🔬 Page 136: Cross-Species Allometric Dose Scaling

Why Simple Weight-Based Scaling Fails Across Species

The most intuitive way to convert an animal drug dose to a human dose is direct proportionality: if a 0.02 kg mouse tolerates 10 mg/kg, a 70 kg human should tolerate the same 10 mg/kg. This assumption — that dose per unit body weight is constant across species — is convenient but physiologically wrong. Metabolic rate, and with it the rate of drug clearance, does not scale linearly with mass across the 3,500-fold weight range separating a mouse from a human.

• ~1:3,500: Mouse : human weight ratio (0.02 kg vs 70 kg)
• ~0.75: Metabolic scaling exponent (Kleiber's law, not 1.0)
• up to 10×: mg/kg overdose risk (naive) (small-animal to human)
• multiple: Historical FIH miscalculations (driving regulatory guidance)

The false assumption behind mg/kg scaling

Simple mg/kg extrapolation treats every kilogram of tissue as metabolically equivalent, regardless of species. In reality, smaller animals have much higher mass-specific metabolic rates: a mouse's heart beats roughly 600 times per minute and it consumes oxygen per gram of tissue at many times the human rate. Its liver processes drugs proportionally faster, meaning a mouse can safely tolerate a much higher mg/kg dose than a human simply because it eliminates the compound faster.

Applying the mouse's mg/kg tolerance directly to a human, without correction, tends to overestimate the safe human dose — sometimes drastically. Conversely, scaling up from a large animal like a dog can underestimate what a smaller species could tolerate. The direction and magnitude of the error depends on which species pair is used and how far apart their sizes are.

Metabolic rate does not scale linearly with mass

Kleiber's law, one of the most robust empirical relationships in biology, states that basal metabolic rate scales with body mass raised to approximately the 0.75 power — not to the first power. Doubling body weight does not double metabolic rate; it increases metabolic rate by only about 68%. This sub-linear relationship holds remarkably well from mice to elephants, spanning many orders of magnitude in body mass.

Because drug clearance is mechanistically tied to metabolic and perfusion processes (hepatic blood flow, glomerular filtration, enzyme turnover), it inherits this same sub-linear scaling behavior. A scaling law built on weight^1.0 therefore systematically diverges from the true biological relationship as the size gap between species widens — small for closely-sized species, large for mouse-to-human conversions.

A dose that is perfectly safe on a mg/kg basis in mice can translate to a dangerously different exposure in humans if converted with simple linear scaling — this is precisely why regulatory guidance mandates allometric, not linear, extrapolation for first-in-human dose estimation.

Body Surface Area as a Better Scaling Proxy

Long before allometric exponents were formalized, oncologists observed that chemotherapy doses translated far more reliably across species and even across individual patients when normalized to body surface area (BSA) rather than body weight. This is because many of the physiological processes that govern drug disposition — heat loss, blood volume distribution, renal filtration, cardiac output — scale more closely with the surface across which exchange occurs than with the volume that contains the tissue.

• ~0.67: BSA scaling exponent (W^(2/3), classic Meeh relationship)
• DuBois: Common BSA formula (BSA = 0.007184·W^0.425·H^0.725)
• mg/m²: Chemotherapy dosing basis (not mg/kg, since 1950s)
• decreases: Surface : volume ratio trend (as body size increases)

Why surface area correlates with physiological rate processes

A sphere's surface area scales with the square of its radius, while its volume scales with the cube. As an organism grows larger, its surface-to-volume ratio necessarily shrinks — a small animal has proportionally much more surface area relative to its mass than a large one. Because heat dissipation, gas exchange, and many transport processes occur across surfaces (skin, capillary beds, alveolar membranes, glomerular filtration surfaces), these processes scale more closely with surface area than with raw mass.

This is why smaller endotherms must generate proportionally more heat per gram of tissue to maintain body temperature — and why their metabolic machinery, including drug-metabolizing enzyme systems, runs proportionally faster. The same geometric logic that governs heat loss also governs the pharmacokinetic parameters (clearance, distribution) most relevant to dosing.

