Frustra laborare? Methodological inefficiencies and epistemic limits in toxicological testing

Toxicological testing seeks to enable robust regulatory decisions while reducing uncertainty. Advances in modern biomedical technologies now allow the generation of data at an unprecedented scale and granularity. However, the proliferation of assays and datasets does not in itself translate into decision-relevant knowledge. This Comment article examines when testing turns into a case of frustra laborare - i.e. expending effort without proportional epistemic gain. We argue that exposure realism and toxicokinetic plausibility function as gatekeepers of relevance: mechanistic effects observed only at implausible internal concentrations rarely alter risk characterisation. Likewise, the act of expanding batteries of new approach methodologies (NAMs) without establishing predefined integration rules, inflates false-positive accumulation and amplifies uncertainty. Reproducibility limits and borderline outcomes are structural features of both animal and non-animal methods, and these factors must be quantified rather than obscured. A value-of-information (VoI) view brings much-needed discipline: the value of a test lies in its ability to shift decisions; those that cannot do so, do not meaningfully contribute. Across all domains discussed, reproducibility stands out as the essential constraint - exposure anchoring, probabilistic integration and uncertainty quantification all depend on assays that perform consistently across laboratories. In order for toxicologists to avoid frustra laborare, testing should be aligned with internal exposure, reproducible, and embedded within decision frameworks that yield actionable outcomes.