A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics
Open accessThe DOI matches its Crossref record; 81 of the 81 references it cites were found in a registry or an index; 43 of the 89 sources its claims cite were checked for support.
Abstract
Translating preclinical safety findings into reliable insights for human risk assessment remains a fundamental challenge in drug development. Prior preclinical-clinical concordance studies have been constrained by limited drug coverage, reliance on identical-term matching for adverse events (AEs), and limited consideration of species, modality, exposure, and biological or mechanistic context. To address these gaps, we assembled a large cross-species concordance dataset, integrating standardized preclinical and clinical safety data for 7,565 marketed and investigational drugs from PharmaPendium and OFF-X. Our framework employs likelihood ratios to reduce prevalence bias and extends concordance assessment beyond identical-term matches to include semantically and mechanistically related AE pairs. Stratified analyses by species, modality, and exposure-matched subsets further refined translational relevance, while integration of on- and off-target annotations supports mechanistic interpretation and potential screening. Using this approach, we identified 850 significant identical-term AEs and 2,833 additional unique endpoints from cross-term associations. To promote reproducibility and transparency in animal research, we provide open access to the analytic code and statistical results via an interactive web application. An accompanying multi-agent AI system (ToxAgents) enables standardized querying and interpretation of concordance results. Together, these resources extend previous foundational efforts and establish a shared, data-driven platform to advance translational safety science, support evidence-based study design aligned with the 3Rs principles, and ultimately contribute to the development of safer medicines to improve human health.
Overview
CoChat Summary
TL;DR
This study assembled a harmonized cross-species safety dataset for 7,565 marketed and investigational drugs from PharmaPendium and OFF-X and used likelihood ratios, species/modality/exposure stratification, and cross-term mapping to assess preclinical-to-human toxicity concordance. It identified 850 significant identical-term adverse events and 2,833 additional unique endpoints, with generally low-to-moderate organ-system concordance but stronger phenotype-level signals, and released a web app plus the ToxAgents multi-agent AI system.
Key findings
- Across all modalities, 850 significant identical-term phenotypes were identified, including 780 for small molecules, 237 for biologics, and 94 in the exposure-controlled subset; cross-term analysis added 2,833 unique endpoints (2,375 clinical, 854 preclinical, 396 shared).
- At the SOC level, “infections and infestations” had the highest predictivity (LR+ >10) for both small molecules and biologics, followed by immune, renal/urinary, and musculoskeletal disorders with moderate LR+ <5; nervous system, psychiatric, investigations, general disorders, and pregnancy/perinatal conditions showed low LR+. Monkeys had higher median LR+ (SM 4.18, Bio 3.74) than other species, and most iLR– values were close to 1.
- At the PT level, small-molecule examples included constipation (LR+ = 21.5), rash (20.0), neutropenia (16.1), and anemia (10.1); biologic examples included infection (116.0), rash (45.2), hyperglycemia (34.4), and neutropenia (21.8). In hepatobiliary/metabolic analyses, immune-mediated hepatitis (321.8), hyperinsulinemia (217.7), and hypertriglyceridemia (113.1) were highly concordant, whereas nonspecific terms such as hepatotoxicity (2.2), hepatic function abnormal (4.2), and liver disorder (3.2) were poor.
- Compared with Giblin et al., 183 of 188 (97%) significant AEs were also significant in this dataset, with Pearson r = 0.61–0.88 across PPV, NPV, LR+, and iLR–; 70% of AEs with LR+ ≥10 in the prior study also met that threshold here. Cross-term semantic similarity quartiles showed median LR+ increasing from 5.6 to 6.7 (Kruskal–Wallis p = 1.8 × 10−100).
Methods
- Data were compiled from PharmaPendium and OFF-X, with AEs coded in MedDRA; 13,949 unique AE terms were retrieved, then filtered to 7,565 drugs with at least one AE in both preclinical and clinical studies. Preclinical species were mouse, rat, rabbit, dog, and monkey; drugs were categorized as small molecule or biologic. Exposure-controlled comparisons required |Δlog10Cmax| or |Δlog10AUC| ≤1.
- Concordance used 2 × 2 contingency tables, two-tailed Fisher’s Exact Test with Bonferroni correction within each SOC (adjusted p < 0.05), and likelihood ratios (LR+, iLR–) plus PPV and NPV. Identical-term and non-identical within-SOC AE pairs were assessed; OpenAI text-embedding-3 cosine similarity quantified semantic relatedness.
- Mechanistic annotation used DrugBank and OFF-X on/off-targets, gprofiler with DisGeNET, STRING PPI networks, and MCODE. Results were made accessible via a Shiny web app and ToxAgents, a GPT-4.1/RAG multi-agent system using 26 ICH safety documents, with case studies and reproducibility/robustness checks.
