Loading field map
How does a chemical cause harm, and how can that harm be measured, predicted, and prevented? Toxicology has wrestled with these questions since its emergence as a distinct scientific field in the early twentieth century. The answers have evolved through five major frameworks, each responding to the limitations of its predecessors while preserving their most useful tools. Today, these frameworks coexist in productive tension, shaping how researchers study chemical hazards and how regulators decide what is safe.
At the start of the 1900s, toxicology was largely descriptive—a catalogue of which substances were poisonous and what symptoms they produced. The first framework to transform this into a predictive science was Dose-Response Toxicology. Its central insight, often summarized as "the dose makes the poison," was that every chemical could be harmful at some exposure level and that the relationship between dose and effect could be measured mathematically. By plotting the proportion of organisms responding against increasing doses, toxicologists generated curves that revealed thresholds, slopes, and potency comparisons. This quantitative approach gave the field its first rigorous language: the LD50 (the dose lethal to half a test population) became a standard benchmark, and the shape of the dose-response curve itself became a tool for comparing chemicals. Dose-Response Toxicology did not explain why a chemical caused harm, but it provided the empirical backbone that all later frameworks would rely on or react against.
While Dose-Response Toxicology treated all organisms as roughly equivalent responders, Selective Toxicity introduced a fundamentally different perspective. Developed in parallel from the early 1900s, this framework asked why a chemical could be harmful to one species or cell type but harmless to another. Paul Ehrlich’s concept of the "magic bullet"—a drug that targets a pathogen without damaging the host—exemplified this thinking. Selective Toxicity reframed toxicity not as an absolute property of a chemical but as a relationship between the chemical and the biological system it encounters. This led to the therapeutic index, a ratio of toxic to therapeutic doses that became essential in drug development. Where Dose-Response Toxicology sought universal curves, Selective Toxicity emphasized differences: metabolic pathways, receptor distributions, and cellular environments that made one organism vulnerable and another resistant. The two frameworks did not conflict so much as address different questions—the first measuring harm, the second explaining why harm was not evenly distributed.
By the mid-twentieth century, toxicologists had accumulated vast dose-response data and recognized many examples of selective toxicity, but they lacked explanations for how chemicals actually produced their effects. Mechanistic Toxicology emerged around 1950 to fill this gap. It moved beyond whole-organism observation to study the molecular and cellular events that translated a chemical exposure into a biological response. Using techniques such as enzyme assays, radiolabeled tracers, and later gene expression analysis, mechanistic toxicologists traced the pathways from absorption to target interaction to cellular damage. This framework transformed the interpretation of dose-response curves: a shallow slope might now be understood as the result of saturable detoxification enzymes, and a threshold could reflect a repair mechanism that operated below a certain dose. Mechanistic Toxicology did not replace its predecessors but deepened them, providing causal stories for the patterns they had described. It also introduced a new kind of evidence—biochemical markers of exposure and effect—that would later become central to risk assessment.
The 1970s brought a new pressure: society demanded that toxicology inform public health decisions. Regulatory Risk Assessment emerged as a framework designed to translate scientific data into policy-relevant judgments about acceptable exposure levels. It borrowed heavily from Dose-Response Toxicology, using dose-response curves as the foundation for setting reference doses, acceptable daily intakes, and exposure limits. But it also incorporated mechanistic data to extrapolate from high-dose animal studies to low-dose human exposures, and from short-term experiments to lifetime risks. A key debate within this framework concerned the shape of the dose-response curve at low doses: should regulators assume a linear no-threshold (LNT) model, which implies that any exposure carries some risk, or should they allow for a threshold below which no harm occurs? This question remains unresolved, with different regulatory agencies adopting different conventions depending on the type of chemical and the strength of mechanistic evidence. Regulatory Risk Assessment thus became a hybrid framework, combining the quantitative rigor of early toxicology with the biological nuance of mechanistic studies, all within a decision-making context that demanded clear, defensible numbers.
By the turn of the twenty-first century, a growing frustration with reductionist approaches had set in. Mechanistic Toxicology had identified hundreds of pathways and thousands of molecular targets, but it struggled to predict how a chemical would behave in a living organism where multiple pathways interact, feedback loops operate, and individual genetic variation matters. Systems Toxicology, emerging around 2000, responded by treating toxicity as an emergent property of complex biological networks. It uses high-throughput data—transcriptomics, proteomics, metabolomics—combined with computational modeling to map how chemicals perturb entire systems rather than single targets. This framework does not reject mechanistic toxicology but absorbs it into a larger picture: instead of asking which single pathway a chemical activates, systems toxicologists ask how the network of pathways reorganizes in response to exposure. The approach has proven especially valuable for predicting chronic toxicities, such as liver injury or developmental toxicity, that arise from the interplay of multiple mechanisms over time. Systems Toxicology is still maturing, but it has already shifted the field’s ambition from explaining individual effects to modeling whole-organism responses.
Today, all five frameworks remain active, each with a distinct role. Dose-Response Toxicology continues to provide the basic quantitative language for comparing chemical potencies. Selective Toxicity guides the design of safer drugs and pesticides by exploiting biological differences. Mechanistic Toxicology supplies the molecular detail needed to interpret dose-response data and to identify biomarkers for early detection of harm. Regulatory Risk Assessment translates these scientific insights into exposure limits that protect public health, even as it wrestles with methodological disagreements over low-dose extrapolation and the weight of mechanistic versus epidemiological evidence. Systems Toxicology is gaining prominence as a way to address the complexity that older frameworks cannot handle alone, particularly for mixtures of chemicals and for long-latency effects.
The leading disagreements today center on how much mechanistic detail is necessary for reliable risk assessment. Some toxicologists argue that regulatory decisions should remain conservative, relying on simple dose-response models that err on the side of caution. Others contend that mechanistic and systems-level data can refine these models, reducing uncertainty and avoiding unnecessary restrictions on beneficial chemicals. A related tension concerns the role of computational predictions: can a computer model of a biological network replace a whole-animal study, or does it merely generate hypotheses that require experimental confirmation? These debates are not signs of weakness but of a healthy, evolving field. Toxicology has never had a single, settled method for determining chemical safety; its strength lies in the interplay between quantitative measurement, biological insight, mechanistic explanation, regulatory pragmatism, and systems-level integration. Each framework has left its mark on the tools and concepts that toxicologists use today, and the field continues to develop as new technologies and new societal demands reshape the questions it asks.