Effector-to-Target Ratios and Saturable Tumor Killing

Why the Effector-to-Target Ratio Matters

A T cell can kill a tumor cell only when effective encounters occur. The same number of cytotoxic T cells may therefore imply different killing pressure in tumors of different sizes. A useful model summary is the effector-to-target ratio: the abundance of cytotoxic effector T cells relative to the abundance of tumor targets in the compartment where killing is represented. This ratio lets a model express how per-target killing changes as T cells become scarce or abundant relative to tumor cells.

This tutorial compares two functions of that ratio. A linear function predicts that per-target killing continues to rise in direct proportion to the ratio. A saturable function predicts an approximately linear rise at low ratios but a finite upper limit at high ratios. These are alternative model assumptions, not universal laws: their usefulness depends on the biological setting and the range of observations being represented.

Define the Ratio and the Killing Rate

Let \(E\) be the abundance of cytotoxic effector T cells and \(T\) the abundance of tumor cells in the same modeled compartment. They may be counts or concentrations, but they must use compatible units and refer to the same compartment. Define \(r=E/T\) for \(T>0\). The ratio is dimensionless when both abundances use the same units.

Definition: The per-target death hazard, \(\lambda(r)\), is the instantaneous tumor-cell death rate per living tumor cell at effector-to-target ratio \(r\). Its units are inverse time. For a short interval over which this hazard is constant, the fraction of tumor cells dying is \(1-\exp[-\lambda(r)\Delta t]\), where \(\Delta t\) is the interval length.

The constant-hazard expression follows from the survival equation \(dT/dt=-\lambda T\): over an interval of length \(\Delta t\), the surviving fraction is \(\exp(-\lambda\Delta t)\), so the fraction that dies is one minus that quantity. This calculation assumes no tumor-cell growth or other gains during the interval and a hazard that does not change over it.

Linear and Saturable Functions

A linear ratio model assigns each target a death hazard proportional to the ratio:

\lambda_L(r)=k_Lr

Here \(k_L\) has units of inverse time per unit of ratio. The model predicts zero killing at \(r=0\), and it has no upper bound as \(r\) increases. Multiplying the per-target hazard by the number of targets gives the total tumor-cell loss rate:

-\frac{dT}{dt}\bigg|_{\mathrm{kill}}=\lambda_L(r)T=k_LE

Thus, when \(E\) is held fixed, the linear ratio function is equivalent to a mass-action total kill rate proportional to the number of effectors. That equivalence is useful, but it also exposes a limitation: it predicts the same total killing rate even as the target population approaches zero, unless the model is constrained or the functional form is changed.

A common saturable function is:

\lambda_S(r)=k_{\max}\frac{r}{r+r_{50}}

Here \(k_{\max}\) is the maximum per-target death hazard, with units of inverse time. The parameter \(r_{50}\) is the ratio at which the hazard is half its maximum. The total tumor-cell loss rate is \(k_{\max}ET/(E+r_{50}T)\). When \(E=0\), the function gives zero killing; as the ratio becomes very large, the per-target hazard approaches \(k_{\max}\), rather than increasing without bound.

Interpretation: In the saturable function, \(r_{50}\) is a ratio, not a cell count. At \(r=r_{50}\), \(\lambda_S=k_{\max}/2\). The maximum \(k_{\max}\) is a per-target hazard, not the maximum number of tumor cells killed per unit time.

At low ratios, the saturable function is approximately linear. Rewriting it makes the approximation clear:

\lambda_S(r)=\frac{k_{\max}}{r_{50}}r\frac{1}{1+r/r_{50}} \approx \frac{k_{\max}}{r_{50}}r \quad\text{when } r\ll r_{50}

The low-ratio slope is therefore \(k_{\max}/r_{50}\). A linear and a saturable model can be matched in this low-ratio region by setting \(k_L=k_{\max}/r_{50}\). They will still differ at higher ratios: the linear hazard keeps rising, while the saturable hazard bends toward its ceiling.