From geometric surface area to a practical scaling proxy

True body surface area for an irregular organism cannot be measured from weight alone, but empirical formulas (Meeh, DuBois & DuBois, Mosteller) approximate BSA well using only weight and height/length. Across mammalian species, when BSA is estimated from weight using the relationship BSA ∝ W^(2/3), it correlates with basal metabolic rate and drug clearance much better than weight alone does.

This observation is the physiological bridge between the naive mg/kg approach and the more defensible allometric approach used today: instead of measuring surface area directly for every species pair, pharmacologists use body weight raised to a fractional exponent close to 2/3 (or empirically closer to 0.75 for metabolic rate specifically) as a practical, calculable proxy for the surface-area-driven physiology that actually governs dose translation.

The Allometric Exponent in Dose Conversion Formulas

The general allometric equation Y = a·W^b — where Y is a physiological parameter, W is body weight, and b is the allometric exponent — underlies essentially all cross-species dose conversion. For basal metabolic rate, b is empirically close to 0.75; for body-surface-area-related processes, b is closer to 0.67. Regulatory dose-conversion practice adopts an exponent in this range, letting a single formula stand in for the full physiological complexity of surface area and metabolic scaling.

• 0.67: Working exponent used here (body-surface-area convention)
• 0.75: Metabolic-rate exponent (Kleiber's law alternative)
• Dₜ = Dₛ·(Wₜ/Wₛ)^b: Formula form (total-dose scaling)
• W^0.67: FDA/ICH guidance basis (human equivalent dose (HED))

Reading the formula: what the exponent actually does

In the dose-scaling formula Dₜ = Dₛ × (Wₜ/Wₛ)^b, the ratio of target-to-source body weight is raised to the fractional exponent b before it multiplies the source dose. Because b is less than 1, the scaling factor grows more slowly than the weight ratio itself: converting from a 0.02 kg mouse to a 70 kg human (a 3,500-fold weight increase) yields a dose-scaling factor of roughly 3,500^0.67 ≈ 240 — not 3,500. This compression is exactly what corrects for the naive linear approach's tendency to overestimate.

The live formula panel and metric tiles in this simulator recompute this factor continuously as the source and target weight sliders move, letting you see how sensitive the scaled dose is to the size gap between species and to the choice of exponent.

Which exponent to use, and why it matters

Different physiological endpoints justify slightly different exponents. Basal metabolic rate across mammals fits W^0.75 (Kleiber's law) extremely well. Renal clearance and many hepatically-cleared drugs often fit closer to W^0.75 as well, while surface-area-driven processes (and some historical FDA guidance for maximum recommended starting dose) use W^0.67. The difference between 0.67 and 0.75 seems small, but compounded across a 3,500-fold weight ratio it changes the resulting scaling factor substantially.

Because no single exponent is correct for every drug and every route of elimination, regulatory guidance treats the allometrically-scaled dose as a starting estimate to be refined — not a final answer. The formula compresses an enormous amount of biological complexity (enzyme expression, plasma protein binding, tissue distribution, active transport) into one exponent, and that compression is where the approximation's limits lie.

Using this simulator's slider defaults — mouse (0.02 kg) as source and human (70 kg) as target — the allometric scaling factor works out to roughly 240×, dramatically different from a naive linear (weight-ratio) factor of 3,500×. That difference is the entire reason allometric scaling exists.

Starting Dose Estimation for First-in-Human Trials

When a new drug moves from animal toxicology studies toward its first administration in humans, allometric scaling of the No-Observed-Adverse-Effect-Level (NOAEL) from the most sensitive animal species provides the Human Equivalent Dose (HED) — a critical anchor point. But the HED is never used directly as the starting human dose: regulatory guidance requires it be divided by a safety factor, typically at least 10-fold, to arrive at the Maximum Recommended Starting Dose (MRSD).

• NOAEL: Anchor value (from most sensitive species)
• HED: Conversion output (human equivalent dose)
• ≥10×: Typical safety factor (default per FDA guidance)
• MRSD: Final output (max. recommended starting dose)

From animal NOAEL to Human Equivalent Dose

Toxicology studies in at least two species (commonly a rodent and a non-rodent, e.g. rat and dog) identify the NOAEL — the highest dose producing no observable adverse effects. Each species' NOAEL is allometrically converted to its Human Equivalent Dose using the W^0.67 relationship. The most conservative (lowest) HED across the tested species is typically selected as the basis for further calculation, since it represents the most sensitive biological signal available.

