Deriving the Logistic Tumor Growth Trajectory

From the Logistic Rate Law to a Trajectory

The logistic growth equation specifies an instantaneous growth rate, but tumor data are usually observed as a sequence of burdens over time. Deriving its trajectory connects those views: it gives the burden predicted at any time from an initial condition and clarifies how the model transitions from nearly exponential growth to slowing near a finite upper limit.

Let \(V(t)\) denote tumor burden at time \(t\), measured on the scale to which the model is applied, such as tumor volume in mm³. The parameter \(r\) is the intrinsic growth-rate constant, with units of inverse time, and \(K\) is the carrying capacity, with the same units as \(V\). For the main derivation, assume \(r>0\), \(K>0\), and \(0<V(0)=V_0<K\). The model also assumes that these parameters are constant over the time interval considered and that no treatment or other modeled input changes the growth law.

The differential equation and its solution describe modeled burden, not automatically every quantity measured in an experiment. For instance, a volume measurement may not equal viable tumor-cell burden; the mapping between those quantities is a separate modeling choice. The earlier material on mapping viable tumor burden to measured tumor volume provides that context.

Separating and Integrating the Equation

For \(0<V<K\), both \(V\) and \(1-V/K\) are positive, so the variables can be separated without dividing by zero:

Since \(1/[V(1-V/K)]=K/[V(K-V)]\), partial fractions give \(1/V+1/(K-V)\). Integrating both sides therefore yields:

The minus sign in the second logarithm follows from \(\int 1/(K-V)\,dV=-\ln(K-V)\). Combining logarithms and exponentiating gives \(V/(K-V)=A e^{rt}\), where \(A\) is a positive constant. Applying \(V(0)=V_0\) fixes that constant as \(A=V_0/(K-V_0)\). Solving for \(V\) produces the closed-form trajectory.

This derivation applies to the interior range \(0<V_0<K\). The differential equation also has equilibrium solutions \(V(t)=0\) and \(V(t)=K\); the separated derivation does not apply at those values because it divides by terms that vanish there. If \(V_0=0\), the model remains at zero, while \(V_0=K\) remains at capacity. For a positive initial burden below capacity, the solution stays positive and below capacity at every finite time.

Reading the Solution at the Limits

The closed form makes two important behaviors explicit. First, when \(V/K\) is small, \(1-V/K\approx1\), so the instantaneous rate is approximately \(rV\), the exponential rate law. This is an approximation for a low-burden range, not a claim that logistic growth is exponential at all sizes. The exact solution also approaches an exponential form when \(V_0/K\) is small and the trajectory is considered before substantial capacity limitation.

Second, for \(r>0\), \(e^{-rt}\) tends to zero as \(t\) increases. Thus the denominator in the closed form tends to one and the burden approaches \(K\). The approach is asymptotic: for \(0<V_0<K\), the model does not reach \(K\) at any finite time.

The same conclusion follows from the rate law. Below capacity, \(dV/dt>0\); at capacity, \(dV/dt=0\). The solution increases toward the equilibrium rather than crossing it. A useful equivalent form is the log-odds relationship, which is linear in time under the model:

This transformation can help inspect whether an idealized trajectory has the expected shape when \(K\) is specified. It is not, by itself, a guarantee that a model fits data: measurement error, an uncertain capacity, and departures from constant parameters can all affect the transformed observations.

Shape, Diagnostics, and Failure Modes

Differentiating the rate law once more identifies the trajectory's change in slope:

For a positive trajectory below capacity, this acceleration is positive when \(V<K/2\), zero at \(V=K/2\), and negative when \(V>K/2\). Thus \(K/2\) is the inflection burden and the point of maximum absolute growth rate. This feature follows from the logistic equation; it does not establish that a biological tumor must follow the model.

A practical diagnostic is to compare observed longitudinal burdens with the predicted trajectory on the original measurement scale, while checking that the assumed observation scale is appropriate. If \(K\) is specified, the log-odds form offers an additional shape check: the transformed burden should change linearly with time under an exact logistic model. Systematic curvature, a changing slope, or persistent residual patterns can indicate that the constant-parameter logistic law is inadequate, that the capacity is misspecified, or that the observations do not correspond directly to modeled burden. Transformation can also magnify measurement problems near zero or near \(K\), where the log-odds become extreme.

The main simple alternative is exponential growth, \(dV/dt=rV\), whose solution is \(V_0e^{rt}\). It is useful over a range where burden is small relative to capacity, but it has no finite carrying capacity and continues increasing without bound. The logistic solution adds a specific size-dependent limitation through \(1-V/K\). A numerical solution of the differential equation may be useful when a model is extended so that no closed form is available, but numerical integration does not remove the need to justify the biological assumptions or check predictions against data.

The logistic trajectory is therefore a compact model with a precise interpretation: it predicts early growth close to exponential when burden is small, a maximum growth rate at half capacity, and an asymptotic approach to capacity. These are mathematical consequences conditional on the model, not universal properties of tumor biology. In application, constant parameters, the chosen burden scale, and the observation process all require scrutiny.