Body weight as a covariate (allometric scaling)
A PK parameter is often scaled as typical value × (weight / reference weight)^exponent. Fitted exponents for antibodies are usually below one, which is why fixed and weight-based dosing give different exposure at the extremes.
Definition
A covariate is a patient characteristic that explains part of the variability of a parameter. For body weight, the usual form is allometric scaling:
parameter = typical value × (weight / reference weight)^exponent
Theoretical exponents are 0.75 for clearances and 1 for volumes, but fitted values for a given drug differ and are reported with uncertainty.
Why the exponent matters for dosing
If clearance scales with weight^0.6, then a dose proportional to weight (mg/kg) gives an exposure that rises slightly with weight (AUC ∝ weight^0.4), while a fixed dose gives an exposure that falls with weight (AUC ∝ weight^−0.6). Near the reference weight the two are about the same; at the extremes they differ in opposite directions.
On this site
The reimplemented pembrolizumab model has weight exponents of 0.595 on clearance and 0.489 on volume, and the nivolumab model 0.566 and 0.597. The anti-PD-1 analysis shows the consequence: with fixed doses the troughs fall in heavy patients, and with weight-based doses the first-interval troughs fall in light patients. The sensitivity analysis moves the exponent to the ends of its 95% interval. See also fixed versus weight-based dosing and clearance.
Frequently asked questions
Is clearance proportional to body weight?
No. Fitted exponents are typically below one, for example about 0.6 for the antibodies here, so clearance rises less than in proportion to weight.
Why not simply dose by weight?
With exponents below one, weight-based dosing over-corrects: exposure then rises with weight. Neither approach equalizes exposure at the extremes, which is the point of comparing them.