Worked Example: Comparing Low- and High-Ratio Predictions

Consider a hypothetical tumor-cell model with \(k_L=0.040\) per hour per ratio unit, \(k_{\max}=0.12\) per hour, and \(r_{50}=0.50\). Compare predictions at ratios \(r=0.20\) and \(r=2.0\), assuming each ratio and its associated hazard remain constant for six hours.

At \(r=0.20\), the linear hazard is \(0.040(0.20)=0.0080\) per hour. The saturable hazard is \(0.12(0.20)/(0.20+0.50)=0.0343\) per hour. The corresponding six-hour death fractions are \(1-\exp[-0.0080(6)]=0.0469\), or 4.69%, for the linear model and \(1-\exp[-0.0343(6)]=0.186\), or 18.6%, for the saturable model.

At \(r=2.0\), the linear hazard is \(0.040(2.0)=0.080\) per hour. The saturable hazard is \(0.12(2.0)/(2.0+0.50)=0.0960\) per hour. Over six hours, the predicted death fractions are \(1-\exp[-0.080(6)]=0.381\), or 38.1%, and \(1-\exp[-0.0960(6)]=0.438\), or 43.8%, respectively.

These particular parameters do not match the functions at low ratios, so the saturable model predicts a larger hazard at both ratios shown. The general comparison is about shape, not which function must predict more killing at every ratio: the low-ratio slopes and parameter estimates determine that.

Calculations with a Saturable Hazard

The next example illustrates the parameter meanings directly. As above, a death fraction calculated from a fixed hazard requires the ratio to be approximately constant over the stated interval. This is a reasonable approximation only when changes in \(E\) and \(T\) over that interval are small or when the ratio is maintained.

Worked Example: Using the Half-Saturation Ratio

Suppose a hypothetical model has \(k_{\max}=0.15\) per hour and \(r_{50}=0.30\). At an effector abundance of 1,200 cells and a target abundance of 6,000 cells, calculate the hazard and the four-hour death fraction under a constant-ratio assumption.

The ratio is \(r=1{,}200/6{,}000=0.20\). The per-target hazard is:

\lambda_S(0.20)=0.15\frac{0.20}{0.20+0.30} =0.0600\ \text{per hour}

This is below half the maximum because \(0.20<r_{50}=0.30\). With a constant hazard for four hours, the predicted death fraction is \(1-\exp[-0.0600(4)]=1-\exp(-0.24)=0.213\). The model therefore predicts that 21.3% of the tumor cells present at the start of this interval die from the modeled killing process. It does not predict that 21.3% of all cells in the body die.

When the Ratio Changes During Tumor Killing

In a dynamic model, \(r\) generally changes as effectors and targets change. For example, if \(E\) stays fixed while \(T\) declines, the ratio rises, and the saturable per-target hazard rises toward \(k_{\max}\). It is then not correct to apply a single initial hazard for an entire interval unless that is an explicit approximation.

The changing-ratio consequence can be derived for a restricted case: no tumor growth, a fixed effector abundance \(E>0\), and saturable killing. The tumor equation is \(dT/dt=-k_{\max}ET/(E+r_{50}T)\). Separating variables and integrating from \(T_0\) at time zero to \(T\) at time \(t\) gives:

dt=-\frac{1}{k_{\max}}\left(\frac{1}{T}+\frac{r_{50}}{E}\right)dT

k_{\max}t=\ln\left(\frac{T_0}{T}\right)+\frac{r_{50}}{E}(T_0-T)

Both terms on the right are nonnegative when \(0<T\leq T_0\), consistent with a positive elapsed time. The logarithmic term reflects the increase in per-target hazard as the target population falls; the second term accounts for the finite initial effector-to-target ratio. This closed-form relation is specific to the stated assumptions. With tumor growth, changing effectors, or additional cell compartments, the corresponding system generally requires numerical solution.

