Two-Compartment Trafficking and Blood Cell Measurements

Why a Blood Count Is Not the Whole Cell Population

After an infusion, a measured circulating cell count describes the cells present in blood at the sampling time. Cells may also occupy tissues, and exchange between those locations can change the blood count even when the total number of viable cells changes little. A model that includes both locations makes that distinction explicit: blood and tissue are separate cell pools, with directional exchange and potentially different losses.

The model developed here is a linear, first-order description of trafficking. It is useful when the goal is to represent how redistribution and compartment-specific loss shape circulating measurements. It does not, by itself, represent cell division, activation, or time-varying trafficking. Those processes require additional model terms and evidence.

The arrows in the subscripts indicate the direction of movement: \(B\) to \(T\) for \(k_{BT}\), and \(T\) to \(B\) for \(k_{TB}\). The model assumes each compartment is well mixed, the rates are constant over the modeled period, and exchange and loss rates are proportional to the number of cells in the compartment from which they occur. There is no external input after the initial conditions, and no cell production term.

Each equation is a balance of inflows and outflows. Blood cells leave blood through tissue trafficking at rate \(k_{BT}B\) and through blood-compartment loss at rate \(d_BB\); they enter blood from tissue at rate \(k_{TB}T\). The tissue equation has the corresponding terms. A negative contribution in one compartment due to trafficking is a positive contribution in the other; loss, in contrast, removes cells from the modeled total.

Consequences of the Balance Equations

Define \(a=k_{BT}+d_B\) and \(c=k_{TB}+d_T\). These are the total first-order removal rates from blood and tissue, respectively, including both movement out of a compartment and local loss. The equations can then be written as a linear system with matrix \(\begin{pmatrix}-a&k_{TB}\\ k_{BT}&-c\end{pmatrix}\). The two decay rates are the negatives of this matrix's eigenvalues.

For nonnegative rate constants, both decay rates are nonnegative. If there is at least one positive loss rate and both compartments communicate, the rates are positive. They describe the two modes of the coupled system, not the loss rate of either compartment in isolation. In general, the observed blood trajectory is a weighted sum of these modes. A coefficient in that sum may be negative, depending on the initial distribution; the total predicted abundance still follows the balance equations.

The form follows by eliminating \(T\) from the coupled equations. Differentiating the blood equation and substituting both balance equations gives \(\frac{d^2B}{dt^2}+(a+c)\frac{dB}{dt}+(ac-k_{BT}k_{TB})B=0\). The characteristic equation is \(r^2+(a+c)r+ac-k_{BT}k_{TB}=0\), whose roots are \(-\lambda_{\mathrm{slow}}\) and \(-\lambda_{\mathrm{fast}}\). Therefore \(B(t)\) is a linear combination of the two exponentials. The initial blood slope, \(\dot B(0)=-aB_0+k_{TB}T_0\), determines the weights along with \(B_0\):

These expressions satisfy both required initial conditions: the coefficients sum to \(B_0\), and their weighted initial slope is \(-\lambda_{\mathrm{slow}}A_{\mathrm{slow}}-\lambda_{\mathrm{fast}}A_{\mathrm{fast}}=\dot B(0)\). If the two decay rates coincide, the usual limiting solution has a term proportional to \(t e^{-\lambda t}\); the coefficient formulas above apply when the rates differ.

Adding the two balance equations cancels the exchange terms: \(\frac{d(B+T)}{dt}=-d_BB-d_TT\). This is a useful check on the model: trafficking alone redistributes modeled cells, while compartment-specific loss decreases their total. If both loss rates are zero, \(B+T\) is conserved. With positive losses, the total declines, although the blood abundance can temporarily rise if return from tissue exceeds blood outflow and blood loss.

Worked Blood Trajectories and Measurements

A laboratory assay often reports a circulating concentration rather than a total abundance. To compare a compartment model with such data, the observation definition must be explicit. If \(B(t)\) is total circulating abundance in cells and \(V_B\) is the effective blood volume in microliters, the idealized concentration is \(C_B(t)=B(t)/V_B\) cells per microliter. If \(B\) is recorded in millions of cells, this becomes \(C_B(t)=10^6B(t)/V_B\). Assay recovery, sampling, and any conversion from a measured fraction to an abundance may require additional observation factors. Those factors affect the measurement mapping, not the trafficking balance equations themselves.

Checking Assumptions and Comparing Representations

The model is most useful when the compartments are meaningful for the measurement question and first-order exchange is a reasonable approximation over the observed period. It assumes that every cell in a compartment shares the same average rates and that the rates do not change with time, cell state, or compartment occupancy. It also treats the tissue pool as one effective compartment, even though real tissues may have different access and retention properties.

Several checks help identify when these assumptions or the measurement mapping may be inadequate:

• Check units and signs. Every rate must have units of inverse time; each transfer must appear with opposite signs in the two compartment equations. The derivative of total abundance must contain losses but no exchange terms.
• Check plausible states. Predicted abundances should be nonnegative over the modeled interval. A fitted trajectory that requires negative compartment abundance signals an unsuitable model, parameterization, or observation mapping.
• Inspect systematic residual patterns. Persistent curvature or time-dependent disagreement between predicted and measured blood counts can indicate changing rates, more than one effective tissue pool, or an inaccurate conversion from abundance to assay measurement. A flexible fit alone does not establish which explanation is correct.
• Check whether the sampling schedule resolves both modes. Sparse measurements may show only an apparent single decline, especially if the fast mode occurs before the first sample or the slow mode is weak. The two-mode formula does not guarantee that both rates can be estimated from the available observations.

A one-compartment apparent-loss model can be appropriate when the purpose is only to summarize the observed blood decline over a limited interval. It is simpler, but its apparent rate combines the effects of redistribution, return, and blood-specific loss; it generally cannot predict tissue abundance. The two-compartment model represents these processes separately and predicts a coupled trajectory, at the cost of more parameters and stronger assumptions. Directly reporting blood measurements without a mechanistic model is also appropriate when only the observed circulating count is needed and no extrapolation to hidden compartments is intended.

The model's limitations are important. A single effective tissue compartment cannot represent organ-specific trafficking or delayed access. Constant first-order rates may be poor approximations if trafficking changes over time or depends on cell phenotype. The equations also omit cell production, activation-dependent changes, and external sources. Finally, blood-only data may not uniquely determine both exchange rates, both loss rates, and the unobserved tissue state. Additional measurements or justified constraints may be needed; fitting a complicated model to sparse observations does not create information that was not measured.