Dosing During Extracorporeal Membrane Oxygenation

Why ECMO Can Change Drug Exposure

Extracorporeal membrane oxygenation (ECMO) adds a circuit to a patient whose physiology may already be changing rapidly. Drug exposure can shift because drug partitions into or binds to circuit materials, the priming volume dilutes blood, organ perfusion and function change, or a circuit component is replaced. These mechanisms do not all act in the same way: an increased apparent distribution volume can lower an early concentration without changing systemic clearance, whereas ongoing drug removal can increase the maintenance input needed to sustain a target concentration.

A dose adjustment should therefore begin by asking what changed, when did it change, and which part of the concentration-time profile matters? A loading dose is principally tied to the amount needed to achieve a target concentration in the accessible distribution space. A maintenance regimen is tied to the rate at which drug is removed over time. Confusing these questions can lead to an unnecessarily large loading dose or a maintenance regimen that does not replace circuit loss.

Loading Dose: Distribution and Early Circuit Loss

For an intravenous bolus in a linear one-compartment model, with immediate mixing, no loss during administration, and a specified target concentration \(C_{\mathrm{target}}\), the loading dose is \(LD=V C_{\mathrm{target}}\). This calculation assumes that the chosen \(V\) represents the relevant accessible distribution volume. If ECMO priming or fluid shifts increase that volume, the same dose produces a lower initial concentration. If drug is captured by the circuit during loading, the administered dose must also supply the amount captured, if that amount can be estimated.

A useful amount balance makes the distinction explicit. During an infusion at rate \(R\), the amount in the accessible system changes through input, systemic elimination, and circuit loss. If \(R_{\mathrm{seq}}(t)\) is the circuit removal rate, then \(dA/dt=R-CL\,C-R_{\mathrm{seq}}(t)\). The term \(R_{\mathrm{seq}}(t)\) may change over time: adsorption can be greatest soon after a new circuit is exposed to drug and then decline as binding sites become occupied. In that situation, treating sequestration as a permanent fixed clearance may be misleading.

Maintenance Dosing: Keep Circuit Loss in the Balance

For a stable linear system with no ongoing circuit loss, the average input rate at steady state equals systemic elimination. For an IV continuous infusion targeting an average concentration \(C_{\mathrm{avg,target}}\), this gives \(R_{\mathrm{maint}}=CL\,C_{\mathrm{avg,target}}\). For intermittent IV dosing with interval \(\tau\), the corresponding average dose rate is \(Dose/\tau=CL\,C_{\mathrm{avg,target}}\). These familiar equations do not account for an additional ongoing circuit sink.

If circuit removal is a constant amount per unit time over the period of interest, steady state requires the input to replace both systemic elimination and that circuit loss. If instead circuit removal is proportional to concentration, it can be represented as \(R_{\mathrm{seq}}=CL_{\mathrm{seq}}C\), giving a combined effective clearance \(CL+CL_{\mathrm{seq}}\) for that model. The appropriate expression depends on the loss mechanism; a transient adsorption process should not be assumed to persist indefinitely at a constant rate.

The same balance applies to intermittent dosing when the target is an average concentration over a dosing interval and the system is at steady state: the average input rate is \(Dose/\tau\). It does not, by itself, select a dose that achieves a desired peak or trough. Those targets depend on the concentration-time profile and require an appropriate pharmacokinetic model.

Perfusion, Organ Function, and Circuit Changes

ECMO can coincide with changes in cardiac output, hepatic and renal perfusion, inflammation, organ function, fluid balance, and protein binding. The effect on a particular drug depends on its elimination pathway and extraction characteristics; it is not valid to apply one universal “ECMO clearance factor.” For a high-extraction hepatic drug, a reduction in hepatic blood flow can reduce hepatic clearance. For a low-extraction drug, intrinsic metabolic capacity and the unbound fraction may be more influential. Renal drug elimination can also change with renal perfusion, injury, and renal replacement therapy. A measured change in total concentration may additionally reflect altered binding rather than a proportional change in unbound exposure.

Circuit changes create a separate time-dependent problem. Replacing a drug-containing circuit with a drug-free one may remove drug in circuit fluid or material, and a new priming volume may dilute the accessible drug. The size and direction of the concentration change depend on the amount in the patient, the amount in the circuit, the replacement volume, and how quickly the compartments mix. A simple calculation can clarify the arithmetic, but it is only as reliable as those assumptions.

Derivation, Diagnostics, and Limitations

The maintenance equations follow directly from conservation of drug amount. At steady state, the amount in the modeled system is not changing, so the derivative in the amount balance is zero. Substituting zero into the balance and rearranging gives the required input. This simple justification also identifies the common error: if circuit removal is included in the model, it must remain in the steady-state equation.

Potential warning signs include an unexpected concentration decline after circuit initiation or exchange, concentrations that change despite an unchanged dose, and a mismatch between predicted and observed concentration-time patterns. These signals do not identify the cause on their own. Sampling time, dose history, circuit configuration, fluid balance, organ function, protein binding, and assay measurement of total versus unbound drug should be checked before changing a dose.

A practical alternative to a single mechanistic calculation is to use therapeutic drug monitoring and adjust from observed concentrations. Monitoring can help detect a poor prediction, but it does not remove the need to specify the target, sample timing, and whether the measurement represents total or unbound concentration. A model-based approach can separate volume, systemic clearance, and circuit loss when data support those components; a simple empirical adjustment may be more defensible when the circuit effect is poorly characterized. Neither approach makes a transient sequestration process equivalent to a stable clearance without evidence.

Common mistake: “The target is 4 mg/L and systemic clearance is 3 L/h, so 12 mg/h must maintain the target.” That conclusion omits any ongoing circuit sink. If the model also specifies a constant circuit loss of 2 mg/h, the balance is \(R=3(4)+2=14\) mg/h. Conversely, adding a fixed loss term when adsorption has already saturated can overestimate the long-term input. State the loss assumption, use it only over the period it represents, and revise it when circuit conditions or concentration data change.

All of these calculations rely on simplified models. The apparent volume may not be a single fixed compartment; sequestration may depend on drug properties, circuit materials, flow, and time; and organ function can change during support. A calculated dose is therefore a model-based starting point, not a substitute for clinical assessment and appropriately timed concentration monitoring when available.