T-Cell Egress, Retention, and Compartment Exchange

Why Exchange Must Be Modelled in Both Directions

A rise in the number of T cells measured in a tumor does not, by itself, show why the rise occurred. More cells may have entered, fewer may have left, or both processes may have changed. A compartment model makes these possibilities explicit by tracking the cell counts in blood, lymphoid tissue, and tumor and assigning a rate to each directional movement. The focus here is movement and cell retention, not the functions T cells perform after arriving in the tumor.

The preceding material on chemokine-driven T-cell recruitment describes entry into tumors. A mass-balanced exchange model adds the complementary point: entry is one side of a flux, and cells may also leave the tumor. If the model omits egress, it treats tumor entry as one-way accumulation unless another loss term is added. That can be a poor representation when cells recirculate between tissues.

This model assumes each compartment is well mixed, so a cell in a given compartment has the same instantaneous transfer rate regardless of its location within that compartment. Rates are nonnegative and may be constant or specified as time-varying functions. In the simplest exchange-only model, cells neither proliferate nor die, and there are no external sources or sinks. These assumptions allow a direct test of mass balance. Later additions such as local cell loss or external input must be written as separate terms rather than concealed inside an exchange flux.

A Three-Compartment Exchange Model

Let \(B(t)\), \(L(t)\), and \(T(t)\) denote T-cell counts in blood, lymphoid tissue, and tumor at time \(t\). Include a directional transfer rate for each connection in both directions: blood to lymphoid tissue and back, blood to tumor and back, and lymphoid tissue to tumor and back. The exchange-only equations are:

Each negative term is an outgoing flux from the compartment on the left; each positive term is an incoming flux. For example, \(k_{BT}B\) leaves blood and enters tumor at the same rate. The rate constants \(k_{BT}\) and \(k_{TB}\) need not be equal: mass balance requires matching each flux between the two equations, not equal rates in opposite directions.

A useful accounting rule is to write every internal flux exactly twice: subtract it from its source equation and add it to its destination equation. This makes conservation visible and helps prevent sign errors. Summing the three equations gives:

Thus, in this exchange-only model, the total \(B+L+T\) remains constant, even while each individual compartment count can change. This is a statement about the model's cell accounting, not a claim that a real organism has no cell production or loss.

For example, the tumor equation with local loss is:

Here \(k_{BT}B+k_{LT}L\) is the total tumor influx, \(k_{TB}T+k_{TL}T\) is egress to blood and lymphoid tissue, and \(d_TT\) is local loss. Egress is not the same process as local death: egress moves cells to another tracked compartment, whereas local loss removes them from this modelled system unless a destination is explicitly represented.

Worked Examples

Retention, Residence Time, and Interpretation

In a tumor equation with constant influx \(I_T\), combined first-order egress \(k_{\mathrm{out}}\), and local loss \(d_T\), the steady-state count is:

The denominator is the total per-cell rate of leaving the counted tumor population, whether by egress or local loss. Lowering egress increases the steady-state tumor count when influx and local loss are fixed. However, retention is a description of reduced departure, not an independent source of cells. If influx is zero and there is no cell production, reducing egress can slow decline but cannot create a rising count.

For a single first-order exit process with rate \(k_{\mathrm{out}}\), the mean time to exit is \(1/k_{\mathrm{out}}\). With competing egress and local loss, the mean time until either event is \(1/(k_{\mathrm{out}}+d_T)\), under constant rates. This distinction matters: a longer residence in the tumor can result from reduced egress, reduced local loss, or both, but those mechanisms have different implications for the number of cells remaining in the tracked system.

Diagnostics, Alternatives, and Limitations

A basic diagnostic is the total-count check. In an exchange-only implementation, compute the sum of the three derivatives. If it is not zero, some internal flux may be missing from one equation or have an inconsistent sign. Check units as well: each term in a count equation must be cells per time. Negative rates, negative counts, or an unexplained total-cell increase indicate a coding, parameterization, or model-structure problem. With sources and local losses, the total derivative should equal the sum of the sources minus the sum of the losses; internal exchange should still cancel.

The model also implies nonnegative counts for nonnegative initial counts when all transfer rates and local loss rates are nonnegative and sources are nonnegative. At a boundary such as \(T=0\), the tumor derivative has no negative tumor-originating terms; with nonnegative influx it cannot point below zero. A numerical simulation that produces negative counts can therefore signal a step-size or solver issue, or an inconsistent model term.

A one-way recruitment model is simpler when tumor entry is the only relevant movement and egress can reasonably be neglected over the time window of interest. It cannot represent recirculation or distinguish a large influx balanced by substantial egress from a smaller influx with strong retention. A bidirectional compartment model is more informative about those balances, but it requires more rates and data to estimate them. A spatial model may be preferable when gradients or distinct tumor regions are central; a single well-mixed tumor compartment cannot resolve where cells are located or how local barriers affect movement.

Tumor counts alone often do not identify influx and egress separately. At steady state, the same count can arise from many combinations of \(I_T\) and \(k_{\mathrm{out}}+d_T\), because only their ratio is directly represented by \(T_{\mathrm{ss}}\). Time-course data, measurements in other compartments, or independent information about particular rates may help distinguish mechanisms, but a mathematical fit does not guarantee that the rates are uniquely determined. Treat estimated exchange parameters cautiously when the observations only constrain net accumulation.