Limitations
- The paper states that source-data assignment of MedDRA terms, especially for preclinical behaviors or nuanced histopathology, may be subjective or inconsistent; missing AEs due to incomplete/selective reporting or extraction failure may mask translational signals, and OFF-X investigational-drug AEs may be neutral or lack severity because companies may not disclose more impactful findings.
- The paper states that AEs are encoded as binary present/absent outcomes without frequency, severity, or explicit dose-response data, limiting use for clinical risk thresholds; biologics were not subtyped because current sources lack consistent subtype labels.
- The paper states that target annotation can be confounded by complex pharmacology, multiple on- and off-targets, metabolites, or uncharacterized off-target interactions; it also notes that true negative counts are estimations because not all possible observations were measured or reported.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures







Verification
Claims the paper makes
30 Valid · 3 Not valid · 23 Not checked · 160 Own results and interpretations
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Advancing drug safety assessment relies on bridging the translation gap between animal toxicology findings and clinical outcomes, a challenge with far-reaching consequences for R&D efficiency, regulatory success and patient safety (Waring et al., 2015; Amorim et al., 2024).”
§ A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics > Introduction
What the sources say
- Not checked — no full text of the sourcecites Waring et al., 2015 source
- Supportedcites Amorim et al., 2024 source
The Introduction states that extrapolating animal data to humans is not always straightforward because of species differences, and reports limited agreement between animal-model toxicity predictions and human outcomes. It also links toxicity prediction to drug-development costs and safety, and notes that drugs must meet regulatory safety standards. No page numbers are provided in our copy. (LLM, conf=0.93)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“The absence of pharmacokinetic exposure control further confounds interpretation, as toxicities observed in animal models, often at high doses, may not accurately represent risks relevant to clinical settings with relevant efficacious doses (Van Norman, 2019).”
§ A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics > Introduction
What the sources say
- Partly supportedcites Van Norman, 2019 sourceExperimental — accuracy not yet measured
The Historical Perspective section reports poor concordance between animal and human toxicity findings. But our copy does not identify a lack of pharmacokinetic exposure control as the confounder or say that toxicities seen in animals at high doses may be irrelevant at clinically efficacious doses. The Price of Wrong Decisions instead gives examples of animals appearing safe at doses far above those that harmed humans; no page numbers were provided. (LLM, conf=0.83)
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“which may relate to differences in immune status, microbial exposure, and the scopes of behavioral assessments between animals and humans (Dinan and Cryan, 2017; Huggins et al., 2019; Burger et al., 2023).”
§ A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics > Results and discussions > SOC-based AE frequency and concordance analysis
What the sources say
- Partly supportedcites Dinan and Cryan, 2017 sourceExperimental — accuracy not yet measured
The Future Research Directions section says that complex human gut microbial ecosystems are difficult to mimic and that translational investigations lag behind preclinical studies. It does not say that infections and psychiatric disorders are underrepresented in animal studies, or attribute such underrepresentation to differences in immune status, microbial exposure, or the scope of behavioral assessments. No page numbers are available in our copy. (LLM, conf=0.76)
- Supportedcites Huggins et al., 2019 source
The Introduction explains that laboratory mice have limited microbial exposure compared with humans and that this difference matters when using mice to study immune fitness. The Discussion adds that SPF housing leaves laboratory mice with an underdeveloped immune system unlike that of adult humans. These passages support the immune-status and microbial-exposure differences for which Huggins et al. is cited; the other citations accompany the behavioral-assessment aspect. (LLM, conf=0.88)
- Partly supportedcites Burger et al., 2023 sourceExperimental — accuracy not yet measured
The Abstract and Discussion report that pathogen-free laboratory mice have different microbial exposure and immune development from mice raised with diverse microbes, supporting the immune-status and microbial-exposure explanation. The paper does not address psychiatric disorders or differences in the scope of behavioral assessments between animals and humans. No page numbers are available in our copy. (LLM, conf=0.86)
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Analytical inconsistencies also arise from the use of different statistical frameworks such as sensitivity, specificity, predictive values, or likelihood ratios, which can yield divergent conclusions (Monticello et al., 2017; Bailey and Balls, 2019; Giblin et al., 2021).”
