Population pharmacokinetic and pharmacokinetic/pharmacodynamic modeling has become an essential component of modern drug development. These methods integrate drug concentration data, patient characteristics, clinical outcomes, and prior biological knowledge to quantify how drug exposure and response vary across patients, and to predict their responses to various dosing regimens.
Despite their widespread use, population PK, PK/PD, and exposure–response (E-R) analyses are still sometimes viewed as specialized analyses conducted after a clinical trial is complete. In reality, their greatest value emerges when they are integrated prospectively into the drug development strategy and used iteratively to answer specific questions.
There are numerous examples in the scientific literature that illustrate applications of population analyses. In the development of sotatercept, a population PK approach showcased its favorable PK characteristics (low to moderate variability and typical bioavailability), which supported it as a SC biological agent for the treatment of patients with pulmonary arterial hypertension (PAH). In a second step, a population PK/PD analysis for a safety endpoint, together with exposure-response analyses for efficacy endpoints, confirmed the optimal sotatercept dosing schedule for balancing efficacy and safety in participants with PAH. Another common practical application of population PK models involves leveraging adult population PK models to predict pediatric dosing regimens. A recent manuscript shows strategies for pediatric dose derivation from population PK models. In another example, an exposure-response approach was used for a study in patients with metastatic renal cell carcinoma, to better define the optimal thresholds of cabozantinib exposure to improve the safety/efficacy balance for patients in a real-world setting.
What population PK modeling contributes
Population PK modeling is a quantitative method that analyzes concentration data using nonlinear mixed-effects models, to:
- Estimate typical PK parameters, such as clearance and volume of distribution in order to characterize drug exposure in the population of interest
- Quantify between-subject and, when feasible, between-occasion variability in PK parameters and describe residual unexplained variability
- Discern and quantify the influence of intrinsic and extrinsic factors on drug exposure
- Derive individual exposure measures (e.g. AUC, , and trough concentrations) for use in exploring relationships between drug exposure and efficacy and safety responses
A major practical advantage of population PK modeling is the ability to combine rich and sparse data. Samples from late stage trials can thus inform pharmacokinetics without requiring every patient to undergo intensive serial blood sampling. This is an important difference from the traditional non-compartmental PK analysis approach, which generally relies on rich PK sampling from each participant to derive individual exposure metrics.
However, the applications of population PK analyses extend well beyond the mere description of PK. The FDA’s population PK guidance recognizes these analyses as an important means of identifying sources of exposure variability and informing individualized dosing recommendations. https://www.fda.gov/media/128793/download?utm_source=chatgpt.com
Population PK modeling can be used to:
- Define dosing regimens in first-in-human trials, translating from preclinical studies
- Provide guidance for optimal sampling and starting dose in early phase clinical trials
- Predict real-time exposures to adjust study design as a study is in progress (e.g. adding or removing cohorts from a study, or changing the dosing regimen); or to inform a Data Safety Monitoring Committee
- Describe PK variability
- Support formulation development
- Justify dose level and dosing regimen selection for subsequent clinical trials
- Evaluate the impact of anti-drug antibodies on PK
- Plan a more efficient study or even eliminate the need for a clinical study, thus minimizing study duration and development program costs
- Guide pediatric development by suggesting pediatric dosing regimens and minimizing pediatric study sample size
- Characterize PK in specific populations
- Evaluate the potential for drug–drug interactions
- Derive exposures to conduct exposure-response analyses
- Assess the probability of success for a given study design
Moving from PK to PK/PD
Population PK describes what the body does to a drug. Population PK/PD modeling then links drug concentrations or exposure parameters to the response; i.e., effectively describing what the drug does to the biological system. The response could be a biomarker, clinical endpoint, safety measure, microbial burden, tumor growth trajectory, or time-to-event outcome. Depending on the underlying biology and available endpoint data, the relationship may be represented by one or a combination of different model structures, including but not limited to:
- Direct effect
- Effect-compartment (biophase)
- Indirect response
- Disease progression
- Transit compartment
- Tumor growth inhibition
- Parametric or semiparametric time-to-event
- Landmark or longitudinal logistic regression
- Proportional Odds
- Count-Data
- Joint longitudinal and survival models
For PK/PD models, mechanistic complexity is not required. The best PK/PD model is the most parsimonious (simplest) model that adequately represents the PK and clinical endpoint data. An empirical exposure–response model may be entirely appropriate for dose selection, while a more mechanistic model may be needed to extrapolate beyond the observed population or dosing regimen.
PK/PD models are particularly useful for separating the timing of exposure from the timing of response in cases where the biological response is delayed because of receptor turnover, physiological feedback, or progression through a biological cascade. In this case, a longitudinal PK/PD model would be an appropriate approach.
Population PK/PD modeling can be used to:
- Describe PK/PD variability
- Evaluate the effect of intrinsic and extrinsic factors on clinical endpoints
- Facilitate choices between similar drug candidates at the preclinical stage
- Evaluate the effect of anti-drug antibodies on clinical outcomes
- Incorporate disease progression to achieve a more refined prediction of outcomes
- Make rational dose decision recommendations to inform regulatory submissions
- Predict clinical outcomes based on PK-model derived exposures via an ER approach
- Simulate clinical outcomes for efficacy and safety endpoints for different dosing regimens to inform future Phase 2/ Phase 3 study designs
- Simulate clinical trial outcomes to assess the probability of success and optimize study design features
Exposure–response analysis: more than plotting outcomes by exposure quartile
Exposure–response analysis evaluates how efficacy or safety changes across the range of drug exposures. When properly designed, it helps determine whether a selected dose is near the plateau of efficacy, whether lower exposures may compromise benefit, and whether higher exposures would increase the risk of toxicity.
Exposure-response modeling can be used to support:
- Phase II and Phase III dose selection
- Benefit–risk assessment
- Confirmation of a proposed commercial regimen
- Evaluation of fixed versus weight-based dosing
- Assessment of alternative dosing intervals
- Dose modification or titration strategies
- Identification of clinically relevant subgroups
If you would like to discuss how population modeling can support your development program, schedule a call with one of our experts.
You can also learn more about physiologically-based pharmacokinetic (PBPK) modeling in this previous post from my colleague, Neil Miller.