NEW YORK – Spanish cancer IVD developer Universal Diagnostics has shifted its commercial strategy from using mass spectrometry for early cancer detection to a liquid biopsy approach that identifies DNA methylation biomarkers to detect precancerous cells in patient blood samples that may indicate early-stage colorectal cancer (CRC) or other tumors.
The Seville-based firm presented data at the 2020 European Society for Medical Oncology's (ESMO) Virtual Congress last weekend showing that its newly developed approach can identify signs of early-stage cancers such as advanced adenoma (AA), a CRC precursor.
Universal Dx originally developed an assay that applies mass spectrometry to measure a panel of 30 small-molecule metabolomic biomarkers in blood samples that are linked to the presence of CRC and precancerous cells.
However, the company encountered challenges with metabolomics in 2017 that prevented it from achieving the specificity needed to commercialize the technology, explained Universal Dx Cofounder and COO Christian Hense. The company therefore began searching for various technological methods to accompany complement the metabolomics platform, including proteomics, transcriptomics, and DNA methylation detection.
"As we were comparing data across these different technologies, we realized methylation was the one we wanted to focus on," Hense said. "We also [eventually] realized that we'd get more robust and accurate results [solely] with methylation."
As such, Universal Dx switched solely to methylation detection to spot precancerous cells in CRC and other cancers. The firm has partnered with Imperial College London; the Austrian Institute of Technology; Liège, Belgium-based Diagenode, and Barcelona's Polytechnical University of Catalonia to develop the methylation-based technology. Hense noted that these collaborators specialize in methylation, sequencing, or bioinformatic elements that the firm included in its platform.
The new approach involves spinning 20 to 40 ml of a patient's blood sample down to plasma, followed by cell-free DNA (cfDNA) extraction. After analyzing the extracted cfDNA using quantitative PCR or next-generation sequencing (NGS), the firm then applies a proprietary algorithm to the data to identify biomarkers of potential CRC, AA, and other early-stage cancers.
While Universal Dx is still finalizing the type of data that it will include in a patient's clinical report, Hense noted that it will most likely provide one of two outcomes to the oncologist. If the algorithm is positive and spots precancerous biomarkers, the firm will recommend that a patient undergoes a colonoscopy. However, if the test is negative, it will advise the clinician to request another one of the tests in one to two years.
After receiving a blood sample from a clinician, the firm believes that it can generate a clinical report within one to two business days. Hense said that the company is currently deciding whether to build its own CLIA-certified, CAP-approved lab in the US, or to collaborate with a third party that has an established lab to provide its services. The firm also plans to build an ISO 13485-certified lab to eventually offer the test in Europe.
While Universal Dx initially built the methylation platform using qPCR, Hense said that it plans to develop the commercialized version of the assay using NGS. In addition, the firm anticipates that the final version of the test will only need 5 to 10 ml of a patient's blood sample.
"NGS allows you to measure more biomarkers, which opens up the possibility to measure for [more] CRC advanced adenomas, but also [other] cancers we want to measure eventually," Hense said. "We will also potentially look at additional biomarkers for the multi-cancer panel."
ESMO presentations
In one ESMO poster, Alexander Perera-Lluna, study senior author and associate professor at the Polytechnical University of Catalonia, and his team developed an open chromatin region (OCR)-based model using Universal Dx's platform to predict AA from plasma cfDNA bisulfite sequencing data.
To train the OCR model, the team performed whole-genome bisulfite sequencing (WGBS) on 10 pooled patient samples, which included two with AA, three with CRC, and five control samples. The group then selected OCRs showing maximum differences between case and controls with low inter-sample variance. The researchers then validated the model on 10 additional patients with AA and 10 control samples.
Perera-Lluna's team found that 1,074 OCRs showed strong case-control separation power in the training data. The cfDNA contribution in colon tissue-specific chromatin was also higher in cases than those of controls.
Overall, the researchers found that the OCR model had 50 percent sensitivity at 90 percent specificity in the validation set.
Perera-Lluna's team therefore believe that the model has potential to act as a method for AA detection and could serve as a basis for early-stage CRC detection.