Worked Example: Solving for Time Under a Changing Ratio

Assume, hypothetically, \(E=1{,}000\) effector cells remain fixed, there is no tumor growth, \(k_{\max}=0.10\) per hour, and \(r_{50}=0.20\). Starting from \(T_0=5{,}000\) tumor cells, calculate how long the saturable model takes to reach \(T=4{,}000\) cells.

Substitute into the integrated equation:

t=\frac{\ln(5{,}000/4{,}000)+(0.20/1{,}000)(5{,}000-4{,}000)}{0.10}

The logarithmic term is \(\ln(1.25)=0.2231\), and the second term is \(0.20\). Thus \(t=(0.2231+0.20)/0.10=4.231\) hours. The ratio rises from \(1{,}000/5{,}000=0.20\) initially to \(1{,}000/4{,}000=0.25\) at the specified endpoint. The calculation includes this changing ratio; it is not the same as applying the initial hazard unchanged for 4.231 hours.

Assumptions, Diagnostics, and Model Choice

Both functions treat the modeled effector and target populations as sufficiently well mixed within the compartment, and summarize variation in encounters and killing with an average rate. Their interpretation also assumes that the \(E\) and \(T\) variables represent cells capable of participating in the modeled interaction. If the measured cell counts include cells that are not functionally active or not in contact, the fitted parameters may absorb those discrepancies rather than represent intrinsic killing kinetics.

A linear ratio function may be useful over a restricted range when proportionality is a reasonable approximation or when a simple mass-action form is desired. A saturable function is useful when evidence supports diminishing increases in per-target killing as effectors become abundant. A constant per-target hazard independent of \(E/T\) is another alternative, but it cannot represent changes in killing pressure as the ratio changes. Choosing among these alternatives is a biological and data-supported modeling decision; a saturable form should not be selected solely because it seems more realistic.

Useful diagnostics include plotting observed or model-predicted killing rates against the ratio, checking whether residual patterns systematically change at high ratios, and comparing predictions over the range actually supported by data. If observations cover only low ratios, \(k_{\max}\) and \(r_{50}\) may be difficult to estimate separately: the data mainly identify their ratio, \(k_{\max}/r_{50}\), which is the low-ratio slope. A fitted plateau far outside the observed range is an extrapolation, not direct evidence of saturation.

A common mistake is to interpret \(k_{\max}\) as a maximum total killing rate and conclude that no more than \(k_{\max}\) tumor cells can die per hour. The correction is that \(k_{\max}\) limits the per-target hazard. The total rate is \(\lambda_ST\), so it also depends on the number of targets. Another mistake is to hold the initial ratio fixed while applying the hazard across a long interval even though tumor abundance changes substantially. In a dynamic system, update the ratio as the state changes or solve the coupled equations; a fixed-ratio death fraction is justified only under its stated approximation.

Key takeaway: A linear ratio function has an unbounded per-target hazard and a total kill rate proportional to effector abundance. A saturable ratio function is approximately linear at low ratios and approaches a finite per-target hazard at high ratios. The ratio must be defined in a shared compartment, and its evolution must be accounted for when predicting tumor dynamics.

Check Your Understanding

Answer each question, then check the concise solutions below.

If \(E=300\) cells and \(T=1{,}500\) cells in the same compartment, what is \(r\)?
For \(k_{\max}=0.20\) per hour and \(r_{50}=0.40\), what is the saturable hazard at \(r=0.40\)?
At low ratios, what linear slope approximates a saturable function with \(k_{\max}=0.12\) per hour and \(r_{50}=0.30\)?
Why can a fixed-hazard death fraction be inaccurate when effectors remain fixed and tumor cells decline?

Solutions. (1) \(r=300/1{,}500=0.20\). (2) At \(r=r_{50}\), the hazard is half maximum: \(0.20/2=0.10\) per hour. (3) The low-ratio slope is \(k_{\max}/r_{50}=0.12/0.30=0.40\) per hour per ratio unit. (4) The ratio \(E/T\) rises as \(T\) falls, so the per-target hazard changes; a single fixed hazard omits that change unless the interval is short enough for the ratio to remain approximately constant.