Deriving benchmark doses for nano- and microplastic-induced neurotoxicity: A systematic tiered framework for fragmented toxicological evidence

The rapid proliferation of micro- and nanoplastics (MNPs) has intensified concerns about their potential neurotoxicity, yet heterogeneous, fragmented toxicological evidence limits quantitative human health risk assessment. Here, a structured, two-phase framework is presented that integrates systematic evidence synthesis with Bayesian benchmark dose (BMD) modeling to derive defensible toxicological benchmarks from mammalian oral exposure studies. Neurotoxicity is used as a proof-of-concept case study within a broader, scalable approach. A PRISMA-guided review across six databases (45,402 records, including an updated February 2026 search) combined large language model-assisted screening with manual evaluation, yielding 28 relevant neurotoxicity studies and 186 endpoints. After quality screening, 102 endpoints were identified with well-constrained dose-response relationships and further prioritized using a multi-criteria decision-making approach that integrated potency, uncertainty, study quality, and exposure duration. This process established a transparent, tiered evidence hierarchy (Tiers 1-3) and identified 30 high-priority (Tier 1) endpoints for hazard characterization. Across studies, MNP exposure was consistently associated with neurobehavioral, neurochemical, and neuropathological alterations, with effects varying by particle size, polymer type, and exposure duration. The framework is designed to support the synthesis of heterogeneous toxicological evidence, although its applicability to other biological systems, exposure routes, and contaminant classes remains to be demonstrated.