Biomarker Validation: Avoiding Misleading Results
- 7th August 2025
- Posted by: Coco Goto-Colverson
- Category: Biomarkers
The Role of Biomarkers in Drug Development
Biomarkers play a huge role in clinical research and real-world medical practice. They’re used to improve diagnostic accuracy, predict disease progression, assess treatment responses, and support personalised medicine strategies.¹ While biomarkers are incredibly valuable, they’re still flawed and can create larger issues further down the pipeline. In some cases, biomarkers may be inaccurate, misleading, or even simply missing. Misuse or misinterpretation can lead to misleading conclusions, ineffective treatments, and poor clinical outcomes, which is why biomarker validation is so important.
What Are Biomarkers?
Biomarkers are an objectively measurable biological characteristic that can be developed into assays and diagnostic tests.1 These markers can be molecular, like proteins, DNA, RNA and metabolites, typically found in biological samples including blood, urine or tissues² or physiological, such as blood pressure. Biomarkers identify specific biological indicators within the body that provide information about an individuals health status, offering insights into normal biological processes, the presence of diseases and how the body will respond to medicines.² They also serve as a foundational tool in the development of precision medicines, particularly in omics-based and oncology research.3
The Evolution of Biomarker Discovery
Interestingly, biomarkers were discovered accidently (an early example being the discovery of haemoglobin (Hb) by F. L. Hünefeld in 1840, who noticed crystals dried in blood samples which were then identified as haemoglobin) and overtime they became useful tools regularly used in research and medical treatment. Today researchers actively search for biomarkers to speed up the process of finding new treatments. Biomarker discovery shifted from being a by-product of clinical practice to a structured and industrialised process, accelerated by technological advances in molecular biology.4
Modern detection methods include:
- Immunoassays
- Molecular techniques
- Mass spectrometry
Combining these technologies with omics approaches allows for a holistic understanding of biological systems and discovery of complex biomarker signatures.2
The Ideal Biomarker
To be clinically relevant, biomarkers should ideally be:
- Specific to a particular disease or condition
- Easily accessible through non-invasive sampling (e.g., saliva or blood)
- Detectable and quantifiable, even at low concentrations
- Sensitive enough to reflect changes in disease state or treatment response2
Types of Biomarkers
Biomarkers are classified based on the type of information they provide:
- Antecedent biomarkers – help assess the risk of developing a disease
- Screening biomarkers – help detect subclinical disease and early-stage disease (before disease symptoms appear)
- Diagnostic biomarkers – recognise overt disease, by confirming (or ruling out) disease in people who are symptomatic or who have tested positive in a screening test
- Staging biomarkers – help categorise disease severity
- Prognostic biomarkers – used to predict disease outcomes or treatment efficacy
These classifications4 demonstrate their varied applications in clinical research and patient care.
The Role of AI and Machine Learning
The field of biomarker detection is continuously evolving, driven by the need for more precise, rapid, and accessible diagnostic tools. Artificial intelligence (AI) and machine learning (ML) are increasingly integrated into biomarker discovery and analysis. These computational tools can process complicated datasets from “omics” technologies, identifying intricate patterns and correlations that might be missed by human researchers.5
However, ML isn’t without pitfalls. It can suffer from:
- Overfitting – where models perform well on training data but fail on new data. This happens when the model can capture patterns specific to the training data but cannot generalise to new datasets. It’s especially problematic in biomarker detection, with small sample sizes and noisy data
- High dimensionality – when the number of variables (e.g. genes, proteins or metabolites) exceed the number of samples, increasing the risk of false positive biomarker detection
- Biases – ML algorithms can inherit human and data-related biases learned during training which can lead to unreliable biomarker predictions
While ML’s strength lies in making predictions and finding complex patterns, classical statistical methods focus on hypothesis testing and biomarker validation. A combined approach is often necessary to strike a balance between predictive power and interpretability, ensuring the identified biomarkers are both accurate and clinically meaningful.5
The Impact of Missing Biomarker Testing
Without proper biomarker information researchers are unable to provide each patient with the most effective therapy. This can lead to patients receiving treatments that offer little to no benefit and risk avoidable side effects.6 Furthermore, this gap leads to poorer clinical outcomes and missed opportunities for better care.
Barriers to Proper Testing:
- Knowledge Gaps: Some researchers may not be fully up to date on evolving biomarker guidelines.
- Logistics: Issues like inadequate biopsy sample quality
- Access Issues: Limited access to specialized testing labs hinders adoption
These barriers highlight the importance of not only using biomarkers but doing so correctly and consistently.6
Misleading Biomarkers
Various measurements can be elevated or thrown off for several reasons, that either don’t necessarily indicate anything harmful, or distract from other serious issues. It’s important to remember that a patient’s health is best understood by considering the full clinical picture, looking into lifestyle and other health factors. Take for example HbA1c, usually taken as a measurement of blood glucose. Heavy alcohol consumption can lower HbA1c, so looking at HbA1c in isolation could suggest a sign of good blood glucose control but with the alcohol intake in mind this may not be indicative of someone who’s following a healthy diet.7 When evaluating an individual’s health, it’s crucial to look past the biomarker alone and take all relevant aspects of their life and health into account.7
Challenges of Biomarker Validation
Once a biomarker has been discovered, it needs to be validated. However, biomarker validation presents a whole new set of challenges:
Reproducibility
Once a biomarker is identified, researchers must prove it can be consistently used as a biological indicator, that means being able to produce the same results over and over, across repeated tests. Biomarker assays that provide different results in different settings or experiments can lead to inconsistent findings. This highlights the importance of developing assays that can be reproduced.
