NEW YORK – South Korean proteomics firm Bertis is moving into the US market with plans to open a San Diego-based CLIA laboratory by the end of the year.
The company will offer diagnostics including a test for the early detection of pancreatic cancer out of the lab as well as pharma research services, said Founder and CEO Seungman Han.
Bertis is also looking to expand its operations in other Asian markets like Singapore, Han said, noting that the company plans in the next several months to launch a funding round targeting around $30 million to $40 million to support these activities. It is also planning to go public in South Korea next year via an initial public offering, he said.
Formally formed in 2014, Bertis emerged from a 2007 proteomics research effort launched by the South Korean government as part of its 21st Century Frontier R&D Program, with the company "inheriting" the findings of the research funding through that project, Han said.
He said the company was moving into the US market to take advantage of the commercial momentum and investment that proteomics has seen in recent years.
"Last year we saw a number of proteomics companies go public on the Nasdaq, and there are a lot of proteomics companies that are putting a lot of drive into their research and development of clinical products," he said. "We thought that while there is this strong momentum in the market, it is best that we make use of it."
Bertis could be too late to the US to ride this most recent wave, though. While proteomics companies including Olink, SomaLogic, Seer, Nautilus Biotechnology, and Quantum-Si went public in 2020 and 2021, these companies' stock prices are all well below their highs, and 2022 has seen many fewer proteomics firms go public.
Bertis' lead product is its proteomic breast cancer test, Mastocheck, which measures the levels of three proteins in patient blood samples to determine whether a woman is likely to have breast cancer. The company received approval in 2019 from the Korean Ministry of Food and Drug Safety to sell the test as an in vitro diagnostic. In September of this year, it launched sales of the test in Singapore through Raffles Medical Group.
Han said the test is intended for use as a complement to traditional mammogram-based breast cancer screening, particularly in women with dense breasts for whom mammograms may not be as effective. Typically, doctors use the test to help determine whether a woman should undergo additional imaging, such as ultrasound or MRI, or a more invasive procedure like a biopsy. Han said the company also sees a potential role for the test as an initial screening assay in populations — particularly in Asian countries — with high levels of hesitancy around mammography.
Despite Mastocheck being available in Asia, Bertis doesn't plan to make the test the first product launched out of its US CLIA lab, Han said. Instead, it plans to lead with a proteomic test for pancreatic cancer that it is developing. Han said Bertis views the US market for pancreatic cancer early detection as larger and less competitive than the market for breast cancer diagnostics.
Bertis plans to position its pancreatic cancer test as a screening tool for individuals at high risk of the disease due to conditions like diabetes or reasons of family history, Han said. It is currently in discussion with several sites around the US about collaborating on a clinical study with the goal of launching the test as a laboratory-developed test by the end of 2023.
Last year, Swedish proteomics firm Immunovia launched its pancreatic cancer early detection test, called the IMMray PanCan-d blood test, in the US and is likewise targeting high-risk patients. In the first half of 2022, the company sold roughly $15,000 worth of IMMray PanCan-d tests.
While Bertis' initial focus in the US will be on the launch of LDTs, the company also plans to expand into pharma research services and companion diagnostics development.
The company runs data-independent acquisition mass spectrometry assays on Thermo Fisher Scientific instruments for its discovery work. For validation and clinical work, it uses targeted multiple-reaction monitoring assays run on Danaher's Sciex instrumentation.
In its discovery workflows Bertis combines extensive fractionation with a machine learning data analysis approach that aims to identify spectra that can help distinguish the condition being studied before determining what peptides are generating those spectra.
According to Han, the company divides samples — typically undepleted serum specimens — into 96 fractions in order to reduce the complexity of the samples and provide for deeper mass spec analysis. He said the company's discovery workflow takes around three hours per samples for mass spec analysis and that it typically looks at around 50 samples in its initial discovery experiments.
Such an approach runs somewhat counter to current trends in proteomics discovery experiments, where many researchers have moved to higher-throughput workflows with the aim of analyzing hundreds to thousands of samples. These higher-throughput approaches typically don't offer the depth of coverage provided by the extensive fraction used by Bertis, but the expectation is that by using large sample sets for discovery work researchers can better account for the biological and analytical variability that has challenged protein biomarker discovery and that the markers that do emerge from such experiments will be more likely to survive validation and, ultimately, prove clinically useful.
Bertis has also purchased proteomics firm Seer's Proteograph Product Suite, which uses nanoparticles to enable an approach to sample fractionation and enrichment more compatible with high-throughput experiments.
The company is also employing a unique approach to proteomic data analysis. In conventional protein biomarker experiments, researchers identify peptides present in the samples of interest based on the mass spectra generated and then analyze this data to look for differential patterns of peptide and protein expression that distinguishes between cases and controls.
Bertis, on the other hand, is using machine learning to look for patterns in the mass spectra that distinguish between cases and controls before it matches those spectra to the peptides that generated them. Having identified the distinguishing spectra, it then goes back and identifies the underlying peptides.
The company's researchers believe this approach will allow them to make use of more of the mass spectra generated during their experiments, said Chief Technology Officer Sangtae Kim, developer of the approach.
"We are discarding more than 50 percent of the spectra [generated by a typical mass spec experiment]," Kim said. "So we are utilizing just a fraction of the information."
By analyzing the spectra before matching it to peptides, Kim and his team are trying to better extract useful information from the spectra they typically discard due to poor quality. He said they are using machine learning language models for this analysis.
"Our first step is to build a classifier using this modeling, and then we reverse-engineer this model" by identifying the peptides responsible for the features important to the model, Kim said.