Body Size And Exposure In Obesity

Why Body Size Can Change an Exposure Comparison

A fixed dose does not guarantee similar exposure across patients. If clearance changes with body size, the same dose can produce different exposure; conversely, scaling dose directly to total body weight can also be inappropriate if clearance rises less than proportionally with weight. In obesity, the relevant question is not simply whether heavier patients receive more drug. It is whether the exposure data support a particular relationship between body size and clearance, and whether a candidate dose strategy reduces meaningful exposure differences without creating new ones.

The analysis below uses body weight as a candidate size measure. That is a modeling choice, not a claim that total body weight is always the best descriptor of drug disposition in obesity. The examples are hypothetical and use simple linear pharmacokinetics to isolate how body-size assumptions affect exposure.

Model and assumptions: For a linear pharmacokinetic model over a dose interval, exposure can be written \(AUC_{\tau,i}=F_i Dose_i/CL_i\), where \(Dose_i\) is the dose in the interval, \(CL_i\) is clearance, and \(F_i\) is bioavailability (with \(F_i=1\) for an intravenous dose). A candidate body-size model is \(CL_i=CL_{70}(WT_i/70)^\theta\), where \(WT_i\) is body weight in kg, \(CL_{70}\) is clearance at 70 kg, and \(\theta\) is the weight exponent. Comparisons using this equation assume linear, time-invariant PK over the relevant range and, when comparing oral exposures, comparable bioavailability. The equation does not account for other patient differences unless they are modeled separately.

A fixed-clearance model is the special case in which clearance is predicted not to change with weight: \(\theta=0\). The power model allows clearance to change with weight. An exponent of 1 would mean clearance increases in direct proportion to weight; an exponent between 0 and 1 means clearance increases less than proportionally. The exponent must be estimated and evaluated for the drug and data at hand; a conventional allometric value is not a universal answer for every drug or obesity population.

How Dose Scaling Follows From the Clearance Model

The exposure implications follow directly from the model. Suppose bioavailability is the same across patients and dose is fixed. Then \(AUC_{\tau}\) is inversely proportional to clearance, so larger predicted clearance means lower predicted exposure. If dose is scaled in proportion to body weight, dose rises as \(WT^1\), which can outpace clearance when \(\theta<1\). A candidate dose proportional to \(WT^\theta\) would, under the model, rise at the same rate as clearance.

AUC_{\tau,i} = \frac{F_i Dose_i}{CL_{70}(WT_i/70)^\theta}

For a common \(F_i=F\), if dose follows \(Dose_i=Dose_{70}(WT_i/70)^\gamma\), then

\frac{AUC_{\tau,i}}{AUC_{\tau,70}} = \left(\frac{WT_i}{70}\right)^{\gamma-\theta}

This result is a model-based comparison, not a dosing recommendation. Fixed dosing has \(\gamma=0\), so exposure varies as \(WT^{-\theta}\). Direct mg/kg dosing has \(\gamma=1\), so exposure varies as \(WT^{1-\theta}\). Matching dose scaling to the estimated clearance exponent gives \(\gamma=\theta\), making the predicted exposure ratio 1 under the assumptions above. Practical dose selection still has to consider available formulations, safety, therapeutic range, uncertainty, and whether the model is credible at the relevant body sizes.

Worked Example: Fixed Dose Versus Weight-Proportional Dose

Assume a hypothetical drug has linear PK, equal bioavailability across patients, and a fitted clearance exponent of \(\theta=0.701\). Compare predicted exposure at 70 kg and 140 kg under a 100 mg fixed dose and under a dose proportional to weight. Use \(CL_{70}=5.00\) L/h and intravenous administration, so \(F=1\).

Clearance ratio. The model predicts \(CL_{140}/CL_{70}=(140/70)^{0.701}=2^{0.701}\approx1.626\). Thus, predicted clearance at 140 kg is \(5.00\times1.626=8.13\) L/h. The exponent is less than 1, so clearance rises less than proportionally to weight.

