What an Indirect Biomarker Can Tell You
A downstream biomarker can change when a receptor is engaged, but that change is not itself a measurement of receptor occupancy. A drug may bind a receptor and alter a pathway that changes the biomarker; it may also change the same biomarker through another receptor, a downstream enzyme, toxicity, or a shift in the cells contributing to the sample. The practical question is therefore not simply whether the biomarker changed, but whether the evidence makes receptor engagement a convincing explanation for that change.
A direct occupancy assay estimates receptor binding or the fraction of target sites occupied, subject to its assay model and calibration. An indirect biomarker records a functional consequence farther along a proposed causal chain. As established in the earlier material on direct receptor occupancy assays and calibrating occupancy assay signals, direct measurements have their own measurement limitations. Nonetheless, direct occupancy evidence and downstream response evidence answer different questions: binding at the target, and a biological change consistent with target activation or inhibition.
Why a Biomarker Change Is Not Unique Evidence
Let \(\theta\) denote the fraction of the specified receptor occupied by the drug, and let \(Y\) be a downstream biomarker. A deliberately simple model for a change from baseline is
where \(\beta\) summarizes the effect of occupancy on the biomarker over the conditions considered, \(Z\) represents one or more alternative influences, \(\gamma\) summarizes their effects, and \(\varepsilon\) is measurement and biological variation. This is a conceptual decomposition, not a universal linear pharmacology model. The relationship between occupancy and response can be nonlinear, and the sign or size of the response depends on receptor function and the biomarker selected.
The identification problem follows directly: observing \(\Delta Y\) alone does not determine \(\Delta\theta\). For example, a biomarker change of 12 units could arise from \(\beta\Delta\theta=12\) and \(\gamma\Delta Z=0\), or from \(\beta\Delta\theta=0\) and \(\gamma\Delta Z=12\). Many mixtures are also possible. Thus, a biomarker change may be consistent with engagement without uniquely demonstrating it.
Useful evidence comes from comparisons that distinguish these explanations. These include vehicle or baseline controls, a selective competitor of binding, receptor removal or suppression, and a rescue condition that restores the receptor. Such tests are informative only when their own effects are measured and when the interventions are sufficiently selective. A concentration-related biomarker response can add support, but it is not decisive by itself: drug concentration can affect both intended and unintended targets.
Worked Examples: Testing Target Dependence
Worked Example: A Selective Competitor Reverses the Biomarker Response
In a hypothetical cell assay, a downstream phospho-biomarker is reported in normalized units. The mean readings are 100 with vehicle, 160 with the test drug, 102 with a selective receptor competitor alone, and 108 with the test drug plus competitor. Assume the samples were processed comparably and the biomarker assay is within its validated range. Does the competitor result support a receptor-mediated explanation?
First calculate the drug-associated increase relative to its matched vehicle condition:
The drug-plus-competitor condition should be compared with competitor alone, not with vehicle, because the competitor may itself affect the biomarker. The remaining increase in the competitor's presence is \(108-102=6\) normalized units. The amount of the drug response suppressed by the competitor is therefore \(60-6=54\) normalized units. The fraction suppressed is
Under the stated assay assumptions, the large suppression supports the interpretation that most of the measured drug response depends on a competitor-sensitive receptor process. It does not prove that exactly 90% of receptors were occupied, nor that the competitor has no other relevant action. The result is stronger if the competitor is known to act selectively, reaches the target, and does not alter the assay independently.
Worked Example: Receptor Removal Leaves a Residual Response
In a hypothetical experiment, a biomarker is measured in wild-type cells and in cells lacking the receptor of interest. The readings are 20 units for wild-type vehicle, 50 for wild-type drug, 18 for receptor-null vehicle, and 38 for receptor-null drug. Assume the same drug exposure and assay procedure in both cell types. How much of the drug response is associated with the presence of the receptor?
Calculate the drug response within each cell type. In wild-type cells it is \(50-20=30\) units. In receptor-null cells it is \(38-18=20\) units. The difference between these responses is
The drug response is 10 units larger in the wild-type cells than in the receptor-null cells. This difference-in-changes supports a receptor-dependent component under the assumptions that the groups are otherwise comparable and that receptor removal has not created relevant compensatory changes. The residual 20-unit response in receptor-null cells is evidence that the drug can affect the biomarker through at least one receptor-independent route, or that the receptor-null system differs in another relevant way. The result does not quantify occupancy.
