Diagnosing Antibody Tissue-Distribution Parameters

Why antibody distribution parameters are hard to diagnose

A plasma concentration-time profile can be measured densely while tissue concentrations remain sparse or unavailable. That does not mean the plasma profile identifies how an antibody enters a tissue, how lymph returns it, or how much is salvaged by FcRn. Several combinations of processes can produce similar plasma curves. The practical question is therefore not only whether a model fits, but whether the available observations distinguish the parameters that generated the fit.

For this tutorial, consider a well-mixed tissue interstitial space with fixed volume \(V_i\), concentration \(C_i\), and amount \(A_i=V_iC_i\), exchanging antibody with plasma at concentration \(C_p\). Let \(PS\) be the permeability-surface area product, \(J_v\) the net fluid filtration rate from plasma to interstitium, \(\sigma\) the vascular reflection coefficient, and \(L\) the lymph flow returning interstitial fluid to plasma. These are idealized lumped quantities; in a spatially heterogeneous tissue they are effective parameters.

Interpretation of these terms depends on what is observed. A model might identify a combination of parameters, such as an effective exchange rate, without identifying each component separately. Also, fitting a parameter successfully is not proof of identifiability: an optimizer can return a number even when many different values fit nearly equally well.

A consistent tissue balance and its implications

For a fixed interstitial volume, fluid entering by filtration must be balanced by fluid leaving. In this simple model, lymph is the only interstitial fluid outflow, so the fixed-volume assumption requires \(L=J_v\). Write their common value as \(J\). The antibody entering by diffusion and solvent drag is \(PS(C_p-C_i)+(1-\sigma)JC_p\); lymph removes antibody at rate \(JC_i\). Thus the net antibody flux from plasma into tissue is:

Since \(A_i=V_iC_i\) and \(V_i\) is fixed, \(dA_i/dt=V_i\,dC_i/dt\). Therefore:

This equation explicitly accounts for both fluid balance and antibody balance. A vascular amount balance must include the same net exchange flux with the opposite sign; systemic input and elimination are additional terms. In a model with changing interstitial volume, this concentration equation would not apply as written: the amount balance would instead include \(d(V_iC_i)/dt=V_i\,dC_i/dt+C_i\,dV_i/dt\).

If \(C_p\) is held constant long enough for the tissue to reach steady state, setting \(dC_i/dt=0\) gives:

A single steady ratio constrains a combination of permeability and lymph flow, not necessarily both separately. The transient response provides another feature: for constant \(C_p\), the approach to steady state has rate constant \(k=(PS+J)/V_i\). That adds information about the sum \(PS+J\), provided \(V_i\) is known and the tissue follows this one-compartment model. Neither result guarantees that real data contain enough information to estimate the parameters precisely.

Use the time course, not just the endpoint

When plasma concentration is approximately constant during tissue equilibration, the equation above has the solution \(C_i(t)=C_{i,\mathrm{ss}}+[C_i(0)-C_{i,\mathrm{ss}}]e^{-kt}\), where \(C_{i,\mathrm{ss}}\) is the steady concentration and \(k=(PS+J)/V_i\). This follows by subtracting the steady-state equation from the dynamic equation: \(d(C_i-C_{i,\mathrm{ss}})/dt=-k(C_i-C_{i,\mathrm{ss}})\). Thus the ratio informs the balance of inward and outward exchange, while the time scale informs their combined rate relative to volume.

Separate distribution from FcRn salvage

FcRn-mediated salvage can alter antibody persistence by rescuing internalized IgG from degradation. Plasma persistence is also affected by distribution: antibody entering tissue can later return to plasma through lymph. Consequently, a long plasma terminal phase is not, by itself, a specific measurement of either FcRn salvage or tissue return. Sparse tissue sampling, residual vascular blood in a biopsy, or measurements that combine interstitial and cellular antibody can further obscure the source of a signal.

Useful discrimination comes from measuring different parts of the system rather than asking one plasma curve to identify every mechanism. Paired plasma and tissue measurements constrain distribution more directly. Measuring interstitial antibody separately from total tissue antibody helps distinguish extracellular exchange from cellular uptake or retention. A perturbation that changes FcRn function, accompanied by suitable controls and relevant tissue measurements, can test the salvage explanation. Such a perturbation is informative only if its effects on other processes, including tissue exchange and systemic physiology, are considered.

Diagnostics, alternatives, and limitations

A practical identifiability assessment asks how the predictions change when a parameter changes while the others are allowed to adjust. If a wide range of \(PS\) and \(J\) values gives almost the same fit, the data do not strongly distinguish them. Profile likelihoods, parameter-correlation summaries, and prediction comparisons can reveal this pattern; a very broad or flat profile is a warning against reporting a precise individual estimate. These checks do not repair a misspecified model, so they should be paired with visual checks of plasma and tissue residuals over time.

Common warning signs include a fitted parameter at a boundary, large changes in estimates from small changes in starting values, implausibly large parameter uncertainty, and systematic residual patterns such as early and late tissue concentrations being underpredicted in opposite directions. Those patterns may indicate weak information, but they can also signal a structural problem: a single well-mixed tissue may be inadequate, or measured “tissue concentration” may include vascular blood or bound and intracellular antibody.

The main alternatives contribute different kinds of information. Plasma-only fitting can characterize systemic exposure but may leave tissue exchange and FcRn effects confounded. Paired plasma and tissue time courses add direct distribution information, but tissue assays must be interpreted in light of vascular contamination and the sampled tissue fraction. Independent physiological measurements of fluid or lymph flow can constrain a flow parameter, although their relevance to the modeled tissue must be justified. Targeted perturbations can be more discriminating than adding more samples under an unchanged condition, but only when their effects are sufficiently specific and the design includes measurements that verify those effects.

No single strategy guarantees identification. A ratio measured at one time can miss transient exchange; a terminal plasma slope can reflect several processes; and measurements from one tissue do not establish that its parameters apply to all tissues. The appropriate conclusion may be that the data identify a combination or a predicted concentration profile, but not each mechanistic parameter. Reporting that limitation is more informative than presenting an unstable estimate as a precise physiological quantity.

A disciplined interpretation workflow