CoChat

Beyond Body Weight: A Comprehensive Review of Allometric Scaling in Drug Development for Human Dose Predictions

Open access

The DOI matches its Crossref record; 97 of the 97 references it cites were found in a registry or an index; 95 of the 346 sources its claims cite were checked for support.

Abstract

Allometric scaling provides a practical framework for predicting human pharmacokinetic (PK) parameters from animal data by relating physiological processes to body size through power-law equations. Despite its simplicity and widespread use in first-in-human (FIH) dose selection, its predictive performance is limited by species-specific differences in absorption, distribution, metabolism, and excretion (ADME). This review summarizes the mathematical foundations, workflows, and diagnostics of allometric scaling, while critically examining where the approach succeeds and where it fails. Core concepts, including clearance, volume of distribution, correction factors, and the rule of exponents, are discussed alongside complementary methods: in vitro–in vivo extrapolation (IVIVE), physiologically based pharmacokinetic (PBPK) modelling, and the Wajima normalized time-course method. Historical clinical failures, including fialuridine, TGN1412, BIA 10-2474, and rofecoxib, illustrate the limits of relying solely on allometry, while thalidomide and the fenfluramine combination exemplify toxicodynamic species-selection failures. Modern advances, including the Extended Clearance Classification System (ECCS), target-mediated drug disposition, FcRn recycling, and emerging artificial intelligence and machine-learning methods, are integrated within a framework. Overall, the review treats allometric scaling as a disciplined starting hypothesis that must be triangulated with mechanistic, experimental, and regulatory evidence to support safer and more reliable human translation.

Overview

CoChat Summary

TL;DR

Allometric scaling remains a useful but limited starting hypothesis for predicting human pharmacokinetics from animal body weight, because species-specific ADME, transporter, target, and toxicodynamic biology often override size-based relationships. The review integrates classical allometry with correction factors, IVIVE, PBPK, ECCS, TMDD, Wajima normalization, and AI/ML into a mechanism-anchored decision framework for first-in-human dose prediction.

Key findings

  • Kleiber’s basal metabolic rate scaling is approximately \(B = aM^{0.75}\), but the review notes the exponent is contested: empirical estimates for physiological and PK parameters typically fall between roughly 0.67 and 0.80, so 0.75 should be treated as a convenient default rather than a biophysical constant.
  • In the worked clearance example, OLS regression gave \(\log_{10}(CL) = 0.603 \times \log_{10}(BW) + 0.931\), projecting human CL at 70 kg as approximately 110 mL/min; despite \(R^2 = 0.939\), the 95% confidence interval for the mean prediction spanned roughly 13–905 mL/min and the 95% prediction interval roughly 6–1970 mL/min.
  • The rule of exponents recommends simple allometry when the exponent is 0.55–0.70, CL × maximum life-span potential when 0.71–1.0, and a two-term CL × brain-weight equation when the exponent exceeds 1.0; brain-weight correction factors were consistently larger than maximum-life-span factors by approximately 1.3–1.9.
  • Lymphatic transport scales differently from systemic clearance: lymph flow and triglyceride mass transport scaled with exponents of approximately 0.84–0.94 and 0.80–0.96, while halofantrine lymphatic mass transport increased with an exponent of approximately 1.3; larger species such as dogs better resemble human lymphatic drug transport than rodents.

Methods

  • Narrative, non-systematic review; sources were selected for conceptual and methodological relevance to interspecies PK scaling and first-in-human dose prediction, with emphasis on primary methodological papers, clinical case histories, and regulatory guidance rather than a registered systematic review protocol.
  • The review synthesizes classical body-weight allometry, correction factors and the rule of exponents, IVIVE, PBPK, ECCS, target-mediated drug disposition, intestinal lymphatic transport, lysosomal sequestration, and AI/ML approaches into a proposed stepwise decision framework.
  • Worked examples in Section 2.2 and Appendix A are hypothetical and constructed for didactic clarity; the Wajima worked example in Section 6.2 uses an independent hypothetical parameter set and is not numerically linked to the clearance example.

Limitations

  • The paper states that predictive performance depends critically on the quality and number of preclinical species, with panels of three to four species typical but providing limited leverage; species choice often reflects historical convention rather than mechanistic relevance.
  • The paper states that the allometry literature is subject to publication bias toward successful predictions, transporter-mediated scaling carries greater uncertainty than CYP-mediated scaling, idiosyncratic toxicities are not captured by any PK scaling framework, and retrospective benchmarks may not generalize to chemically novel modalities.
  • Evident from the text, though not framed as a limitation: the review is non-systematic and does not perform a formal quantitative synthesis or meta-analysis, and its worked examples are hypothetical rather than empirical datasets.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/pharmaceutics18070824

