Scaling In Vitro Transporter Abundance to Pediatric Tissues

Why Abundance Scaling Needs More Than an Age Curve

An in vitro transporter measurement describes the assay system that was measured. A pediatric tissue estimate must also account for how that system relates to the target tissue and how transporter abundance in that tissue differs with age. These are distinct conversions: a tissue scaling factor addresses the assay-to-tissue bridge, while an ontogeny ratio addresses the age-specific difference. Combining them carelessly can omit a conversion or count one twice.

The earlier material on Transporter Ontogeny for Hepatic Uptake and Efflux establishes the role of age-dependent transporter profiles. Here the focus is how to put such a profile on the same abundance scale as an in vitro measurement. The result is an estimate of transporter abundance in tissue, not a direct estimate of transport activity or drug clearance.

Definition: Let $B_{\mathrm{vitro}}$ be measured transporter abundance in an in vitro system, in pmol per mg of the specified assay protein. Let $K_{\mathrm{tissue}}$ be an adult tissue scaling factor, in mg of protein-equivalent per g of tissue, calibrated to map that assay abundance to adult tissue abundance. Let $q(a)$ be the dimensionless abundance ratio at age $a$, defined as age-$a$ tissue abundance per g divided by adult tissue abundance per g. Then the age-specific tissue abundance is $$ C_{\mathrm{tissue}}(a)=B_{\mathrm{vitro}}K_{\mathrm{tissue}}q(a), $$ with units pmol/g tissue.

The factor $K_{\mathrm{tissue}}$ can be constructed from two quantities measured on compatible protein bases. If $B_{\mathrm{adult,tissue}}$ is adult tissue transporter abundance in pmol per mg of the relevant tissue protein, and $P_{\mathrm{adult,tissue}}$ is that protein amount in mg per g of tissue, then $$ K_{\mathrm{tissue}}=\frac{B_{\mathrm{adult,tissue}}}{B_{\mathrm{vitro}}}P_{\mathrm{adult,tissue}}. $$ The abundance ratio is dimensionless, so $K_{\mathrm{tissue}}$ has units mg/g. Multiplying by $B_{\mathrm{vitro}}$ yields adult tissue abundance in pmol/g; multiplying by $q(a)$ then gives the age-specific value.

This formulation requires $q(a)$ to compare abundance per g tissue at age $a$ with adult abundance per g tissue. Ontogeny data reported per mg protein do not automatically provide that ratio: if tissue protein yield per g changes with age, that yield must also be included. An alternative is to use the age-specific abundance per mg protein together with the age-specific protein yield. The two approaches agree only when their denominators and conversions are aligned.

Assumptions Behind the Conversion

The product formula is simple, but its interpretation depends on several assumptions. First, the in vitro and tissue measurements must refer to the same transporter, with compatible definitions of abundance and protein denominator. A measurement per mg total cell protein cannot be combined as though it were per mg membrane protein without an explicit conversion. Second, the tissue scaling factor must represent the target tissue and assay system; a factor calibrated in one tissue or expression system is not automatically transferable to another.

Third, the ontogeny ratio must correspond to the same tissue, transporter, and abundance metric as the adult reference. The ratio is not a generic pediatric multiplier. Finally, the method treats the measured abundance as relevant to the tissue quantity being estimated. It does not by itself correct for differences in transporter localization, membrane recovery, post-translational state, or activity per molecule. Those differences can make abundance an imperfect proxy for functional transport.

Worked Example: Scaling an Assay Measurement to an Infant Tissue

Suppose a hypothetical in vitro assay measures $8.0$ pmol transporter per mg of assay protein. Adult target tissue contains $2.4$ pmol per mg of the corresponding tissue protein, and the tissue protein yield is $40$ mg/g tissue. An age-stratified dataset estimates that abundance per g tissue in the infant age group is $0.45$ times the adult value. Calculate the infant tissue abundance.

First calculate the adult tissue-to-assay abundance ratio: $$ \frac{2.4\ \mathrm{pmol/mg}}{8.0\ \mathrm{pmol/mg}}=0.300. $$ It is dimensionless because the numerator and denominator use the same abundance units. The tissue scaling factor is therefore $$ K_{\mathrm{tissue}}=0.300\times40\ \mathrm{mg/g}=12.0\ \mathrm{mg/g}. $$ The adult tissue abundance implied by the assay and scaling factor is $8.0\times12.0=96.0$ pmol/g tissue. Applying the age ratio gives $$ C_{\mathrm{tissue}}(\mathrm{infant})=8.0\ \mathrm{pmol/mg}\times12.0\ \mathrm{mg/g}\times0.45=43.2\ \mathrm{pmol/g}. $$ The estimate is $43.2$ pmol/g tissue. This calculation assumes the $0.45$ ratio is defined per g tissue; if it were measured per mg protein, the age-specific protein yield would also be needed.