§ A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics > Introduction
What the sources say
- Not checked — no full text of the sourcecites Monticello et al., 2017 source
- Supportedcites Bailey and Balls, 2019 source
In the main-text section “Monticello et al.” (page unavailable), Bailey and Balls say Monticello et al. relied chiefly on negative predictive value to support animal testing, whereas examining the likelihood ratios yields a different assessment. The paper also explains that likelihood ratios incorporate sensitivity and specificity, supporting the claim that the choice of statistical metric can affect conclusions. (LLM, conf=0.90)
- Partly supportedcites Giblin et al., 2021 sourceExperimental — accuracy not yet measured
The “Quantifying risk” section says likelihood ratios offer a different, potentially better interpretation of concordance than sensitivity, supporting the point that the choice of metric matters. It does not discuss specificity or predictive values as alternative frameworks, or show that using these metrics yields divergent conclusions. (LLM, conf=0.84)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“This result is consistent with previous reports (Bailey et al., 2014; Clark, 2015; Clark and Steger-Hartmann, 2018).”
§ A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics > Results and discussions > SOC-based AE frequency and concordance analysis
What the sources say
- Not checked — no full text of the sourcecites Bailey et al., 2014 source
- Supportedcites Clark, 2015 source
The Abstract states that the absence of adverse events in nonclinical studies is not a good predictor of safety in humans. This supports the citing sentence’s characterization of Clark (2015) as a previous report consistent with that result. (LLM, conf=0.93)
- Not checked — no full text of the sourcecites Clark and Steger-Hartmann, 2018 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
103 more claims · 13 with their cited sources checked
Create a free account to see all 216 claims and every source verdict.
- Probably not a substantial claim. 53
- It may state two separate claims as one. 79
Citations
Cites
Szklarczyk, D., Kirsch, R., Koutrouli, M., Nastou, K., Mehryary, F., Hachilif, R., Gable, A. L., Fang, T., Doncheva, N. T., Pyysalo, S., et al. (2023) · Nucleic Acids Research · doi:10.1093/nar/gkac1000 · Cited by 9,726 · cited 1×
“Protein-protein interaction (PPI) analysis was performed using the STRING database (Szklarczyk et al., 2023) to investigate potential interactions among targets.”
An automated method for finding molecular complexes in large protein interaction networks
Bader, G. D., & Hogue, C. W. (2003) · BMC Bioinformatics · doi:10.1186/1471-2105-4-2 · Cited by 6,072 · cited 1×
“To refine the network and focus on key interactions, the MCODE algorithm (Bader and Hogue, 2003) was applied to identify densely connected sub-networks, which were further analyzed for functional enrichment to determine their biological roles and contributions to toxicity.”
Immune-Related Adverse Events Associated with Immune Checkpoint Blockade
Postow, M. A., Sidlow, R., & Hellmann, M. D. (2018) · New England Journal of Medicine · doi:10.1056/nejmra1703481 · Cited by 4,849 · cited 2×
“This is consistent with the strong concordance observed in mice, which are widely used to model T cell–mediated hepatic injury and immune-related AE (Postow et al., 2018; Shojaie et al., 2021).”
Piñero, J., Bravo, À., Queralt-Rosinach, N., Gutiérrez-Sacristán, A., Deu-Pons, J., Centeno, E., García-García, J., Sanz, F., & Furlong, L. I. (2017) · Nucleic Acids Research · doi:10.1093/nar/gkw943 · Cited by 2,680 · cited 1×
“Pathway and functional enrichment analyses were conducted in R with gprofiler (Kolberg and Raudvere, 2025), employing the DisGeNET database (Piñero et al., 2017) to explore relevant biological pathways and processes.”
Concordance of the toxicity of pharmaceuticals in humans and in animals
Olson, H., Betton, G., Robinson, D., Thomas, K., Monro, A., Kolaja, G., Lilly, P., Sanders, J., Sipes, G., Bracken, W., et al. (2000) · Regulatory Toxicology and Pharmacology · doi:10.1006/rtph.2000.1399 · Cited by 2,050 · cited 2×
“Early work by Olson et al. (2000) within the International Life Sciences Institute’s Health and Environmental Sciences Institute (ILSI-HESI) quantified cross-species sensitivity by analyzing 150 drugs (both investigational and approved).”
An analysis of the attrition of drug candidates from four major pharmaceutical companies
Waring, M. J., Arrowsmith, J., Leach, A. R., Leeson, P. D., Mandrell, S., Owen, R. M., Pairaudeau, G., Pennie, W. D., Pickett, S. D., Wang, J., et al. (2015) · Nature Reviews Drug Discovery · doi:10.1038/nrd4609 · Cited by 1,450 · cited 1×
“Advancing drug safety assessment relies on bridging the translation gap between animal toxicology findings and clinical outcomes, a challenge with far-reaching consequences for R&D efficiency, regulatory success and patient safety (Waring et al., 2015; Amorim et al., 2024).”