Cites

97 of 97 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Allometric scaling of clearance for the hypothetical drug. (a) Raw log–log scatter of the four preclinical species (mouse, rat, monkey, dog), shown without overlay so the empirical dispersion can be judged independently of any model. (b) The same data with the ordinary least squares (OLS) regression line log10(CL) = 0.603 × log10(BW) + 0.931, R2 = 0.939 and the projected human clearance at 70 kg (110 mL/min, star) overlaid.
Figure 1 Allometric scaling of clearance for the hypothetical drug. (a) Raw log–log scatter of the four preclinical species (mouse, rat, monkey, dog), shown without overlay so the empirical dispersion can be judged independently of any model. (b) The same data with the ordinary least squares (OLS) regression line log10(CL) = 0.603 × log10(BW) + 0.931, R2 = 0.939 and the projected human clearance at 70 kg (110 mL/min, star) overlaid.CC BY · PubMed Central
Figure 2 Representative log–log scaling relationships for key PK parameters. (a) Clearance versus body weight on a log–log scale. The slope approximates 0.75, consistent with Kleiber’s law for metabolic rate, but species expressing unique CYP isoforms may deviate. (b) Volume of distribution versus body weight (log–log). The exponent typically approaches 1.0, reflecting that body size is linearly related to drug distribution space. Highly lipophilic or tissue-binding compounds may show marked deviations. (c) Absorption rate constant (ka) versus body weight (log–log). The negative slope reflects longer gastrointestinal transit times in larger species. Interspecies variability in gastric pH, transporter expression, and first-pass metabolism limits the predictive power of this relationship. (d) Half-life versus body weight (log–log). Since t½ = 0.693 × VD/CL, prediction accuracy depends on the fidelity of both clearance and volume estimates. Crucially, these diagnostics certify only the quality of a fit within the observed range; they cannot reveal whether the underlying disposition biology is conserved across species, which is exactly where body-weight scaling most often breaks down.
Figure 2 Representative log–log scaling relationships for key PK parameters. (a) Clearance versus body weight on a log–log scale. The slope approximates 0.75, consistent with Kleiber’s law for metabolic rate, but species expressing unique CYP isoforms may deviate. (b) Volume of distribution versus body weight (log–log). The exponent typically approaches 1.0, reflecting that body size is linearly related to drug distribution space. Highly lipophilic or tissue-binding compounds may show marked deviations. (c) Absorption rate constant (ka) versus body weight (log–log). The negative slope reflects longer gastrointestinal transit times in larger species. Interspecies variability in gastric pH, transporter expression, and first-pass metabolism limits the predictive power of this relationship. (d) Half-life versus body weight (log–log). Since t½ = 0.693 × VD/CL, prediction accuracy depends on the fidelity of both clearance and volume estimates. Crucially, these diagnostics certify only the quality of a fit within the observed range; they cannot reveal whether the underlying disposition biology is conserved across species, which is exactly where body-weight scaling most often breaks down.CC BY · PubMed Central
Figure 3 Wajima multi-species worked example after a 1 mg/kg IV bolus in mouse, rat, monkey, and dog. (a) Absolute plasma concentration–time profiles, plotted on a log y-axis, differ markedly across species in both timescale and concentration range. (b) After Wajima normalization (x-axis rescaled to t/MRT, y-axis rescaled to C/(Dose/Vss)), the four profiles collapse onto a single common curve, demonstrating the superposition principle on which forward projection to humans relies. Parameter values and exposure-metric dispersion are tabulated in Appendix A (Table A4 and Table A5).
Figure 3 Wajima multi-species worked example after a 1 mg/kg IV bolus in mouse, rat, monkey, and dog. (a) Absolute plasma concentration–time profiles, plotted on a log y-axis, differ markedly across species in both timescale and concentration range. (b) After Wajima normalization (x-axis rescaled to t/MRT, y-axis rescaled to C/(Dose/Vss)), the four profiles collapse onto a single common curve, demonstrating the superposition principle on which forward projection to humans relies. Parameter values and exposure-metric dispersion are tabulated in Appendix A (Table A4 and Table A5).CC BY · PubMed Central
Figure 4 Practical decision tree for FIH human PK prediction. The algorithm begins by partitioning the molecule into small molecules versus biologics, then queries the ECCS/BDDCS class and whether linear or nonlinear pharmacokinetics are anticipated. For linear small molecules within CYP-dominated ECCS classes, simple multi-species allometry with ROE-informed correction provides the empirical starting point, cross-checked against IVIVE. For transporter-substrate compounds, simple allometry is downweighted and transporter-aware PBPK with explicit hepatic uptake modules is prioritized. For non-linear or saturable PK, Michaelis–Menten parameterization replaces simple allometric extrapolation. For biologics, TMDD-aware PK/PD modelling and MABEL-anchored dose selection govern, with allometric exponents used only to bound central-compartment volume and FcRn-mediated linear clearance empirically. All four branches converge at a triangulation node, where MRSD (BSA-scaled), MABEL (potency-anchored), and PBPK-simulated exposures are compared, with the most conservative, defensible value selected for FIH initiation and the full chain documented for regulatory submission. The tree is intended as a teaching aid and pre-data planning instrument; mechanistic findings during model qualification may legitimately override its default branches. Abbreviations: FIH, first-in-human; ECCS, Extended Clearance Classification System; BDDCS, Biopharmaceutics Drug Disposition Classification System; PK, pharmacokinetic; CYP, cytochrome P450; ROE, rule of exponents; IVIVE, in vitro-in vivo extrapolation; PBPK, physiologically based pharmacokinetic; TMDD, target-mediated drug disposition; PK/PD, pharmacokinetic/pharmacodynamic; MABEL, minimum anticipated biological effect level; Vc, central volume of distribution; FcRn, neonatal Fc receptor; CL, clearance; MRSD, maximum recommended starting dose; BSA, body surface area.
Figure 4 Practical decision tree for FIH human PK prediction. The algorithm begins by partitioning the molecule into small molecules versus biologics, then queries the ECCS/BDDCS class and whether linear or nonlinear pharmacokinetics are anticipated. For linear small molecules within CYP-dominated ECCS classes, simple multi-species allometry with ROE-informed correction provides the empirical starting point, cross-checked against IVIVE. For transporter-substrate compounds, simple allometry is downweighted and transporter-aware PBPK with explicit hepatic uptake modules is prioritized. For non-linear or saturable PK, Michaelis–Menten parameterization replaces simple allometric extrapolation. For biologics, TMDD-aware PK/PD modelling and MABEL-anchored dose selection govern, with allometric exponents used only to bound central-compartment volume and FcRn-mediated linear clearance empirically. All four branches converge at a triangulation node, where MRSD (BSA-scaled), MABEL (potency-anchored), and PBPK-simulated exposures are compared, with the most conservative, defensible value selected for FIH initiation and the full chain documented for regulatory submission. The tree is intended as a teaching aid and pre-data planning instrument; mechanistic findings during model qualification may legitimately override its default branches. Abbreviations: FIH, first-in-human; ECCS, Extended Clearance Classification System; BDDCS, Biopharmaceutics Drug Disposition Classification System; PK, pharmacokinetic; CYP, cytochrome P450; ROE, rule of exponents; IVIVE, in vitro-in vivo extrapolation; PBPK, physiologically based pharmacokinetic; TMDD, target-mediated drug disposition; PK/PD, pharmacokinetic/pharmacodynamic; MABEL, minimum anticipated biological effect level; Vc, central volume of distribution; FcRn, neonatal Fc receptor; CL, clearance; MRSD, maximum recommended starting dose; BSA, body surface area.CC BY · PubMed Central