Fixed 100 mg dose. At 70 kg, \(AUC=100\text{ mg}/5.00\text{ L/h}=20.0\text{ mg·h/L}\). At 140 kg, \(AUC=100\text{ mg}/8.13\text{ L/h}\approx12.3\text{ mg·h/L}\). The exposure ratio is \(12.3/20.0\approx0.615\), also given by \(1/1.626\). Under this model, the fixed dose predicts about 38.5% lower exposure at 140 kg.

Weight-proportional dose. Doubling body weight doubles the dose, from 100 mg to 200 mg. At 140 kg, \(AUC=200\text{ mg}/8.13\text{ L/h}\approx24.6\text{ mg·h/L}\). Relative to 20.0 mg·h/L at 70 kg, the exposure ratio is \(24.6/20.0\approx1.23\). Because the dose doubled while clearance increased by only a factor of 1.626, the weight-proportional strategy predicts about 23.0% higher exposure at 140 kg.

These calculations show why fixed dosing and mg/kg dosing can both produce weight-related exposure differences. They do not establish which strategy is appropriate for a real drug.

Estimating and Comparing Candidate Clearance Relationships

A size relationship is only useful if supported by data spanning the body-size range of interest. A simple way to interpret a fitted power relationship is to compare two clearances. If clearance at two weights is \(CL_1\) and \(CL_2\), and both follow the same power model, then \(\theta=\log(CL_2/CL_1)/\log(WT_2/WT_1)\). This expression follows by dividing the two model equations, which cancels the reference clearance, and taking logarithms. In a population analysis, an exponent is generally estimated using all relevant data and uncertainty in individual clearance estimates; a two-person calculation is illustrative, not a reliable population estimate.

\frac{CL_2}{CL_1} = \left(\frac{WT_2}{WT_1}\right)^\theta \quad\Longrightarrow\quad \theta = \frac{\log(CL_2/CL_1)}{\log(WT_2/WT_1)}

Worked Example: Interpreting a Weight Exponent

Suppose hypothetical individual clearance estimates are 5.00 L/h at 70 kg and 8.13 L/h at 140 kg. Calculate the exponent implied by these two estimates, then state what it means.

The weight ratio is \(140/70=2\), and the clearance ratio is \(8.13/5.00=1.626\). Therefore, \(\theta=\log(1.626)/\log(2)\approx0.701\).

The implied relationship is \(CL\propto WT^{0.701}\) over these two points. Doubling weight is associated in this model with a 1.626-fold clearance, rather than a twofold clearance. The calculation describes a relationship in the hypothetical data; by itself, it does not show that weight caused the clearance difference or that the exponent generalizes beyond the measured range.

The central comparison is between residual patterns under candidate models, not merely between their fitted exponents. An exposure residual can be defined as observed exposure minus predicted exposure; for comparisons across different magnitudes, a relative residual is often easier to interpret: \((AUC_{\mathrm{obs}}-AUC_{\mathrm{pred}})/AUC_{\mathrm{pred}}\). A positive relative residual means observed exposure exceeds the prediction; a negative one means it falls below. If a model systematically overpredicts exposure at high weights and underpredicts it at low weights, or the reverse, its structure may be missing a size-related pattern.

Worked Example: Checking Exposure Residuals Across Weight

Consider hypothetical intravenous 100 mg observations with estimated clearances of 4.00 L/h at 50 kg, 5.00 L/h at 70 kg, and 8.40 L/h at 140 kg. The corresponding observed AUC values are \(100/4.00=25.0\), \(100/5.00=20.0\), and \(100/8.40\approx11.9\) mg·h/L. Compare a fixed-clearance prediction of 5.80 L/h with a size-based prediction using \(CL_{70}=5.00\) L/h and \(\theta=0.701\).

Fixed-clearance model. It predicts \(AUC=100/5.80\approx17.2\) mg·h/L at every weight. The relative residual at 50 kg is \((25.0-17.2)/17.2\approx+45.0\%\). At 70 kg it is \((20.0-17.2)/17.2\approx+16.0\%\). At 140 kg it is \((11.9-17.2)/17.2\approx-30.9\%\). In this small example, the fixed prediction is low at the lower weights and high at the highest weight.

