There’s never been a better time to be interested in biomarkers.
We can measure more, analyse more and connect more types of biological information than ever before. Multiomics can bring together layers of information from genomics and transcriptomics to proteomics and metabolomics, while AI is increasingly being used to make sense of complex biological and clinical datasets.
Exciting? Absolutely!
But there’s a slightly uncomfortable question beneath all this progress… are we sometimes collecting data just because we can?
It’s easy to see why biomarker panels grow. One cytokine becomes ten. Ten evolve into a multiplex panel, and then perhaps there’s exploratory proteomics analysis, some genetic data and a few additional samples collected “just while we’re here”.
Before long, the study has generated an impressive amount of information. But an impressive dataset is not necessarily an informative one.
The real value of a biomarker is not how much data it produces. It’s whether that data helps answer an important biological or clinical question. That’s why biomarker strategy matters.
A good biomarker strategy always starts with a question.
Is the drug interacting with its intended target? Is the proposed mechanism of action supported by the biology? Is there a pharmacodynamic response? Are participants responding differently? Could a biomarker help identify a population, or help us understand treatment response?
These questions ultimately lead to very different biomarker strategies.
Imagine a programme where the main question is whether a new treatment is producing the expected pharmacodynamic response. The answer might lie in a carefully selected set of biomarkers that directly or indirectly reflect the underlying biology. Adding dozens more measurements might generate interesting associations, but that doesn’t necessarily make the evidence any more useful.
It’s not about doing less science. It’s about making sure every measurement has a job to do.
This idea isn’t just a good principle for study design. It’s also reflected in regulatory thinking.
The FDA uses the concept of Context of Use (COU) to describe the specific purpose a biomarker is intended to serve in drug development. Examples include evaluating treatment response, supporting dose selection or enriching a clinical trial for a particular population or event. The intended use helps define the exact evidence needed to support that application.
In other words, before asking “What can this biomarker tell us?”, it’s worth asking “What do we need this biomarker to tell us?”. Sounds obvious, but it’s surprisingly easy to lose sight of the overarching question once the technology becomes interesting.
The industry is moving incredibly quickly. AI and multiomics are opening up opportunities to explore high-dimensional datasets and uncover patterns that would have been difficult to see using individual measurements alone.
But more sophisticated technology also brings a more complex problem: how do you turn all that information into something reproducible, interpretable and clinically useful?
A recent Nature Reviews Drug Discovery perspective makes a similar point about AI in drug discovery, arguing that the focus needs to move beyond what a technology can technically do towards whether it actually improves meaningful development decisions. The same principle applies to biomarker research.
The goal isn’t to win the data race. It’s to get closer to the right answer.
A biomarker that looks promising in an exploratory experiment still has a long journey ahead.
be measured reliably? In the right sample? At the right time point? In the intended population? Can the assay perform for the purpose it’s needed for? And perhaps, most importantly, will the resulting data tell us something useful when it is considered alongside the clinical study?
These questions can sound less exciting than discovering a new biomarker. They’re often the questions that determine whether a biomarker is actually useful.
That’s where biomarker development becomes more than just biomarker discovery. It becomes about connecting the biomarker, sample, assay, timepoint, clinical endpoint and analysis into one strategy.
Maybe the better question isn’t “What can we measure?” it is “What do we need to know?”
Starting there doesn’t mean doing less science; it’s quite the opposite. It means making sure the science has somewhere meaningful to go after.
For sponsors, that can mean a more focused biomarker strategy, a clearer rationale for what is being measured and a better chance of generating evidence that can support the next development decision.
This is where we see the value of bringing scientific, laboratory, clinical and statistical expertise together. At hVIVO, we help sponsors connect biomarker identification and assay development with clinical study design, analysis and the development decisions that follow.
Because in the current biomarker gold rush, the most valuable thing may not be another nugget of data. It’s knowing which ones are worth digging for.