Verification

Claims the paper makes

69 Valid · 20 Not valid · 92 Not checked · 24 Own results and interpretations

90 more claims · 31 with their cited sources checked

Create a free account to see all 205 claims and to verify the 8 references still open.

See all claims — free account
  • It may state two separate claims as one. 98
  • Probably not a substantial claim. 18

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 97 references

Research works printed but not linked: 1

  1. Institute of Medicine (US) Committee to Review the Fialuridine (FIAU/FIAC) Clinical Trials Review of the Fialuridine (FIAU) Clinical Trials Manning F.J. Swartz M. Institute of Medicine (US) Washington, DC, USA 1995 25121268

Metadata

Authors
Mallillin, M. C., Silva, D. A., Miller, N. A., Zhao, S., Salami, M., Löbenberg, R., & Davies, N. M.
Affiliations
  • University of Alberta (CA)
  • Simulations Plus (United States) (US)
  • Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Edmonton, AB T6G 2H7, Canada
Corresponding author
  • Löbenberg, Raimar
  • Davies, Neal M.
Venue
Pharmaceutics
Publisher
MDPI AG
Published
2026-07-03
Volume
18
Issue
7
Pages
824
ISSN
1999-4923 (ISSN-L)
OpenAlex
W7167236904
Cited by
2
References
96
Citations by year
2026: 2
FWCI
2.79
Type
journal article
Language
en
Open access
Locations
  • Pharmaceutics (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
Topics
  • Health Sciences › Medicine › Pediatrics, Perinatology and Child Health › Pharmaceutical studies and practices
  • Health Sciences › Veterinary › Small Animals › Animal testing and alternatives
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Clinical Trials
Keywords
allometric scaling, first-in-human dose selection, interspecies extrapolation, clearance, volume of distribution, correction factors, in vitro-in vivo extrapolation, physiologically based pharmacokinetic modelling, ADME, target-mediated drug disposition, thalidomide, fenfluramine, rofecoxib, TGN1412, FIAU, drug development

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

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.