Deriving and Using the Scaling Factor

The scaling equation follows by connecting quantities that share the same abundance basis. By definition, the adult tissue-to-assay ratio is $R=B_{\mathrm{adult,tissue}}/B_{\mathrm{vitro}}$. Thus $B_{\mathrm{adult,tissue}}=B_{\mathrm{vitro}}R$, in pmol per mg protein. Multiplying by tissue protein yield converts the denominator from mg protein to g tissue: $$ C_{\mathrm{adult,tissue}}=B_{\mathrm{vitro}}R P_{\mathrm{adult,tissue}}. $$ Since the age ratio is $q(a)=C_{\mathrm{tissue}}(a)/C_{\mathrm{adult,tissue}}$, rearrangement gives $C_{\mathrm{tissue}}(a)=B_{\mathrm{vitro}}R P_{\mathrm{adult,tissue}}q(a)$, which is the stated product formula with $K_{\mathrm{tissue}}=RP_{\mathrm{adult,tissue}}$. This derivation also shows why age and tissue conversions must use matching denominators: the cancellation is valid only when the compared quantities are defined consistently.

Worked Example: Checking Protein Denominators Before Applying Ontogeny

In a hypothetical study, the adult tissue-to-assay ratio is $0.25$, and adult protein yield is $36$ mg/g tissue. The assay abundance is $10$ pmol/mg assay protein. An ontogeny dataset reports $0.60$ for a pediatric-to-adult ratio measured per mg tissue protein. Suppose pediatric and adult protein yields are $30$ and $36$ mg/g tissue, respectively. Find the pediatric abundance per g tissue.

The adult scaling factor is $0.25\times36=9.0$ mg/g, so adult abundance per g tissue is $$ 10\ \mathrm{pmol/mg}\times9.0\ \mathrm{mg/g}=90.0\ \mathrm{pmol/g}. $$ The given $0.60$ is a ratio per mg protein, not per g tissue. The corresponding per-g-tissue ratio is $$ q_{\mathrm{per\ g}}=0.60\times\frac{30\ \mathrm{mg/g}}{36\ \mathrm{mg/g}}=0.500. $$ Therefore pediatric tissue abundance is $90.0\times0.500=45.0$ pmol/g. Directly multiplying $90.0$ by $0.60$ would give $54.0$ pmol/g and would incorrectly assume equal protein yields.

Uncertainty in the Product

Uncertainty can arise from the assay abundance, the tissue scaling factor, and the age ratio. A useful approximation is to represent each positive input as lognormally distributed and to assume independence. For a lognormal quantity with coefficient of variation $CV$, the variance on the log scale is $\ln(1+CV^2)$. For independent inputs, the log variance of their product is the sum of their log variances. Independence is an assumption, not a default truth: shared assay methods or shared reference samples can induce dependence.

Justification: If $Y=B_{\mathrm{vitro}}K_{\mathrm{tissue}}q(a)$, then $\ln Y=\ln B_{\mathrm{vitro}}+\ln K_{\mathrm{tissue}}+\ln q(a)$. Under independent lognormal inputs, the variance of a sum is the sum of variances. Hence $$ \sigma_{\ln Y}^2=\ln(1+CV_B^2)+\ln(1+CV_K^2)+\ln(1+CV_q^2). $$ The corresponding central 95% interval for a lognormal product with median $Y_{\mathrm{med}}$ is $Y_{\mathrm{med}}\exp(\pm1.96\sigma_{\ln Y})$.

Worked Example: Combining Assay, Scaling, and Ontogeny Uncertainty

Use the previous infant estimate of $43.2$ pmol/g as the median product estimate. Suppose the hypothetical coefficients of variation are 20% for assay abundance, 25% for tissue scaling, and 15% for the age ratio. Assume independent lognormal inputs and that the stated central values are medians.

The log standard deviation is $$ \sigma_{\ln Y}=\sqrt{\ln(1+0.20^2)+\ln(1+0.25^2)+\ln(1+0.15^2)} =\sqrt{0.03922+0.06062+0.02225}=0.3494. $$ The 95% multiplicative limits are $\exp(-1.96\times0.3494)=0.504$ and $\exp(1.96\times0.3494)=1.984$. Thus the interval is $$ 43.2\times(0.504,\ 1.984)=(21.8,\ 85.7)\ \mathrm{pmol/g}. $$ Under the specified lognormal and independence assumptions, this is a model-based 95% interval for the product estimate. It is not a 95% probability statement that a fixed true abundance lies in this particular interval. If inputs are correlated or the uncertainty is dominated by an unmodeled assay-to-tissue difference, this interval may understate or misrepresent uncertainty.

