Loading field map
How does the body process a drug? From the moment a tablet is swallowed or an injection given, the drug embarks on a journey: it is absorbed into the bloodstream, distributed to tissues, metabolized by organs, and eventually excreted. These four processes—absorption, distribution, metabolism, and excretion, collectively known as ADME—determine the concentration of the drug at its site of action over time. Predicting that concentration curve is the central challenge of pharmacokinetics. For much of the twentieth century, the field has wrestled with a fundamental tension: how to build mathematical models that are simple enough to be useful yet realistic enough to be trustworthy. The history of pharmacokinetic frameworks is a story of successive attempts to balance empirical convenience against physiological fidelity.
The first systematic attempt to turn ADME into a quantitative science came with compartmental modeling. In the 1930s, the Swedish physiologist Torsten Teorell published a series of papers that laid the foundation for this approach. His insight was to treat the body as a system of interconnected compartments—typically a central compartment representing blood and well-perfused organs, and one or more peripheral compartments representing tissues where the drug distributes more slowly. The movement of drug between compartments was described by differential equations, with rate constants that captured how quickly the drug entered and left each space.
Compartmental modeling was a breakthrough because it gave researchers a mathematical language to describe drug behavior. A two-compartment model, for example, could explain why a drug's concentration in blood first drops rapidly (distribution into tissues) and then declines more slowly (elimination). The framework became the workhorse of early pharmacokinetics, used to estimate key parameters such as half-life, clearance, and volume of distribution. Yet its power came with a cost: the compartments were abstract, not anatomical. A "peripheral compartment" might lump together muscle, fat, and bone, even though those tissues handle drugs very differently. The model fit the data, but it did not necessarily reflect physiology.
By the 1970s, the limitations of compartmental modeling had become apparent. Choosing the wrong number of compartments could distort parameter estimates, and the assumption that drug movement followed first-order kinetics did not always hold. Researchers needed a method that did not force the data into a preconceived compartmental structure. Noncompartmental analysis (NCA) emerged as that alternative.
NCA relies on statistical moment theory rather than differential equations. It calculates key pharmacokinetic parameters—such as area under the concentration-time curve (AUC), mean residence time, and clearance—directly from the observed data, without assuming a specific compartmental model. This model-independence made NCA especially attractive for regulatory applications. When two formulations of a drug are compared for bioequivalence, for instance, regulators require NCA because it provides a standardized, assumption-light way to assess whether the products behave similarly in the body. Today, NCA remains the default method for many routine pharmacokinetic analyses, particularly in industry and clinical settings where speed and reproducibility matter more than mechanistic insight.
Compartmental modeling and NCA both shared a limitation: they treated each individual as a separate experiment. A typical study would involve a small number of subjects, each sampled intensively, and the results would describe that specific group. But clinicians and drug developers wanted to understand variability across a broader population—why some patients clear a drug faster, or why kidney disease alters exposure. Population pharmacokinetics (PopPK), pioneered by Lewis Sheiner and Stuart Beal in the 1980s, addressed this question directly.
PopPK uses nonlinear mixed-effects modeling to analyze data from many individuals simultaneously, even when each individual contributes only a few blood samples. The framework separates the observed variability into two components: fixed effects (such as age, weight, or kidney function) and random effects (unexplained inter-individual and residual variability). Crucially, PopPK often uses compartmental models as its structural backbone—the same differential equations from the 1930s—but now embedded in a statistical framework that can handle sparse, unbalanced clinical data. This marriage of compartmental structure with population-level statistics transformed pharmacokinetics from a descriptive tool into a predictive one. PopPK became central to model-informed drug development, allowing sponsors to simulate clinical trials, optimize dosing regimens, and support regulatory decisions with quantitative reasoning.
Despite the advances of PopPK, a gap remained. Neither compartmental models nor NCA nor PopPK could predict how a drug would behave in a new scenario—say, in a patient with liver disease, or in a child whose organs are still developing—without first collecting data in that population. Physiologically based pharmacokinetic (PBPK) modeling, which gained traction in the 1990s, aimed to close that gap by grounding the model in real anatomy and physiology.
PBPK models represent the body as a network of organ compartments—liver, kidney, lung, brain, muscle, fat—each with a realistic blood flow rate, tissue volume, and drug-binding properties. Instead of fitting rate constants to observed data, PBPK models use independent physiological and biochemical parameters: organ blood flows from the literature, tissue partition coefficients from in vitro experiments, and metabolic clearance from enzyme kinetics. This mechanistic foundation allows PBPK to extrapolate beyond the conditions in which it was built. A PBPK model validated in adults can be scaled to predict pediatric dosing, or to forecast drug-drug interactions before a clinical study is run. Regulatory agencies now routinely accept PBPK simulations in lieu of dedicated clinical trials for certain questions, such as evaluating the impact of renal impairment on drug exposure.
Today, all four frameworks remain active, and their division of labor reflects their distinct strengths. NCA is the workhorse of bioequivalence testing and early-phase clinical trials, prized for its simplicity and regulatory acceptance. Compartmental modeling, though less dominant than it once was, still serves as the structural core of many PopPK analyses. PopPK has become the standard for analyzing sparse clinical data, quantifying variability, and designing optimal dosing regimens. PBPK, meanwhile, is the framework of choice for mechanistic prediction and extrapolation, especially in special populations and drug-drug interaction assessment.
The leading frameworks today—PopPK and PBPK—agree on a fundamental principle: pharmacokinetic models should be quantitative, data-driven, and validated against observed outcomes. They disagree, however, on how much mechanistic detail is necessary. PopPK practitioners argue that a well-chosen compartmental model, combined with careful covariate analysis, is sufficient for most drug development decisions. PBPK advocates counter that only a physiologically realistic model can reliably predict behavior in untested scenarios. In practice, the two approaches are increasingly integrated: PBPK models can inform the structural priors for PopPK analyses, and PopPK data can refine PBPK parameter estimates. This hybrid strategy, sometimes called "middle-out" modeling, represents the current frontier of the field.
What unites all four frameworks is the recognition that no single model captures the full complexity of the human body. Each framework makes trade-offs—between simplicity and realism, between empirical fit and mechanistic understanding, between individual precision and population generality. The history of pharmacokinetic modeling is not a story of one framework replacing another, but of a field gradually learning to match the right tool to the right question. For a student entering the discipline today, the challenge is not to choose one framework, but to understand how they complement each other in the ongoing effort to predict what a drug will do inside the body.