Size-based model. At 50 kg, predicted clearance is \(5.00(50/70)^{0.701}\approx3.95\) L/h, giving predicted AUC \(100/3.95\approx25.3\) mg·h/L and relative residual \((25.0-25.3)/25.3\approx-1.3\%\). At 70 kg, predicted AUC is \(100/5.00=20.0\) mg·h/L, giving a 0.0% residual. At 140 kg, predicted clearance is \(5.00(140/70)^{0.701}\approx8.13\) L/h, giving predicted AUC \(100/8.13\approx12.3\) mg·h/L and relative residual \((11.9-12.3)/12.3\approx-3.2\%\).

The size-based predictions track these three hypothetical observations more closely, but three observations cannot establish model adequacy or a dosing strategy. A larger analysis should assess the residual pattern across the full data, uncertainty in the estimated relationship, and whether other patient characteristics account for apparent weight effects.

Diagnostics, Alternatives, and Limits

A body-size model should be examined across the observed range, with special attention to the heaviest participants rather than relying only on an overall fit statistic. Useful checks include plotting individual or population exposure residuals against body weight and predicted clearance, inspecting observed versus predicted exposure, and examining whether the model systematically misses one end of the weight distribution. Residuals should be interpreted relative to their variability: a small visual trend in a sparse dataset may be noise, while a consistent trend across a well-sampled range suggests remaining structure.

The principal alternatives answer different questions. A fixed-clearance model is simpler and may be adequate when the data do not support a weight effect. A power model estimates a continuous weight relationship. A categorical comparison, such as separate predictions for a defined obesity group, can describe group differences but discards information within groups and depends on the chosen cut point. Direct mg/kg dosing is a dosing rule, not itself evidence that clearance scales linearly with weight. These approaches should be compared using predictive performance and residual behavior, while accounting for uncertainty and complexity; a better fit to the development data alone does not guarantee reliable prediction in new patients.

Several limitations are especially important in obesity. Total body weight may be an imperfect proxy for the physiological determinants of clearance; lean or adjusted body size might be plausible alternatives, but each requires evidence and careful evaluation. Renal function, maturation, disease severity, concomitant medication, and other covariates can be associated with weight and confound an apparent size relationship. Bioavailability can also vary, particularly for oral dosing, so an exposure difference cannot automatically be attributed to clearance. Finally, linear dose-exposure scaling may fail if the drug has nonlinear PK, and predictions beyond the observed body-size range are extrapolations.

A candidate dose strategy should therefore be presented as conditional on its assumptions. Matching dose to a fitted exponent may reduce predicted exposure differences in the model, but it does not address formulation constraints, safety limits, efficacy, or between-patient variability. Exposure targets and clinical dose decisions require evidence beyond a clearance scaling calculation.

Common Pitfall

Common mistake: “Clearance rises with body weight, so dose should be proportional to total body weight.” This reasoning silently assumes \(\theta=1\). If the fitted exponent is below 1, mg/kg dosing increases dose faster than the model predicts clearance to rise, and predicted exposure increases with weight as \(WT^{1-\theta}\). The correction is to estimate and diagnose the clearance relationship, calculate the exposure consequences of each candidate strategy, and treat the result as a model-based comparison rather than an automatic clinical recommendation.

Check Your Understanding

Answer each question using the assumptions stated in the question. Solutions follow.

If \(\theta=0.60\), what is the predicted clearance ratio for a patient at 120 kg relative to 60 kg?
With \(\theta=0.60\), how does predicted exposure under mg/kg dosing change when weight doubles, assuming equal bioavailability?
A fixed-clearance model has positive exposure residuals at low weights and negative residuals at high weights. What pattern does this suggest?
Why does a weight-related exposure difference after oral dosing not, by itself, prove a clearance effect?

Solutions. (1) The weight ratio is \(120/60=2\), so the clearance ratio is \(2^{0.60}\approx1.516\). (2) With mg/kg dosing, \(\gamma=1\), so the exposure ratio is \(2^{1-0.60}=2^{0.40}\approx1.320\); predicted exposure rises about 32.0%. (3) The fixed model tends to predict exposure too low at low weights and too high at high weights, consistent with a residual size trend; other covariates or sampling variability could also contribute. (4) Oral exposure depends on both bioavailability and clearance, \(AUC=F Dose/CL\). A weight-related change in \(F\), or differences in dose or other factors, could contribute to the observed pattern.