explained, but few are indicated for biomarker-defined subsets of patients, let alone epigenetic biomarker-defined subsets of pancreatic cancer patients.
Given this untapped opportunity, along with the lack of targeted treatment options for pancreatic cancer, the company plans to first evaluate its ATAC-array as a potential companion diagnostic for the EZH2 inhibitor tazemetostat (Epizyme's Tazverik), which is FDA-approved for several hematologic malignancies, most recently EZH2 mutation-positive follicular lymphoma.
Tazemetostat has not been explored in pancreatic cancer as of now, according to Dhara. Episteme is already mapping out tentative plans for this trial, which is contingent on the success of further validation studies and FDA clearance.
Of course, these ambitions require funding, which is top of mind for the Episteme team as they work on raising a seed round of roughly $2.5 million and securing a pending NIH grant. These funds will get the ball rolling on the validation trial and regulatory activity.
Fortunately, Leach explained, Dartmouth's entrepreneurial track record has forged a path for Episteme. Indeed, it was Dartmouth's researchers who in 1987 launched Medarex, the company that developed the first checkpoint inhibitor immunotherapies, ipilimumab and nivolumab (now Bristol Myers Squibb's Yervoy and Opdivo). BMS acquired Medarex in 2009, but in the years since, Leach explained, "there have been a host of biotech startups emanating from our cancer center following Medarex's lead."