Theriot, C. M., Koenigsknecht, M. J., Carlson, P. E., Hatton, G. E., Nelson, A. M., Li, B., Huffnagle, G. B., Z. Li, J., & Young, V. B. (2014) · Nature Communications · doi:10.1038/ncomms4114 · Cited by 974 · cited 1×
“Similarly, C. difficile colitis is supported by mouse models that recapitulate key clinical and histopathologic features and enable mechanistic studies (Chen et al., 2008; Theriot et al., 2014).”
Neurobiology of food intake in health and disease
Morton, G. J., Meek, T. H., & Schwartz, M. W. (2014) · Nature Reviews Neuroscience · doi:10.1038/nrn3745 · Cited by 748 · cited 1×
“These clusters point to CNS-mediated appetite regulation involving hypothalamic neurocircuitry and suggest that hyperphagia may reflect direct drug effects on neural and hormonal satiety control (Morton et al., 2014; Zhuang et al., 2017; He et al., 2021; Wang et al., 2024; Qiu and Fu, 2025).”
Limitations of Animal Studies for Predicting Toxicity in Clinical Trials
Van Norman, G. A. (2019) · JACC: Basic to Translational Science · doi:10.1016/j.jacbts.2019.10.008 · Cited by 732 · cited 3×
“These events often lack precise analogs in animal models, are subject to broad etiologic variability, or are under-detected due to behavioral and physiological differences across species (Van Norman, 2019; Lu et al., 2023; Wu et al., 2023).”
A mouse model of clostridium difficile-associated disease
Chen, X., Katchar, K., Goldsmith, J. D., Nanthakumar, N., Cheknis, A., Gerding, D. N., & Kelly, C. P. (2008) · Gastroenterology · doi:10.1053/j.gastro.2008.09.002 · Cited by 510 · cited 2×
“This observation is consistent with established mouse models of antibiotic-induced C. difficile colitis that recapitulate key features of the human disease, supporting the strong translational relevance of this phenotype (Chen et al., 2008; Best et al., 2012).”
Research works printed but not linked: 1
- Shreffler, 2024, Diagnostic testing accuracy: sensitivity, specificity, predictive values and likelihood ratios
Cited by
Artificial Intelligence Across the Drug Development Lifecycle
Demyashkin, G. A., Алексеевич, П. М., Zyryanov, S., Yavorskiy, A., Shegai, P., & Kaprin, A. (2026) · Medical Sciences · doi:10.3390/medsci14020248 · Cited by 6
“This pattern carries a direct implication for AI, because a model trained predominantly on rat hepatotoxicity data will learn to predict rat hepatotoxicity [191].”
Domingo, J. L. (2026) · Qeios · doi:10.32388/lrqpsd.2
Filipova, D., Fox, L., Goßmann, M., Herrmann, K., Knight, A., Lickiss, B., Maaß, C., Neumann, G., Ritskes-Hoitinga, M. J., & Zietek, T. (2026) · NAM journal. · doi:10.1016/j.namjnl.2026.100136
Groenen, P. M. A. (2026) · NAM journal. · doi:10.1016/j.namjnl.2026.100135
Metadata
- Authors
- Liu, X., & Fan, F.
- Affiliations
- Johnson & Johnson (United States) (US)
- Corresponding author
- Liu, Xin
- Venue
- Frontiers in Toxicology
- Publisher
- Frontiers Media SA
- Published
- 2026-04-02
- Volume
- 8
- Pages
- 1731947-1731947
- ISSN
- 2673-3080 (ISSN-L)
- PMC
- PMC13082757
- OpenAlex
- W7147433783
- Semantic Scholar
- 3f7e708816873f6ec98fa5233bbeb1b8421dd228
- Cited by
- 4
- References
- 81
- Citations by year
- 2026: 4
- FWCI
- 26.68
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Frontiers in Toxicology (publishedVersion, cc-by)
- bioRxiv (Cold Spring Harbor Laboratory) (submittedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Health Sciences › Veterinary › Small Animals › Animal testing and alternatives
- Life Sciences › Pharmacology, Toxicology and Pharmaceutics › Toxicology › Pharmacovigilance and Adverse Drug Reactions
- Life Sciences › Pharmacology, Toxicology and Pharmaceutics › Drug Discovery › Computational Drug Discovery Methods
- Keywords
- drug safety, small molecules, biologics, likelihood ratios, prevalence bias, 3Rs, human risk assessment
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-07, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-07 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics. https://app.cochat.ai/papers/10.3389/ftox.2026.1731947
- The paper
- Liu, X., & Fan, F. (2026). A large-scale concordance study of toxicity findings across preclinical species and humans for small molecules and biologics. Frontiers in Toxicology. https://doi.org/10.3389/ftox.2026.1731947
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.