Individualization of piperacillin dosage based on therapeutic drug monitoring with or without model-informed precision dosing: a scenario analysis

Abstract

Background: Model-informed precision dosing (MIPD) combines population pharmacokinetic knowledge with therapeutic drug monitoring (TDM) to optimize dosage adjustment. It could improve target concentration attainment over empirical TDM, still widely practised for broad-spectrum antibiotics.

Objectives: To evaluate the respective performance of TDM and MIPD in achieving target piperacillin exposure.

Methods: Measurements from 80 courses of intermittent piperacillin infusions, each with two TDM samples, were retrospectively submitted to our MIPD software TUCUXI. We considered six dosage adjustment strategies: identical dosage for all (4000 mg q8h), actual initial dosage (chart-based), actual empirical adjustment following first TDM, a priori MIPD-based dosage, a posteriori MIPD-based adjustment after first TDM and MIPD including both TDM measurements. Dosing strategies were compared regarding daily dosage, trough levels distribution and PTA (with target trough 8-32 mg/L).

Results: Median trough concentration fell within 8-32 mg/L for all strategies except a priori MIPD-based dosage (42 mg/L). Distributions of trough concentrations predicted with the six dosage adjustment strategies showed significant differences, with both a posteriori MIPD-based strategies best reducing their standard deviation (P < 0.001). PTA of 32%, 32%, 55%, 29%, 83% and 94% were estimated, respectively for the six strategies (P < 0.001). Poor performance of a priori MIPD-based dosage did not hinder a posteriori MIPD-based strategies from significantly improving target attainment.

Conclusions: Whilst empirical TDM improves exposure standardization and target attainment compared with no TDM, MIPD can still bring further improvement. Prospective trials remain warranted to confirm MIPD benefits not only on target attainment but also on clinical endpoints.

Figure 1.

Derivation of the six dosing strategies studied, namely uniform dosage of 4000 mg q8h in all patients (strategy 1U), actual initial dosage based on a reference dosing chart (strategy 1A), actual empirical dosage adjustment following first TDM (strategy 2A), a priori MIPD-based dosage (strategy 1P), a posteriori MIPD-based adjustment after first TDM (strategy 2T) and MIPD-based recommendation based on both TDM measurements (strategy 3T).

Figure 2.

Piperacillin concentrations measured on first and second TDM. The variance ratio (F test) on log-transformed values is 64% (****P < 0.001). The single point below the lower limit of quantification (LLOQ = 0.08 mg/L) was treated as LLOQ/2.

Figure 3.

Concordance analysis between first TDM measurements and a priori predictions (left panel) and between second TDM measurements and a posteriori predictions (right panel). Dotted line: identity; dashed line: log-linear regression. The single point below the lower limit of quantification (LLOQ = 0.08 mg/L) was treated as LLOQ/2 (left panel).

Figure 4.

Predicted piperacillin trough concentrations with all six dosing strategies. Median concentration and proximity to 16 mg/L of scenario 1P are significantly higher in comparison with the five other dosing strategies (****P < 0.001). Significant differences in variance between the six strategies (****P < 0.001) with post hoc comparisons grouping strategies 1U and 1A together, strategies 2A, 1P and 2T together and leaving strategy 3T alone (see Text and Table S8).

Figure 5.

Probability of target attainment, defined as the percentage of total trough concentrations predicted in the 8–32 mg/L range, for the six dosing strategies. White and grey scares illustrate total trough concentrations predicted below and above this range.