Imagine predicting how a chemical will behave in the human body without administering a single dose. This is the promise of PBPK modeling, a revolutionary tool that is reshaping food safety.
For decades, safeguarding our food from harmful chemicals has relied heavily on animal testing. While this approach has offered invaluable protection, it is often time-consuming, costly, and raises ethical questions. Crucially, results from animals do not always translate perfectly to humans. The need for more accurate, efficient, and human-relevant safety assessments has never been greater.
In 2021, a pivotal workshop convened by the UK Food Standards Agency (FSA) and its Committee on Toxicity (COT) explored a powerful solution: Physiologically Based Pharmacokinetic (PBPK) modeling 1 . This sophisticated computer simulation technology creates a "virtual human" to predict the journey of a chemical through the body, offering a groundbreaking shift from traditional methods to a more precise, data-driven future for food safety.
At its core, a PBPK model is a mathematical representation of the human body. It simulates how a substance is Absorbed, Distributed, Metabolized, and Excreted (ADME)—a process known as pharmacokinetics 3 .
Think of it as a sophisticated GPS navigation system for chemicals. Instead of mapping city streets, it charts the pathways of blood flow, organ tissues, and biochemical processes. You can input data about a specific chemical, and the model predicts its path: how much is absorbed from the gut, how it spreads to the liver or brain, how it's broken down by enzymes, and how quickly it's eliminated 2 3 .
It can translate data from cell-based tests (in a petri dish) into predictions of what would happen inside a living human (in vivo) 1 . This helps scientists understand if a concentration that causes harm in a lab dish would ever be reached in a human organ.
For chemicals where human testing is unethical, PBPK modeling provides a scientifically robust way to estimate safe exposure levels, potentially reducing reliance on animal data 1 .
The COT FSA workshop served as a critical forum for regulators, academics, and industry experts to debate the future of this technology. Their conclusions were both optimistic and cautious 1 :
Experts agreed that PBPK modeling is a powerful and applicable tool for refining food safety assessments.
A significant barrier identified was the lack of in-house PBPK expertise within most regulatory bodies, unlike agencies like the U.S. FDA or EMA which have embraced it more fully 1 .
A major takeaway was the urgent need for harmonized international guidance, such as from the OECD, to ensure models are built and used consistently and transparently 1 .
This workshop marked a clear consensus: PBPK modeling is no longer a futuristic concept but an essential component of modern risk assessment, ready to be integrated into the regulatory toolkit.
To understand the power of PBPK, let's examine a real-world application from the pharmaceutical world, which often paves the way for food chemical assessment.
Gemfibrozil (a lipid-lowering drug) and repaglinide (an anti-diabetic drug) are sometimes prescribed together. However, gemfibrozil inhibits enzymes that are crucial for metabolizing repaglinide. When taken simultaneously, repaglinide levels can soar, causing dangerous hypoglycemia (low blood sugar) 3 . Predicting the severity of this interaction is vital for patient safety.
Researchers used a PBPK model to simulate this interaction in a virtual human population through a step-by-step process 2 3 :
They first gathered physiological data (organ sizes, blood flow rates) for a virtual population.
They input the specific physicochemical properties of both gemfibrozil and repaglinide, including how they are metabolized and which enzymes they interact with.
The model was run under two scenarios: repaglinide taken alone, and repaglinide taken with gemfibrozil.
The model's predictions were compared against actual clinical data to ensure its accuracy.
The PBPK simulation successfully predicted a significant increase in repaglinide exposure when co-administered with gemfibrozil. The model quantified this change, showing that the area under the curve (AUC)—a measure of total drug exposure in the body—could increase several-fold.
| Scenario | Simulated AUC (ng*h/mL) | Change vs. Alone |
|---|---|---|
| Repaglinide Alone | 100 | Baseline |
| Repaglinide + Gemfibrozil | 500 | +400% |
This data is critical because it allows regulators and doctors to make informed decisions. Instead of a vague warning of a "potential interaction," the model provides a quantitative estimate of the risk, guiding specific advice like "dose reduction is required" or "co-administration should be avoided" 3 .
Building a reliable PBPK model requires a suite of data and software. The table below details the key "research reagents" or components essential for this virtual science.
| Component | Function | Example in Food Safety |
|---|---|---|
| Physiological Parameters | Defines the virtual body's anatomy and physiology (organ volumes, blood flow rates) 2 . | Using data on human liver size and blood flow to model how a food contaminant is processed. |
| Compound-Specific Data | Describes the chemical's properties (solubility, permeability, protein binding) 4 . | Inputting data on a pesticide's solubility and its affinity for fat tissues. |
| Biochemical Parameters | Quantifies how the body handles the chemical (metabolic rate constants, excretion rates) 2 . | Incorporating data on how quickly a preservative is broken down by liver enzymes. |
| In Vitro to In Vivo Extrapolation (IVIVE) | Translates data from cell-based assays into parameters for the whole-body model 1 2 . | Using liver cell data to predict the metabolism of a new sweetener in humans. |
| Specialized Software | The platform that integrates all data and runs the complex simulations. | Commercial tools like GastroPlus and Simcyp or open-source platforms like PK-Sim 4 . |
The journey of PBPK modeling is just beginning. The 2021 workshop highlighted several exciting frontiers 1 2 :
AI can help predict missing chemical properties, speeding up model development and making it applicable to a wider range of substances.
PBPK models will be linked with toxicodynamic models. This means predicting not just where a chemical goes (kinetics), but also what biological damage it causes at that site.
As called for in the workshop, increased training and familiarization for regulators will break down adoption barriers and build in-house expertise 1 .
| Industry | Primary Application | Regulatory Stance |
|---|---|---|
| Pharmaceuticals | Predicting drug-drug interactions, dosing in special populations 3 . | Well-established acceptance by FDA and EMA 3 4 . |
| Agrochemicals | In vitro to in vivo extrapolation for risk assessment 1 . | Growing acceptance, particularly for refining exposure assessments. |
| Food Safety | Refining exposure assessments, filling data gaps for chemicals with narrow safety margins 1 . | Emerging field, with active discussion and development of guidance. |
The move toward PBPK modeling, as championed by the COT FSA workshop, represents a profound shift in food safety science. It moves us from broad-stroke estimates to nuanced, human-specific predictions. This technology promises not only to make our food supply safer but to do so with greater efficiency and ethical responsibility. The next time you enjoy a meal, know that behind the scenes, sophisticated virtual humans are working to ensure every bite is safe.