Pharmaceutical regulations provide a framework for qualifying non-mutagenic impurities (NMIs) in drug substances and finished products. However, when new impurities emerge or known ones appear at higher levels, especially if they weren’t part of earlier safety studies—it's not always clear what steps to take. To address this, the European Medicines Agency (EMA) has issued a draft reflection paper that proposes a more structured and adaptable strategy for evaluating their potential risks.

This article summarises the paper’s key recommendations, including the use of risk assessment frameworks, approaches to determining acceptable exposure thresholds, and the increasing use of non-animal testing alternatives. We also contrast the EMA’s evolving stance with current US Food and Drug Administration (FDA) guidance.

Overall, these recommendations represent a shift toward a more flexible, evidence-informed approach, one that is particularly valuable when traditional toxicological data are incomplete or unavailable.

Impurity classification

Impurities are defined as any component of the drug product that is not the drug substance or an excipient. These may include residual reagents (e.g., catalysts, ligands), compounds introduced during manufacturing (e.g., leachables from equipment or filters), heavy metals, and degradation products formed through hydrolysis, oxidation, or other chemical reactions.

Impurities are typically categorised as organic impurities, inorganic impurities, or residual solvents. Structural identification can be technically challenging, often requiring multiple orthogonal analytical techniques such as LC–MS/MS, NMR, IR spectroscopy, and chromatographic methods—as discussed in our recent white paper 'Characterisation of non-compendial impurities – what defines ‘Good enough? '

Once identified, impurities must be assessed for toxicological relevance. This includes identifying potential toxicophores—structural features associated with known toxic effects. However, some toxic effects only emerge after metabolic activation, further complicating risk assessment.

Understanding toxicity risk

The reflection paper presents the Threshold of Toxicological Concern (TTC) as a valuable tool for evaluating low-level impurity exposure, particularly when chemical-specific toxicity data are lacking or insufficient for a full risk assessment.

When the structure of an impurity is known, it can be evaluated against TTC values for structurally similar compounds. This enables classification of substances as low, moderate, or high-risk, based on structural alerts and available toxicology data.

Modelling complex exposure scenarios

The paper encourages a robust risk framework for evaluating NMIs, considering both individual substances and the combined effect of multiple impurities. A multifactorial approach is advocated, recognising that toxicity depends on multiple variables, including exposure level, administration route, bioavailability, degradability, and patient specific factors (e.g., age or genetics).

Given this complexity, the use of in silico tools is recommended. Advances in machine learning, AI, and quantitative structure–activity relationship (QSAR) models continue to enhance the predictive capabilities of computational toxicology.

Beyond traditional toxicology

In the absence of conventional data, the draft guidance highlights the growing importance of New Approach Methodologies (NAMs)—including high-throughput in vitro systems, microphysiological systems, read-across approaches, toxicogenomics and metabolomics.

It also supports the integration of physiologically based pharmacokinetic (PBPK) modelling to estimate systemic exposure and support toxicological evaluation.

While animal testing may still be appropriate in some cases, the paper notes limitations in sensitivity, and ethical concerns. Where needed, studies of sufficient duration—such as 28-day repeat-dose protocol—are suggested to support meaningful evaluation.

Determining impurity acceptability

The new draft guidelines propose a structured process for deriving acceptable levels (ALs) of NMIs, based on either single-factor or multifactorial risk assessments, depending on the number and complexity of impurities identified. ALs represent product-specific safe exposure limits and are informed by:

• Permitted daily exposure (PDE) - based on toxicity data and uncertainty factors.

• Bioavailability – which may be similar to the API.

• Read-across (RAX) – using structurally similar compounds to fill data gaps.

• Product-specific considerations – including formulation, pharmacology, or immunogenicity.

Integrating multiple data sources through a weight-of-evidence approach is encouraged to support more informed and reliable impurity risk assessments.

Summary and regulatory take-way: EMA vs FDA

The EMA’s draft reflection paper introduces greater clarity and flexibility for the qualification of NMIs. It explicitly supports the use of NAMs, including computational models and in vitro assays, to reduce reliance on in vivo studies and enable more efficient impurity assessment.

In contrast, while the FDA follows guidance such as ICH Q3A(R2) and Q3B(R2)—which do not exclude alternative methods —these are often interpreted as favouring reliance on traditional toxicology data. Although the FDA supports the use of NAMs, these approaches are generally treated as complementary rather than primary evidence—particularly in the absence of established toxicology data.

Overall, the EMA’s evolving framework offers greater flexibility in incorporating NAMs as part of a science-based, risk-informed approach—providing new opportunities to streamline impurity qualification.

About the Author

Dr Joe Lackey holds a PhD in Molecular Physiology from the University of Dundee, where his research focused on early drug discovery and the role of the PI-3 kinase signalling pathway in tumour growth and development. Over eight years as a Technical Specialist at LGC, Joe has written widely on drug discovery and personalised medicine, while also exploring interests in food and environmental toxicology.