Worked Example: A Raw Biomarker Change Is Explained by Cell Loss
In a hypothetical culture experiment, the total biomarker signal in a sample falls from 40 units before treatment to 24 units after treatment. The number of viable cells falls from 100 to 60 in the same units of cell count. Is the raw signal decline evidence that the drug changed biomarker expression per cell?
The raw signal change is \(24-40=-16\) units, a decline of \(16/40=0.400\), or 40.0%. But the mean signal per viable cell is
The signal per viable cell is unchanged in these data. A reduction in the number of cells is sufficient to explain the 40.0% decrease in total signal. Without a suitable normalization or other evidence separating cell loss from a per-cell biomarker response, the raw decline does not support the proposed receptor-mediated change. Even an unchanged per-cell value would not rule out occupancy; it means this particular biomarker measurement provides no evidence for a per-cell response in this comparison.
How to Assess an Indirect Biomarker
A defensible interpretation links the proposed target to the observed measurement while actively checking the competing explanations. The following sequence is useful when planning or reviewing a study.
Name the receptor, the measured biomarker, and the expected direction of change. State whether the proposed response is activation, inhibition, or another defined functional effect.
Use an appropriate baseline or vehicle condition and, where possible, a matched control for each perturbation. Check that sample handling, exposure conditions, and assay range allow the groups to be compared.
Ask whether a selective competitor, receptor suppression, or receptor removal changes the drug-associated biomarker response. Include intervention-only controls so that the intervention's own effect is not mistaken for reversal.
Consider off-target pharmacology, downstream pathway effects, cell number or composition, viability, sample dilution, and assay interference. Normalize to an appropriate measure where the experimental design supports it.
Describe the biomarker as consistent with or supportive of target-dependent action when alternatives have been tested. Reserve claims of receptor occupancy for an appropriate direct occupancy measurement.
Assumptions, Diagnostics, and Limits
The comparisons above rely on assumptions that should be examined rather than silently accepted. A competitor must reach the receptor and affect the relevant binding process; if it has its own effect on the biomarker, that effect needs a matched control. Receptor deletion or suppression must meaningfully reduce the target without introducing changes that dominate the measured response. Assay readings must be in a range where the measurement is interpretable, and differences in cell number, sample quality, or normalization must not account for the pattern.
Several warning signs suggest that an indirect biomarker is being overinterpreted:
- The biomarker changes, but no control establishes that the change exceeds ordinary baseline or assay variation.
- A competitor is described as reversing the response, but its effect alone was not measured.
- The response remains similar after effective receptor removal, yet the entire response is attributed to that receptor.
- A total-sample signal is interpreted as a per-cell change despite treatment-related changes in cell number or composition.
- Only one biomarker is measured, even though the drug could affect it through multiple pathways or directly interfere with the assay.
Alternative approaches answer different parts of the question. A direct receptor occupancy assay is more direct evidence of binding, but does not by itself establish a downstream functional effect. A downstream biomarker can demonstrate a response, but requires specificity evidence to support a receptor-mediated explanation. Selective competition or receptor perturbation tests dependence, but can be limited by incomplete selectivity or adaptation. Using complementary evidence is generally more informative than treating any single measurement as conclusive.
Finally, receptor engagement and biomarker response need not be proportional. Receptor reserve, efficacy, subtype contributions, and pathway context can all affect the relationship between occupancy and response, as covered in the earlier material on occupancy and agonist and antagonist efficacy and receptor reserve. Consequently, a modest biomarker change does not necessarily imply little binding, and a large change does not establish high occupancy. The purpose of an indirect biomarker is to contribute evidence about a proposed mechanism, not to substitute for a binding measurement or to identify every causal step.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- A drug changes a downstream marker, but the marker was not measured in a vehicle group. What key comparison is missing?
- A competitor alone changes the biomarker by 8 units. Why should the drug-plus-competitor condition be compared with competitor alone rather than with vehicle?
- In a receptor-null system, a drug still changes the biomarker by the same amount as in wild-type cells. What conclusion is justified, and what conclusion is not?
- A total biomarker signal falls by 30%, while viable cell number also falls by 30%. What additional calculation or measurement would help evaluate a per-cell response?
Solutions. (1) A suitable vehicle or baseline comparison is needed to determine the drug-associated change under comparable conditions. (2) The competitor's own effect must be accounted for; the matched comparison isolates the drug-associated increment in the competitor's presence. (3) The observed response does not require the receptor under the tested conditions, so the biomarker change alone does not support a receptor-dependent explanation. It does not show that the drug failed to bind the receptor. (4) Normalize the total signal to viable cell number if that normalization is valid, or use another design that separates per-cell signal from cell loss; the raw total alone cannot establish a per-cell change.