The Rapid Rise of Biomarker Research and Its Significance
Until around 30 years ago, biomarkers were viewed primarily as indicators used in routine health checkups. For example, elevated blood CRP levels signal inflammation somewhere in the body, while persistently high circulating γ-GTP levels are common in heavy drinkers. In short, biomarkers were traditionally used as general barometers of health and disease.
In recent years, however, molecules such as Her2 and PD-1/PD-L1 in oncology as well as amyloid-β and phosphorylated tau in dementia have emerged as critical drug targets, leading to major therapeutic breakthroughs.
In infectious diseases, the foreign pathogen itself serves not only as a biomarker but also as the direct drug target. This is why such diseases were among the first to be effectively treated—therapeutic strategies are relatively straightforward, relying on vaccines and antibiotics.
In contrast, diseases like cancer or dementia arise insidiously from within the body. Their underlying causes are difficult to pinpoint, as the precise “moment of disease onset” remains obscure. As a result, medical researchers have had to shift focus: rather than identifying the true cause directly, they explore biological features that are closely associated with it.
This approach has led to the recognition of Her2 and phosphorylated tau as biomarkers—keys to understanding disease mechanism and guiding therapeutic development.
Antibody drugs such as trastuzumab (Herceptin) and Enhertu, targeting Her2, have brought dramatic benefits to breast and gastric cancer patients, while pembrolizumab (Keytruda), targeting PD-1/PD-L1, has shown remarkable efficacy across multiple cancers. These success stories demonstrate that biomarkers have evolved from “diagnostic indicators” into true “drivers of drug discovery.”
Rethinking What a Biomarker Is
A biomarker, simply put, is a molecular or cellular feature that characterizes a disease. In medical science, it provides the starting point for “making disease visible and measurable.” In drug discovery, it defines the “target on which drugs should act.”
Because all drugs exert their effects by interacting with specific molecules, identifying molecular targets relevant to a disease is essential. This principle underpins the current paradigm of “molecular medicine” and “molecular therapeutics.” As scientific understanding has advanced toward viewing the body at the molecular level, drug-target interactions became a prerequisite for rational drug design.
In the 20th century, many biomarkers were discovered serendipitously — Her2 and PD-1 being notable examples. In the 21st century, the approach has shifted toward genome- and proteome-wide analyses, uncovering candidates such as phosphorylated tau.
Yet, decisive biomarkers remain elusive for many diseases. In oncology alone, pancreatic cancer, cholangiocarcinoma, small-cell lung cancer, glioblastoma, and anaplastic thyroid cancer lack well-established biomarkers. Neurodegenerative and autoimmune diseases present even greater challenges.
Interestingly, the reverse scenario also occurs: when different diseases share the same molecular target, a single drug can be effective across multiple conditions. For example, tofacitinib (Xeljanz®), originally developed by Pfizer for rheumatoid arthritis, was later approved for ulcerative colitis. JAK inhibitors have also shown efficacy in treating atopic dermatitis, illustrating how targeting common cytokine signaling pathways can yield therapeutic benefits across distinct diseases.
This highlights how biomarker discovery fundamentally reshapes both medicine and drug development.
A New Methodology for Biomarker Discovery
Molecular structure refers to the relative positions of the atoms that make up an amino acid. Atoms can only be seen with short-wavelength electron beams or X-rays; they cannot be seen with light. Because it requires sophisticated technology such as an electron microscope or X-ray diffraction equipment, it is not possible to casually survey them, and there was no methodology to discover realistic structural features.
COGNANO has developed a unique methodology: analyzing vast antibody datasets to identify disease-specific molecules through a reverse approach to conventional methods. By extracting antibodies that preferentially bind to diseased, but not healthy, molecules and cells, we have established a novel system called IBMET® (Inverse Biomarker Exploring Technology).
Because antibodies recognize specific structural features of antigens, identifying their binding sites helps illuminate disease-specific profiles of the target molecule. Moreover, antibodies developed through this method can themselves serve as therapeutic agents, which is a significant advantage. Using antibodies as probes to examine protein structures reflects a kind of retrospective engineering―like repurposing an old gear into a novel instrument.
