Mutagenicity assessment is no longer debated; it is embedded within regulatory science. The more important question today is not whether in silico models should be used, but how confidently and transparently their outputs are interpreted.
Under ICH M7, the mutagenic potential of pharmaceutical impurities must be evaluated using two complementary in silico methodologies, one expert rule-based and one statistical, followed by expert review. Over the past decade, this framework has reshaped impurity hazard characterisation, establishing (Q)SAR as a trusted and practical component of regulatory submissions.
Yet the true value of complementary modelling extends beyond compliance. When applied thoughtfully, it strengthens scientific confidence, supports defensible decision-making, and enhances the transparency regulators increasingly expect.
Why complementary models matter
The principle behind (Q)SAR is straightforward: chemical structure informs biological activity. However, different modelling approaches interrogate that relationship in different ways. At Lhasa, we have two complementary (Q)SAR models:
| Derek Nexus | Sarah Nexus |
|---|---|
|
Expert rule-based system. |
Statistical-based system. |
|
Encodes mechanistic knowledge into structural alerts. |
Learns directly from curated experimental data. |
|
Provides transparency around known toxicophores, |
Identifies structural features associated with activity |
|
chemical reactivity, and established mechanisms linked to a broad |
based on patterns observed in historical studies for |
|
range of toxicity endpoints, including mutagenicity. |
mutagenicity and chromosome damage. |
Each methodology brings distinct strengths, and this complementary approach is precisely why ICH M7 and the draft OECD revised guidance on the definition of residue require both, and why researchers and regulators are now looking at their potential for broader genotoxicity evaluations.
From mutagenicity to broader genotoxicity
Mutagenicity, typically assessed via the Ames test, has long served as a surrogate endpoint for carcinogenic potential in impurity assessment. Regulatory confidence in in silico mutagenicity prediction is supported by decades of accumulated data and mechanistic understanding.
There is increasing interest in broader genotoxicity endpoints, including chromosome damage such as clastogenicity and aneugenicity. These endpoints introduce additional complexity due to varied mechanisms, historical data variability, and differences in assay sensitivity. Nevertheless, the same core principles apply: transparency, high-quality training data, defined applicability domains, and structured expert interpretation remain essential.
Regulatory discussions increasingly recognise the role that in silico approaches can play within a weight-of-evidence framework, particularly where experimental data may be limited or evolving. For pesticide metabolites and impurities assessed in line with expectations from bodies such as the European Food Safety Authority (EFSA), there is growing interest in how complementary (Q)SAR approaches can support early hazard identification and prioritisation. Validation exercises have shown that combining rule-based (Derek Nexus) and statistical (Sarah Nexus) approaches in a consensus framework can:
Improve balanced accuracy
Increase sensitivity for detecting
genotoxic compounds
Maintain specificity, helping to avoid
unnecessary over-classification
For mutagenicity under ICH M7, this translates into greater confidence in both positive findings requiring control and negative classifications supporting higher acceptable limits. For chromosome damage endpoints, where mechanistic diversity and data variability are greater, combining methodologies strengthens the overall weight of evidence.
The role of expert review
It is important to emphasise that in silico tools do not replace expert judgement. Under ICH M7 and the OECD residual definition guidance, expert review is a required step in the assessment process.
The role of the expert is to:
- Confirm applicability domain
- Evaluate concordance between methodologies
- Consider mechanistic plausibility
- Interpret supporting examples and data
- Assign the final impurity classification
Complementary (Q)SAR systems provide structured evidence to inform that review. By presenting mechanistic alerts alongside data-driven hypotheses and supporting examples, they enable clear and defensible scientific reasoning.
The outcome is not simply a prediction, but a documented conclusion.
Applying complementary (Q)SAR in practice
While the principles of complementary modelling are well understood, their value becomes clearer when viewed within real assessment workflows.
ICH M7: Hazard characterisation in context
Under ICH M7, hazard characterisation typically involves:
Identification of actual and potential impurities arising during synthesis or storage
Review of available carcinogenicity and Ames data
In silico mutagenicity assessment using complementary (Q)SAR methodologies
Expert review and impurity classification
Within this workflow, complementary systems such as Derek Nexus and Sarah Nexus support the mutagenicity assessment step by providing:
- Mechanistic evaluation of structural alerts
- Data-driven analysis based on curated Ames datasets
- Transparent supporting examples
- Structured evidence to inform expert review
The software output is not the end of the process. It informs the expert decision that ultimately determines impurity classification and control strategy.
Agrochemicals: In silico assessment of metabolites
A similar structured approach applies in agrochemical assessments:
Identification of potential metabolites from residue metabolism studies
Review of available carcinogenicity and assay data
In silico genotoxicity evaluation using complementary (Q)SAR methodologies
Expert assessment to conclude genotoxic potential
Here, complementary (Q)SAR approaches play an important role in early hazard identification and prioritisation, particularly where experimental data are limited.
By combining mechanistic alerts with statistical evidence for mutagenicity and chromosome damage, toxicologists can build a weight-of-evidence assessment that is transparent and scientifically defensible. This aligns with evolving expectations from regulators such as EFSA 2016 and the OECD Revised Guidance on the Definition of Residue, where justification and documentation are increasingly important.
Supporting confident regulatory decisions
As regulatory science evolves, expectations around transparency and justification continue to grow. Organisations must demonstrate not only that assessments were performed, but that conclusions are scientifically sound and reproducible.
Combining rule-based and statistical approaches helps to:
- Increase confidence in positive findings
- Provide reassurance when assigning negative classifications
- Strengthen weight-of-evidence arguments
- Support consistent documentation for submission
In silico models play a key part in modern genotoxicity assessments, particularly for early screening and prioritisation. Tools like Derek Nexus and Sarah Nexus are enabling:
More confident early
decision-making
Reduction in unnecessary
testing
Increased transparency
and scientific justification
Greater alignment with
regulatory expectations
If you would like to explore how Derek Nexus and Sarah Nexus can strengthen your mutagenicity or genotoxicity assessments, please get in touch.
Request more informationLast Updated on February 23, 2026 by lhasalimited