This step alone already builds in considerable conservatism: choosing the most sensitive species' converted value, rather than averaging across species, protects against the possibility that humans respond more like the more sensitive animal model.

Applying the safety factor to reach a starting dose

The HED is not administered directly to first-in-human trial participants. A safety factor — by default at least 10-fold, and often larger for drugs with steep dose-response curves, irreversible toxicity, or a narrow therapeutic index — is applied by dividing the HED. The result is the Maximum Recommended Starting Dose (MRSD), the ceiling for the very first, lowest dose cohort in a Phase 1 trial.

Additional considerations can push the safety factor even higher: pharmacologically active dose (PAD) estimates from in vitro potency data, target receptor occupancy modeling, and the steepness of the toxicity curve observed in animal studies are all weighed alongside the allometric HED before a final starting dose is locked in. The scaled dose computed by this simulator's formula should be read as an illustrative HED-like value — real MRSD determination always incorporates this additional conservative reduction.

A 10-fold (or greater) safety factor exists precisely because allometric scaling is an approximation of animal toxicology, not a guarantee of human safety — the infamous 2006 TGN1412 trial, where six healthy volunteers suffered severe cytokine release syndrome despite dosing calculated from animal data, remains a stark reminder of why this margin is non-negotiable.

Limitations Requiring Species-Specific Validation

Allometric scaling is a statistically-derived approximation built on population-level trends across many drugs and species — it is not a guarantee for any individual compound. Species differ substantially in the specific enzyme pathways, transporters, and receptor biology that determine how a given molecule is actually handled, and these differences can cause real pharmacokinetics to deviate meaningfully from the allometric prediction.

• partial: CYP450 isoform overlap (species-specific enzyme sets)
• notable subset: Drugs poorly predicted by W^b (renally-cleared, actively transported)
• microdosing: Validation approach (human PK bridging studies)
• estimate, not proof: Regulatory stance (HED requires confirmation)

Where allometric predictions break down

Allometric scaling works best for drugs eliminated primarily through processes that scale predictably with body size — passive renal filtration, generalized hepatic blood flow. It performs far less reliably for compounds whose clearance depends on specific enzyme isoforms or active transporters that are expressed differently, or are entirely absent, across species.

Cytochrome P450 enzymes are a prime example: humans and preclinical species express overlapping but non-identical sets of CYP isoforms, with different relative abundances and substrate specificities. A drug metabolized predominantly by a CYP isoform poorly represented in the toxicology species can show a clearance rate in humans that diverges sharply from what weight-based allometry would predict — in either direction.

Additional sources of species-specific deviation

Beyond enzyme pathway differences, several other factors can cause actual human pharmacokinetics to diverge from allometric predictions:

• Plasma protein binding: differing albumin and α1-acid glycoprotein binding affinities between species change the free (active) drug fraction available for clearance and effect • Active transport: efflux and uptake transporters (P-glycoprotein, OATPs) are expressed at different densities across species and tissues, affecting distribution and elimination independent of body size • Gut microbiome and first-pass metabolism: oral bioavailability can differ substantially between species for compounds subject to significant pre-systemic metabolism • Receptor and target biology: even when exposure is correctly predicted, the pharmacodynamic response to a given exposure can differ if the drug target itself is expressed differently or has different affinity across species

Empirical validation closes the gap

Because of these limitations, allometric scaling is treated throughout drug development as a starting estimate that must be confirmed, not a final answer. Microdosing studies (administering sub-pharmacological "exploratory" doses to a small number of human volunteers) allow early empirical measurement of actual human pharmacokinetic parameters, which can then be compared against — and used to correct — the allometrically-predicted values before full-scale dosing decisions are made.

Throughout Phase 1 escalation, each dose cohort's observed pharmacokinetics is compared against the model prediction, and the dosing plan for subsequent cohorts is adjusted accordingly. This iterative empirical correction, layered on top of the initial allometric estimate and its safety-factor margin, is what ultimately makes cross-species dose translation safe in practice — the formula gets the process started; real human data is what gets it right.

No allometric formula, however refined, replaces direct measurement. Every regulatory framework treats the scaled dose as a hypothesis to be tested conservatively — never as a substitute for empirical species-specific and eventually human-specific validation.