The Significance of Biomarkers
The essence of biomarkers is to "indicate abnormalities, particularly at the molecular level, that correlate with disease." Nucleic acids and proteins are two highly practical disease indicators, and proteins have historically and operationally played a leading role as drug discovery targets. The theory that "amino acid sequence determines protein structure, and structure determines function" is a central tenet supporting drug discovery. Based on this principle, humanity has evolved structural biology, eventually reaching the point where structure can be predicted using machine learning algorithms.
So, can structure prediction algorithms such as AlphaFold discover new biomarkers? Because structural training data in diseases is extremely scarce, predictions that "accurately identify the structural characteristics of disease" are not possible. On the other hand, when a biomarker molecule is identified, the ability to design candidate chemicals for the specific structure of the molecule has become quite feasible. However, these technologies are not yet at the level of biomarker discovery.
IBMET®, on the other hand, is a methodology for discovering new biomarkers that approaches the cause of disease more closely than structure prediction algorithms. It is an engine for discovering new biomarkers by taking advantage of information belonging to drugs, namely antibodies, and has the power to revolutionize conventional drug discovery methods. To test its effectiveness, we are preparing trial samples that will be available to researchers and pharmaceutical companies around the world. COGNANO has established a group of antibodies whose biomarker performance has been confirmed in pancreatic cancer tissue, and plans to provide them free of charge to researchers conducting non-profit research and publishing.
By accumulating the results obtained in this way, COGNANO hopes to share and popularize the new concept of "structural biomarkers" with researchers.
AI and the Next Frontier of Drug Discovery
While AI is widely applied in life science today―with AlphaFold making a significant breakthrough, AI-driven drug discovery still requires clearly defined disease targets to begin. Moreover, post-translational protein modifications and intrinsically disordered proteins (IDP) remain unresolved challenges that current prediction models struggle to address.
COGNANO is pioneering an antibody-based approach by immunizing alpacas with disease-representative cells and molecules, then analyzing their resulting antibody repertoire to identify disease-specific biomarkers. This work represents one of the few global examples of computer-assisted biomarker discovery―a field we refer to as Structure-Omics.
The biomarkers discovered by Structure-Omics (IBMET®) are, strictly speaking, "marker structures" rather than "marker molecules." Until now, drug discovery relied on molecular names because the target structure was unknown. For convenience, drug seeds were designed by "replacing the target with a molecular name." In fact, the more accurate term is "structural targets" rather than "molecular targets."
In contrast, VHH antibodies function as structural probes with billions of pixels, enabling precise mapping. Since their full sequences can be digitized and barcoded, they allow us to build a Google Maps-style atlas of human molecular structures.
To accelerate AI-assisted drug discovery, we are organizing an AI competition to predict the fine structural regions (epitopes) that antibodies recognize on target molecules. Epitope identification currently relies on resource-intensive techniques such as proteomics or cryo-EM, highlighting the urgent need for AI-driven prediction. An epitope is the geometric area of a target to which an antibody binds. Resolving the challenge of epitope prediction will likely lead to new disease concepts and new drug discovery flows that were previously undetectable through conventional structure prediction.
This will also be the final piece of the puzzle for COGNANO's "automated drug discovery," bringing us one step closer to our goal. We hope many people will take part in this AI competition, working toward the realization of structure-targeted medicines, which will break through the limitations of conventional drug discovery, which relies on molecular names as guideposts
Conclusion
Biomarkers are essential for understanding diseases and a prerequisite for drug development. COGNANO is building a new system for discovering disease-specific molecular structures using antibodies, powered by global data-sharing and AI-driven prediction.
The discovery and application of biomarkers will play an increasingly pivotal role in shaping a healthier future for humanity. Once biomarker prediction becomes a reality, structure prediction algorithms such as AlphaFold will be able to play an integral role as part of the drug discovery workflow. We hope that distributing pancreatic cancer biomarker antibodies and launching the international AI competition will mark a significant leap forward on our journey.