Standardisation
A lack of standardised protocols for measuring and reporting biomarkers contributes to reproducibility issues. This inconsistency creates difficulties for researchers to compare findings and make conclusions across studies.
Analytical Validation
Although it can be a time-consuming process, without analytical validation researchers may find that biomarkers are used incorrectly which could result in misdiagnosis or inappropriate treatment.
Clinical Relevance
Biomarkers need to have proven clinical significance otherwise they can’t offer any meaningful insights into patient care, which makes them useless in practical applications.
Population Diversity
Biomarkers need to be applicable across diverse populations, so that they are effective for the general public. However, lots of the biomarker research conducted is based on limited, homogeneous groups. To avoid health disparities biomarkers must be inclusive to all populations.
Regulatory Hurdles
Unsurprisingly there are strict regulatory hurdles that researchers and sponsors must adhere to, in order to qualify their biomarkers. Regulatory agencies such as the FDA and EMA, demand strong evidence of biomarker accuracy, reliability and clinical relevance before giving approval. This includes laboratory validation, proof of impact on patient outcomes and demonstration of clinical benefit.
Longitudinal Studies and Economic Considerations
Biomarker validation can require longitudinal studies that often take years to complete. Securing funding for such projects involves ongoing costs over a long period of time.
Integration Challenges
Integrating new tools into clinical workflows presents challenges and requires collaboration between several stakeholders, e.g. researchers, providers and regulatory organisations.8
Improving Biomarker Validation and Discovery
The current progress in biomarker identification is considered to be disappointing, at least when measured by the number of biomarker discoveries that have reached clinical application. Reportedly a contributing factor to this failure is poor study design when conducting biomarker discovery. However, this issue can be solved easily, by us.
Fios Genomics can provide you with study design optimisation, multi-omics integration and advanced statistical analysis. With flexibility across platforms, our team work to identify meaningful biological signals, improve assay reproducibility and ensure clinical relevance. By validating findings we can help to avoid false biomarker leads, and increase the discovery and validation of actionable biomarkers.
Fios Genomics offer a comprehensive range of bioinformatics services. If you would like to learn how we can support your particular goals with bioinformatics, just use the form below to tell us about your research. We’ll then get in touch to let you know the different ways we can support your project.
Sources
1 Paver, E.C. and Morey, A.L. (2023). Biomarkers and biomarker validation: a pathologist’s guide to getting it right. Pathology. doi: https://doi.org/10.1016/j.pathol.2023.11.002.
2 Biology Insights. (2025). What Is Biomarker Detection and How Does It Work? [online] Available at: https://biologyinsights.com/what-is-biomarker-detection-and-how-does-it-work/
3 Prabhakar, P.K. (2025). The Potential of Cancer Biomarkers. Elsevier.
4 Puntmann, V.O. (2009). How-to guide on biomarkers: biomarker definitions, validation and applications with examples from cardiovascular disease. Postgraduate medical journal, [online] 85(1008), pp.538–45. doi: https://doi.org/10.1136/pgmj.2008.073759.
5 Ng, S., Masarone, S., Watson, D. and Barnes, M.R. (2023). The benefits and pitfalls of machine learning for biomarker discovery. Cell and Tissue Research, 394(1), pp.17–31. doi: https://doi.org/10.1007/s00441-023-03816-z.
6 Myadlm.org. (2025). Overlooked and underserved, patients missed for cancer biomarker testing risk worse outcomes. [online] Available at: https://myadlm.org/cln/articles/2025/januaryfebruary/patients-missed-for-cancer-biomarker-testing-risk-worse-outcomes
7 Team, D. (2019). When Biomarkers Can be Misleading. [online] Designs for Health. Available at: https://www.casi.org/node/1164
8 Salib, V. (2025). Understanding biomarker validation, qualification challenges. [online] Pharma Life Sciences. Available at: https://www.techtarget.com/pharmalifesciences/feature/Understanding-biomarker-validation-qualification-challenges.
Also: Frangogiannis, N.G. (2012). Biomarkers: hopes and challenges in the path from discovery to clinical practice. Translational Research, 159(4), pp.197–204. doi: https://doi.org/10.1016/j.trsl.2012.01.023. This was read by the author to inform their general understanding and is not specifically referenced in the article above.
Author: Coco Goto-Colverson, Marketing Intern, Fios Genomics
Reviewed by Fios Genomics bioinformatics experts to ensure accuracy
See Also
Data Analysis for Clinical Development