In another presentation, Universal Dx and its academic collaborators performed WGBS on three control samples, as well as samples from three patients with AA and four with CRC to identify differentially methylated regions in advanced colorectal neoplasia patients. The team then evaluated regions in plasma samples in a prospective cohort of 110 patients.
The researchers first applied Universal Dx's qPCR methylation-based method to target candidate regions in plasma cfDNA. They then tested the methylation panel by dividing the 110 samples into training and testing sets, ranking the markers to build a prioritizing algorithm.
The researchers found that the algorithm, built on 35 markers, had an area under the receiving curve (AUC) of 80 percent, with a clinical sensitivity of 63 percent and 88 percent specificity, to detect AAs. The algorithm also had varying sensitivity and specificity for high- and low-grade dysplasia, tubulovillous AA patients, and certain gastrointestinal diseases.
The study authors therefore believe that the plasma methylation markers could help detect AAs, especially for adenomas with high-grade dysplasia. In addition, the results suggest that the markers could be used to develop a "highly accurate and minimally invasive blood-based screening test" to help guide downstream treatment" and cancer patient evaluation.
In a third presentation, James Kinross, study senior author and clinical senior lecturer in colorectal surgery at Imperial College London and his colleagues applied Universal Dx's methylation platform to build a panel to test for several cancers, including CRC, lung, pancreatic, and breast cancer, in patient plasma samples.
Using data from the Cancer Genome Atlas to select initial methylation markers, the group evaluated methylation marker regions in 101 plasma samples and 71 controls. The group then used the platform to target regions in plasma cfDNA.
Building an algorithm to detect and test the biomarkers, Kinross' team established one panel for overall cancer detection and one for identifying the local tissue of origin. By running the algorithm on the samples, the researchers found that they could identify differential methylation patterns on a single-marker level.
The first methylation-based panel, which included 10 biomarkers, had a clinical sensitivity of 79 percent and clinical specificity of 90 percent, for overall cancer detection. The Stage I cancer panel had a sensitivity of 75 percent. In addition, the individual markers showed differential methylation levels between the different cancer types (CRC, pancreatic, lung, and breast cancer).
The second panel, which contained 16 biomarkers, successfully identified the correct tissue of origin in 80 percent of CRC, 78 percent of lung cancer, 75 percent of pancreatic cancers, and 62 percent of breast cancer patients.
However, Kinross acknowledged in an email that the study's biggest challenges involved defining the different patient cohorts for complex diseases like cancer.
"This not only means robustly quality assuring the staging strategy across cancer types to ensure that tumor burden and hence detection limits are precisely defined but also [that] control groups are also robustly phenotyped," Kinross noted. "In this case, it meant performing endoscopy in a representative screening population."
But Kinross and his team believe that Universal Dx's methylation-based method could serve as the basis to develop a simple, accurate, and minimally invasive blood-based multi-cancer screening assay. He pointed out that methylation changes occur very early on in the disease's development and are organ-specific.
"The method is scalable … and can be deployed across populations for nationalized screening programs," Kinross said. "A blood test is [also] likely to be more acceptable to patient cohorts than a fecal test for bowel cancer."
Commercial aspirations
Universal Dx has filed six patents with the US Patent and Trademark Offices for IP related to specific biomarkers informative for CRC and AA, as well as biomarkers informative for lung, breast, and pancreatic cancers.
Since launching in 2012, Universal Dx has raised roughly $27 million in funding to develop the methylation-based method, primarily from European undisclosed private investors, including biotech funds and high net-worth individuals.
The company hopes to finalize development of the CRC screening assay by the end of 2021, then internally validate the test followed by commercialization.
Because Universal Dx is currently in discussions with multiple different NGS providers to run its methylation panels, Hense declined to disclose which specific instrument the firm will select for the final version of the test.
Universal Dx will initially focus on commercializing the CRC test in the US market and expects to launch the CRC assay as a laboratory-developed test in 2022, potentially followed by seeking US premarket authorization approval. Universal Dx is also pursuing a CE mark to eventually offer the CRC test in Europe as it develops LDT version for the US.