Diagnostics, Alternatives, and Limitations

Before interpreting a scaled value, check the measurement basis from end to end. Confirm the transporter identity, tissue, assay system, protein denominator, and units for every input. Recalculate the units in the product; the final result should have the intended units, such as pmol/g tissue. Check whether an age ratio is per g tissue or per mg protein, and verify that the adult reference used to define it matches the adult tissue scaling estimate. A ratio that is zero, implausibly large, or inconsistent across assay platforms should prompt examination of detection limits, normalization, and assay comparability rather than automatic use in a model.

The main alternative is direct measurement of transporter abundance in age-specific tissue. It avoids some extrapolation steps but may be limited by the availability and quality of pediatric tissue samples. A second approach is relative abundance scaling from an in vitro system to adult tissue, followed by an age ratio; this is the product method developed here and is useful when direct pediatric measurements are sparse. Activity-based scaling uses transport activity rather than measured abundance. It may better reflect functional transport under the assay conditions, but activity can also depend on assay-specific conditions and does not isolate abundance from other functional differences. These approaches answer related, not identical, questions.

Worked Example: Treating Assay Disagreement as Structural Uncertainty

Two hypothetical assay platforms produce tissue-to-assay ratios of $0.30$ and $0.18$ for the same transporter. Keep the adult tissue protein yield at $40$ mg/g and the assay abundance at $8.0$ pmol/mg. Use an age ratio of $0.45$ per g tissue to compare the resulting pediatric estimates.

For the first platform, $K=0.30\times40=12.0$ mg/g, giving $8.0\times12.0\times0.45=43.2$ pmol/g. For the second, $K=0.18\times40=7.2$ mg/g, giving $8.0\times7.2\times0.45=25.9$ pmol/g. The second estimate is $25.9/43.2=0.600$, or 40.0% lower than the first. This difference is a cross-platform sensitivity range, not a confidence interval. Until the discrepancy is understood or justified, reporting only one estimate with a narrow measurement interval would conceal an important source of uncertainty.

A common mistake is to multiply an in vitro abundance by a tissue-to-assay factor and then apply an age ratio that already includes the age-specific protein yield, while also separately multiplying by that yield. The reasoning may sound like “each age-related adjustment should be included,” but it counts the same protein-yield change twice. The correction is to define the age ratio precisely: use a per-g-tissue ratio with the adult tissue scaling factor, or use per-mg-protein ontogeny with age-specific protein yield. Do not mix the two formulations.

Even a well-scaled abundance estimate has limits. Assay differences, tissue sampling, antibody or quantification performance, and variation in membrane recovery can create systematic differences not captured by a simple coefficient of variation. Abundance also does not establish the fraction of transporter at the relevant membrane or its transport capacity. The estimate should therefore be reported with its tissue, age, normalization basis, scaling assumptions, and uncertainty sources rather than as a universal pediatric value.

Key takeaway: Pediatric tissue abundance is obtained by multiplying an in vitro abundance by a tissue scaling factor and a correctly defined age ratio. The factors must refer to compatible transporter measurements and protein bases; uncertainty from assay measurement and cross-system mismatch should be made explicit.

Check Your Understanding

Answer each question using consistent units and definitions.

An assay measures $6.0$ pmol/mg. The adult tissue-to-assay ratio is $0.20$, tissue protein yield is $50$ mg/g, and the age ratio per g tissue is $0.40$. What is the age-specific abundance?
Why can an ontogeny ratio measured per mg protein not always be applied directly to an adult estimate in pmol/g tissue?
What assumptions are needed to add log variances when combining uncertainty in assay abundance and tissue scaling?
Two assay platforms give different tissue-to-assay scaling ratios. Why should that disagreement not automatically be described as a confidence interval?

Solutions. (1) $K=0.20\times50=10.0$ mg/g, so abundance is $6.0\times10.0\times0.40=24.0$ pmol/g. (2) The ratio uses a protein-mass denominator; if protein yield per g tissue differs by age, that difference must be included to obtain a per-g-tissue ratio. (3) The inputs must be modeled on the log scale, with independence assumed for simple addition of variances; correlated inputs require covariance terms. (4) Platform disagreement may reflect systematic method differences or structural uncertainty, not repeated-sampling uncertainty summarized by a confidence interval.