Steady-State Versus First-Dose Target Attainment

Why the Treatment Horizon Matters

A regimen can produce different target attainment after its first dose and after repeated dosing. If drug remains in the body when the next dose is given, concentrations accumulate. A dose that does not reach an immediate-effect target may still reach a chronic-treatment target after accumulation; conversely, a target intended to be reached promptly should not be evaluated only after steady state has developed.

The comparison is meaningful only after specifying what quantity is targeted, when it is assessed, and which treatment horizon the target represents. This tutorial uses concentration at a stated time as a simple example. The same planning principle applies when a dose-selection analysis uses another already-defined PK/PD index, but the calculations for those indices are not interchangeable.

A Simple Accumulation Model

To derive the comparison, consider a hypothetical one-compartment model with intravenous bolus doses. Let \(x=D/V\) be the concentration increment from one dose, in mg/L; let \(k\) be the elimination rate constant, in h\(^{-1}\); and let \(\tau\) be the dosing interval, in hours. Assume \(x>0\), \(k>0\), identical doses at intervals of \(\tau\), and linear, time-invariant elimination. Define \(q=e^{-k\tau}\), so \(0<q<1\). A single dose produces concentration \(x e^{-kt}\) at elapsed time \(t\) after dosing.

After \(m\) doses have been administered at times \(0,\tau,\ldots,(m-1)\tau\), each earlier dose contributes a decayed amount to the concentration. At elapsed time \(t\), where \(0\leq t<\tau\) after the most recent dose, superposition gives:

The finite geometric sum is valid because \(q\ne1\). Since \(0<q<1\), \(q^m\) tends to zero as \(m\) increases. Thus the steady-state concentration at elapsed time \(t\) after a dose is:

Immediately before the next dose, \(t\) approaches \(\tau\). The first-dose trough is \(xq\), whereas the steady-state trough is \(xq/(1-q)\). After \(m\) doses, the pre-next-dose trough is \(xq(1-q^m)/(1-q)\), which is \(1-q^m\) of the steady-state trough. These results show why steady state is approached gradually and why its concentration is not generally the same as the first-dose concentration.

Define the Target Before Simulating

For individual \(i\), let \(\theta_i\) denote the individual's PK parameters and covariates, and let \(C_i^{(h)}\) denote the simulated concentration under horizon \(h\), where \(h\) is either first dose or steady state. A target rule must specify the quantity and assessment time. For example, an immediate target might require \(C_i^{(\mathrm{first})}(1\text{ h})\geq L_{\mathrm{imm}}\). A chronic target might instead require \(C_i^{(\mathrm{ss})}(\tau^-)\geq L_{\mathrm{ch}}\). The thresholds and times should come from the intended clinical interpretation, not from whichever calculation is easiest.

Here \(I_i^{(h)}=1\) when simulated individual \(i\) meets that horizon's target rule and is zero otherwise; \(N\) is the number of simulated individuals. For a fair horizon comparison, use the same simulated individuals and their same parameter draws at both horizons. This paired calculation isolates the impact of the treatment horizon from random differences between two independently generated virtual populations. The simulated population and target rule must also match the dose-selection estimand.

Use Clinically Matched Rules

Immediate and chronic targets can differ in both assessment time and threshold. An immediate target is evaluated on the first-dose profile at the clinically relevant early time. A chronic target is evaluated under the maintained regimen, at the time or summary specified for chronic effect. A steady-state threshold should not be substituted for an immediate target merely because simulated concentrations are higher after accumulation. Nor does a first-dose failure imply chronic failure.

The next example uses different, explicitly defined rules to show why PTA values alone do not tell the whole comparison. Because the rules differ, the difference in their estimates is not a pure measure of accumulation; it reflects both the horizon and the clinical target definition.

Simulation Workflow and Diagnostics

In a population simulation, each replicate represents a virtual individual generated from the intended covariate and random-effects distributions. For each individual, calculate or simulate the first-dose outcome and the repeated-dose outcome, apply the relevant target rule, and then average the resulting indicators. If a steady-state value is computed analytically, verify it against a sufficiently long repeated-dose simulation for selected individuals. In this linear example, both approaches should agree up to numerical precision.

Useful checks include confirming that steady-state concentrations are no lower than first-dose concentrations at the same elapsed time under this positive, linear accumulation model, and confirming that the repeated-dose profile approaches its analytical steady-state value as doses are added. Check that a trough is measured immediately before a scheduled dose, not after it. Inspect unusually large or small simulated values and verify that the parameter and covariate draws are plausible for the model. If profiles fail to converge or the expected accumulation relationship fails, investigate dose timing, indexing, units, model assumptions, and simulation code before interpreting PTA.

Alternatives, Limitations, and a Common Pitfall

An analytical steady-state formula is efficient and exact under the stated model. Direct simulation over many dosing intervals is more flexible when the model has additional structure, but requires a defensible criterion for having reached steady state and additional computation. In either approach, the target must match the clinical question. More replicate subjects reduce Monte Carlo error in the estimated PTA; they do not fix a misspecified PK model or an inappropriate target.

The formulas above describe a one-compartment IV bolus model, not every dosing route or disposition pattern. For more complex but linear models, accumulation can still be represented through contributions from prior doses, but the one-exponential expressions do not generally describe the full profile. Nonlinear PK can invalidate superposition. Time-varying physiology, nonadherence, and changes in dose can also make a fixed-regimen steady state an unsuitable representation. Such cases require simulations consistent with the more complex model and regimen.