Injection-Site Precipitation and Dissolution-Limited Release

Why an Injection Can Produce Delayed Input

An intramuscular dose may leave the syringe as a solution yet form a precipitate after dilution in tissue fluid. A suspension already contains particles that must dissolve. In either case, drug must become locally dissolved before it can cross into systemic circulation. If that dissolution is slow relative to absorption from the local dissolved pool, the injection site acts as a release-limiting depot: systemic input can be much slower and more delayed than a model that absorbs the whole dose directly.

The key modelling distinction is between release from the solid depot and absorption from the dissolved depot. These are separate processes unless absorption is sufficiently rapid that dissolved drug does not accumulate locally. The equations below describe mass at the injection site; the systemic input rate is the amount entering the circulation per unit time.

A Sequential Dissolution–Absorption Model

First consider an injection that leaves a solid depot of mass \(D_s\), with no dissolved drug initially and no local loss. Assume the solid dissolves at a rate proportional to its remaining mass, and dissolved drug is absorbed at a rate proportional to its local mass:

The first equation says that dissolution transfers mass out of the solid pool at rate \(k_dM_s\). The second says that this transfer enters the local dissolved pool, from which drug leaves for the systemic circulation at rate \(k_aM_l\). Thus \(k_dM_s\) is not generally the systemic input rate: it is the rate entering the dissolved pool. The two rates are equal only at times when the dissolved pool is not changing.

Solving the first equation gives \(M_s(t)=D_se^{-k_dt}\). Substituting this into the second equation and solving with \(M_l(0)=0\) gives, for \(k_a\ne k_d\):

When \(k_a=k_d=k\), the corresponding solution is \(M_l(t)=D_skte^{-kt}\), so \(R_{\mathrm{sys}}(t)=D_sk^2te^{-kt}\). These expressions are nonnegative for positive rate constants; when the denominator \(k_a-k_d\) is negative, the difference of exponentials is negative as well.

Peak Timing and Mass Balance

For unequal rates, the peak systemic input occurs where the derivative of \(R_{\mathrm{sys}}(t)\) is zero. Differentiating the expression above and setting the result to zero gives the condition \(k_ae^{-k_at}=k_de^{-k_dt}\). Rearranging yields:

This time is positive for either ordering of \(k_a\) and \(k_d\). At the peak, \(dM_l/dt=0\), so the dissolved pool is momentarily constant and \(k_dM_s=k_aM_l=R_{\mathrm{sys}}\). This equality at the peak does not mean that dissolution flux and systemic input are identical throughout the profile.

With no local loss, all the initial solid mass eventually dissolves and is absorbed. More formally, integrating the local mass balance gives \(M_s(t)+M_l(t)+\int_0^t R_{\mathrm{sys}}(u)\,du=D_s\). As \(t\) tends to infinity, both local pools tend to zero, so cumulative systemic input tends to \(D_s\). If an overall systemic availability fraction \(F\) is applied after the local absorption step, the systemic input and eventual systemic amount are multiplied by \(F\).

Adding Local Loss or Ongoing Precipitation

Dissolved drug may be lost locally, for example through degradation or a route of removal not represented as systemic absorption. With first-order local loss \(k_lM_l\), the balances become:

For a dose initially all in the solid pool and \(k_a+k_l\ne k_d\), the dissolved mass has the same sequential form with \(k_a\) replaced by \(k_a+k_l\). The fraction of dissolved drug that ultimately enters the systemic circulation is \(k_a/(k_a+k_l)\), because absorption and local loss are competing first-order exits from that pool. This fraction is not an additional bioavailability term to apply if the model already represents those pathways explicitly.

If an initially dissolved injection precipitates gradually after entering tissue, a simple extension allows reversible transfer between the local dissolved and solid pools. Under constant pseudo-first-order rates, precipitation at rate \(k_{\mathrm{prec}}M_l\) and dissolution at rate \(k_dM_s\) give:

Precipitation is a transfer between local pools, not a loss of total drug. These equations assume the precipitation rate can be represented by a constant proportional rate; a solubility threshold or concentration-dependent nucleation may require a different formulation. The parameters \(k_{\mathrm{prec}}\) and \(k_d\) should not be interpreted as measured physical constants unless the model and data support that interpretation.

Assumptions, Diagnostics, and Alternatives

The sequential model is useful when its assumptions are plausible: each rate is approximately first order over the simulated range; the depot is adequately represented by one well-mixed solid pool; rate constants do not change with time; and there is no unmodelled input, loss, or return from systemic circulation to the injection site. The no-loss version also assumes that all dissolved drug can ultimately be absorbed. A heterogeneous suspension, changing particle surface area, limited solubility, local inflammation, or changing tissue perfusion may violate these simplifications.

Useful checks are based on the predicted shape and mass balance. A pure solid depot predicts zero systemic input at time zero, a delayed rise, and a later decline. A fitted profile that requires a large instantaneous systemic input may indicate an initial dissolved fraction or a separate rapid component. If measured exposure is lower than the administered amount can support, include a justified loss or availability term rather than allowing a dissolution rate to absorb the discrepancy. If the model predicts negative local mass, systemic input, or cumulative amount, there is an implementation or parameterization error. Also verify that rate constants use the same time unit as the time variable and that mass and mass-per-time quantities are not interchanged.

A direct first-order absorption model is simpler and may be sufficient when dissolution is rapid relative to local absorption, or when the available data do not distinguish these processes. It represents input as a single exponential decline from the available dose and does not separately predict solid mass. The sequential model is more mechanistic about the injection site and can represent a delayed peak, but it has more parameters and requires information capable of distinguishing dissolution from absorption. A lag-time model can shift an otherwise specified input profile, but does not itself represent a solid pool or ongoing dissolution. These alternatives answer different questions; added mechanistic detail is valuable only when it is defensible and useful for the prediction task.