Explore Our Blog

Exposure-Response Part II: Measuring Exposure and Response

This is Part II of our exposure-response blog series. In our previous post, we introduced the FDA Guidance for Industry, “Exposure-Response Relationships — Study Design, Data Analysis, and Regulatory Applications,” provided some insight into approaches to understand exposure-response relationships, and offered a few points to consider when designing exposure-response studies. In the current post, we will discuss study design in more detail, paying particular attention to approaches for measuring systemic exposure and response.

The Many vs. The One

The exposure-response guidance allows for both individual and population dose ranging (population studies are more common). Population designs in which each subject receives a single dose level are especially useful in later stage studies where the pharmacodynamic (PD) effect is delayed, for example, when looking at factors such as hemoglobin A1c, viral load, or tumor reduction.

With individual dose ranging, each participant receives a range of doses (i.e., a crossover design), which provides information on the distribution of individual exposure-response relationships and permits a better estimation of variability in the half maximal effective concentration (EC50) and maximum response (Emax). Using such crossover designs requires consideration of sequence and duration of dosing, as there is a potential for sequence and carryover effects.

Measuring Exposure

Exposure is nearly always related to measurements of active drug moieties, but measuring other molecules, such as optical enantiomers, complex mixtures, endogenous ligands, or determining the level of unbound drug may also be appropriate in some situations, albeit less commonly. There are a number of ways to represent exposure and the choice of which to use is ultimately dependent upon the study objectives, study design, and the nature of the exposure-response relationship.

If a single PD response is measured per day, then simplified exposure metrics such as AUC (area under the drug concentration-time curve), Cmax (peak concentration), or Cmin (trough concentration) are generally appropriate. However, sorting through which of these exposure measures is the “best” (i.e., the most predictive) can be difficult, especially given the high correlation between them. AUC can be used to compare exposure after multiple doses to that following a single dose and is especially useful when correlated at steady state to effects related to long-term exposure (e.g., cancer). Cmax is useful when examining rapid-onset endpoints such as toxicity (e.g., nausea, cardiac events), but due to the potential for large interindividual variability in time to peak concentration, an appropriately thorough PK sampling schedule is critical and consideration must be given to expected differences in PK profiles due to demographics, disease states, and food effects. Cmin (measured just before administration of the next dose) is a less commonly used exposure parameter to relate to response, but is useful in studies where collection of multiple plasma samples over a dosing interval is not practical, when a drug acts slowly relative to the rates of absorption, distribution, and elimination, or when the objective is to determine the lowest effective dose (e.g., antibiotics). In contrast to Cmax, Cmin does not reflect drug absorption processes and is often proportional to AUC.

In cases where response varies substantially with time within a dosing interval, measuring the concentration of active moieties over time (i.e., rich PK sampling) provides detailed, time-dependent exposure information (information that cannot be derived by AUC or Cmin) that can be correlated to the observed response. This approach allows calculation of AUC but also the determination of concentration-time profiles over a dosing interval for each individual and the study population.

When detailed PK information is required but rich PK sampling is not feasible from a practical or ethical standpoint (e.g., special population studies involving pediatric or geriatric subjects), sparse PK sampling (only a few samples per subject, taken at randomly selected times or at prespecified but different times) can offer a viable solution. Population PK analysis combined with Bayesian estimation methods can be used to approximate population and individual PK parameters, generating an exposure variable that can be correlated to response.

Measuring Response

Both positive (efficacy) and negative (safety) responses can be characterized using various measures or endpoints, including clinical outcomes (i.e., direct determinations of clinical benefit or toxicity), biomarkers, and surrogate markers. Biomarkers are measurements such as blood pressure, cholesterol, viral load, and magnetic resonance imaging [MRI] measures, among others, that reflect the activity of a disease process and generally quantitatively correlate with disease progression. When biomarkers are accepted for use in clinical trials as a substitute for clinically meaningful endpoints (i.e., as predictors of therapeutic effect), they are classified as surrogate markers (e.g., blood glucose, C-peptide, Alzheimer’s plaques). Somewhat surprisingly, there are very few validated surrogates in existence, as they each must demonstrate a consistent pattern of correlation to clinical effect across multiple drugs.


It can be challenging to convince stakeholders that understanding exposure-response relationships early in drug development is a wise investment. Usually, the initial focus is on providing an early demonstration of efficacy, with concerns about determining the optimal dose, dosing interval, or any adjustments that need to be made for special populations being put off until later in the development cycle. But, the history of drug development is full of drugs that have failed, not because they fundamentally did not work, but because the opportunity to understand the exposure-response relationship in early development was missed in the rush to get to registration trials.

Nuventra has a great deal of experience in helping design and analyze trials that allow the understanding of exposure-response relationships with little to no impact on development timelines. Critically, these design decisions need to made as early as possible in the development program, as it becomes increasingly difficult and sometimes nearly impossible to salvage a program that has inadequate exposure-response data in the final lead-up to filing the marketing application.

Contact a senior consultant about how to effectively incorporate exposure-response into your development program