QSAR Modelling for Pesticide Registration

Quantitative structure-activity relationship (QSAR) models are theoretical models that can be used to quantitatively or qualitatively predict the physicochemical, biological (e.g. an (eco) toxicological endpoint) of a chemical substance based on its chemical structure. In the high-stakes environment of pesticide registration, where the safety of active ingredients, metabolites and impurities must be demonstrated to protect human health and the environment, QSAR provides a robust, ethical, and cost-effective alternative to traditional in vivo animal testing. CIRS provides full-range QSAR modelling services to support your pesticide registration and hazard screening. 

Regulatory Acceptance of QSAR Modelling in the European Union (EU)

In the EU, QSAR modelling is often used in Technical Equivalence assessment. The assessment is divided into two tiers: Tier I or for Tier II.

In Tier I asessment, substance identity, chemical composition, analytical profile of five representative batches, description of the analytical method used for the determination of the substance are assessed. 

If technical equivalence cannot be established based in Tier I due to the presence of significant impurities present in quantities ≥ 1 g/kg or relevant impurities, the applicant may request authorities to perform a Tier II assessment. Laboratory studies or QSAR modelling reports can be used as supporting data to assess the hazard profiles of impurities. 

In addition, the European Food Safety Authority (EFSA) utilizes QSAR extensively for the assessment of pesticide metabolites in food and groundwater. If a pesticide degrades into a metabolite that was not tested in the original animal studies, EFSA allows the use of QSAR to screen for genotoxicity and other hazardous properties.  This is often part of an Integrated Approach to Testing and Assessment (IATA), where QSAR results are combined with the Threshold of Toxicological Concern (TTC) to determine if a metabolite poses a risk to consumers.

Regulatory Acceptance in the United States (USA)

The U.S. Environmental Protection Agency (EPA) manages pesticide registrations under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA). The EPA has long championed the use of QSAR, particularly for predicting the environmental impact of chemicals on non-target species.

In the USA, QSAR is part of the IATA framework, where structural alerts and predictions are integrated to streamline data requirements. If QSAR models show that a new pesticide is structurally similar to a well-studied compound with low toxicity, the EPA may grant waivers for certain animal studies, accelerating the registration process.

Regulatory Acceptance of QSAR Modelling in Brazil

Brazil is one of the world's largest pesticide markets and maintains a robust regulatory framework overseen by MAPA (Agriculture), ANVISA (Health), and IBAMA (Environment).

In December 2025, ANVISA published Resolution RDC No. 1.006, which formally introduced a tiered toxicological assessment system for Technical Grade Active Ingredient (TGAI) equivalence. The main requirement for QSAR modelling is listed as follows:

●      Multi-System Verification: Registrants are now explicitly required to provide independent predictions from at least 3 different expert systems.

●      Tier II Equivalence: If the impurity profile of a technical product differs from the reference product, a QSAR report is mandatory. This report must provide a definitive conclusion on the toxicological relevance of impurities, particularly regarding mutagenic or carcinogenic potential.

●      Data Transparency: Submission of raw data and a signed "Declaration of Toxicological Relevance of Impurities" (DRTI) are now mandatory for registration dossiers.

Open-Source and Commercial QSAR Modelling Tools

Several specialized software tools have become standard in the global pesticide registration process, ranging from open-source models to advanced commercial expert systems.

●      OECD QSAR Toolbox: The preeminent software for regulatory gap-filling.  It allows users to group chemicals into categories and use read-across techniques to predict missing data. Its vast database, containing over 3 million data points, is essential for assessing pesticide metabolites and impurities.

●      EPA Tools (EPI Suite and ECOSAR): ECOSAR specifically estimates the toxicity of chemicals to aquatic organisms like fish, daphnia, and algae. These programs are foundational for screening-level assessments in the United States.

●      VEGA and T.E.S.T.: VEGA HUB provides a platform for multiple models, focusing on the Applicability Domain Index (ADI) to evaluate prediction reliability.The Toxicity Estimation Software Tool (T.E.S.T.) uses a consensus approach, averaging predictions from different models to increase stability.

In high-tier regulatory submissions, particularly for genotoxicity and mutagenicity, commercial platforms developed by Lhasa Limited are frequently employed to provide "complementary" methodologies.

●      Derek Nexus: This is an expert rule-based system that draws on over 40 years of research to provide manually curated structure-activity relationships. It identifies toxicophores—specific structural regions associated with hazards—and provides mechanistic rationales for endpoints like mutagenicity, carcinogenicity, and skin sensitization.

●      Sarah Nexus: This is a statistical-based tool that utilizes a machine-learned self-organizing hypothesis network (SOHN). It learns directly from curated experimental data to generate fragment-based structural hypotheses, providing a data-driven prediction of mutagenicity and chromosome damage.

The integration of these two systems is often required by regulatory frameworks such as ICH M7, which dictates that two different modeling methodologies (one rule-based and one statistical) should be applied to maximize the sensitivity and reliability of mutagenicity assessments. This approach has become standard for the evaluation of pesticide metabolites and residue definitions in accordance with EFSA guidance.

The OECD Principles for Model Validation and Regulatory Acceptance

The global adoption of QSAR in pesticide registration is underpinned by a set of five validation principles established by the Organization for Economic Cooperation and Development (OECD) in 2004. These principles were designed to ensure that in silico predictions are sufficiently reliable for legal and safety decisions.18

Principle 1: A Defined Endpoint

A model must predict a specific toxicological or physicochemical result that is relevant to the regulatory assessment.

Principle 2: An Unambiguous Algorithm

The internal logic of the model must be transparent and reproducible. Regulators require documentation of how descriptors are calculated and how they are mathematically combined to produce a prediction. This transparency is essential for the legal defensibility of the registration decision.

Principle 3: A Defined Domain of Applicability (AD)

One of the most critical aspects of QSAR is understanding its limitations. A model is only valid for chemicals similar to those in its training set. The Applicability Domain defines the structural and property boundaries within which the model’s predictions are reliable. If a pesticide metabolite falls outside this domain, the prediction is considered unreliable, and traditional experimental testing may be required.

Principle 4: Statistical Evaluation of Goodness-of-Fit and Predictivity

Models must undergo rigorous statistical testing to prove they work.1 Internal validation, such as "leave-one-out" cross-validation, checks the model's consistency, while external validation—using a set of chemicals the model has never seen—proves its true predictive power.

Principle 5: Mechanistic Interpretation

Where possible, the model should explain the biological or chemical reason for its prediction. For instance, if a model predicts that a pesticide will be a skin sensitizer because it contains an electrophilic group, this provides a mechanistic rationale that aligns with toxicological theory.This alignment significantly increases the confidence of regulatory reviewers.

Our QSAR Modelling Services

Quantitative Structure-Activity Relationships have transitioned from a specialized tool in computational chemistry to an indispensable foundation of modern pesticide registration.

Our experienced team and toxicologists at CIRS Group can prepare QSAR reports that are accepted for regulatory use to help you expedite pesticide registration and save costs.

Our services include:

• Hazard screening of new compounds or impurities using QSAR
• QSAR modelling using both rule-based and statistical models
• Preparation of QSAR reports for regulatory use

For more information on how we can assist with your pesticide registration needs in EU, contact us today via Email: service@cirs-group.com.