Bertis Publishes Findings on 9 Protein Biomarkers and
Algorithm-based Test Model for Breast Cancer Early Detection in International Journal
- Propose a test model with AUC 0.9105 using 9 proteins as multiple biomarkers for breast cancer in stages 0-2
- Develop an innovative platform for discovering and validating disease biomarkers based on its self-constructed quantifiable peptide library
On June 21st, Bertis, a company specializing in the development of proteomics-based precision medicine technology (led by CEOs Dong-young Noh and Seung-man Han), presented its findings on developing a nine types of protein biomarkers blood test for early diagnosis of breast cancer using its library for quantitative analysis of proteins and artificial intelligence (AI) technology.
Through this research, designed to develop an advanced blood test for breast cancer diagnosis, the Bertis research team combined the quantitative values of 9 protein biomarkers to present a diagnostic test model with an average AUC 0.9105 for stages 0-2 breast cancer, which is an improvement from that of ‘MASTOCHECK®,’ the world’s first proteomics-based blood test for early diagnosis of breast cancer developed and commercialized by Bertis.
The research also revealed a platform for discovering and validating biomarkers that can drastically reduce time and cost in developing the proteomics-based diagnostic test. It uses 'PepQuant libraryTM', a library for quantifiable protein analysis developed by Bertis, which enables the discovery of biomarkers in the setting of a validation process. Instead of the existing difficult and expensive biomarker discovery method, this method uses the data pool of the library to perform discovery and validation together, saving time and cost and enabling a system that is easy to apply in the actual clinical environment.
a. AUC to predict breast cancer b. Predicted Probability of 9 biomarkers for early diagnosis of breast cancer and deep learning algorithm
(Normal, Stages 0-1, Stages 2-3, Other cancer (ovarian cancer, thyroid cancer, lung cancer, colon cancer and pancreatic cancer)
The research team generated the library that is composed of 852 quantifiable peptides that cover 452 human blood proteins, and then selected and presented 9 proteins from them as biomarkers for breast cancer screening. The team also developed an algorithm for early screening of breast cancer using its deep learning technology specialized in simultaneous multiple analysis of proteins. The study was published in ‘Scientific Reports’ on June 2nd.
Sungsoo Kim, Head of Bio Manufacturing subdivision, who led the research said, “These findings hold significance in that they show improved performance over MASTOCHECK®, which is currently commercially available, in screening for breast cancer. They also present an analysis platform that can be applied not only to breast cancer but also to various other diseases through the library.” He added, "Based on the results, Bertis will spare no efforts in future research and development to provide improved diagnostic tests to the medical field."
Seung-man Han, CEO of Bertis, said, “This study has established an efficient biomarker discovery and validation system for the development of proteomics-based simultaneous multiple testing based on quantifiable proteins library and AI technology. Therefore, we expect that the introduction of innovative tests into the medical field will gain momentum.”
Bertis is devoting itself to discovering biomarkers for major diseases and developing clinical solutions by incorporating proteomics technology accumulated over 10 years with the latest bioinformatics and AI technologies. The company has succeeded in commercializing proteomics technology through MASTOCHECK®, the world's first proteomics-based blood test solution for early breast cancer diagnosis, and PASS (Pan-omics Analysis Service & Solution), a pan-omics integrated analysis solution. It is also spurring its R&D efforts for early diagnosis solutions for various cancers, including pancreatic cancer and ovarian cancer, and analysis of blood proteomics with age.