📄 medicine

Implications of Ceftazidime-Avibactam Clearance Correlation for Target Selection and Single-Analyte Therapeutic Drug Monitoring During Continuous Infusion: A Pharmacometric Simulation and Individual Patient Data Analysis

This study demonstrates that the clearance correlation between ceftazidime and avibactam in critically ill patients is lower than previously assumed, rendering single-analyte therapeutic drug monitoring less reliable and highlighting the critical impact of target selection on dosing strategies.

Original authors: Vahhab Piranfar

Published 2026-08-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Vahhab Piranfar

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the high-stakes environment of modern hospitals, doctors often turn to powerful antibiotic combinations to fight infections that have learned to resist standard treatments. One such combination pairs two drugs, ceftazidime and avibactam, which work together to dismantle the defenses of dangerous bacteria. To make these drugs work as hard as possible, doctors sometimes administer them as a continuous drip rather than a single shot, keeping a steady level of medicine in the patient's blood. However, because these patients are often critically ill with changing kidney function, the amount of drug their bodies clear out can vary wildly. This creates a difficult puzzle: how does a doctor know if the right amount of medicine is actually reaching the infection? The standard approach has been to measure the level of just one of the two drugs, assuming that if the first one is at the right level, the second one must be too. This assumption relies on the idea that the two drugs are cleared from the body in perfect lockstep, like two runners holding hands and moving at exactly the same speed.

A new study challenges this long-held assumption, revealing that the two drugs do not move in perfect unison after all. The researchers, led by Vahhab Piranfar, set out to test whether measuring just one drug is truly enough to guarantee that the other is also doing its job. They built a massive computer simulation involving one hundred thousand virtual patients with varying levels of kidney function to see how the drugs behaved under different conditions. They also analyzed real data from a small group of twenty-one critically ill adults to measure how closely the clearance rates of the two drugs actually matched. The study found that the two drugs are indeed linked, but not as tightly as previously believed. Instead of moving in near-perfect sync, their clearance rates were found to be correlated at a level of 0.70, significantly lower than the 0.94 value that had been used in medical models for years.

This difference might sound small, but it has a profound impact on patient safety. When the researchers used the older, higher correlation value, they found that measuring only the first drug was a very reliable way to guess the level of the second drug. However, when they applied the newly measured, lower correlation, the reliability dropped. In the new scenario, relying on a single measurement led to a situation where nearly one in ten patients was given a false sense of security. These patients appeared to have enough of the second drug based on the first drug's level, but in reality, they did not. The study suggests that for the most vulnerable patients, checking only one drug might leave them under-treated without the doctor knowing it.

The research also highlighted that the definition of "enough" drug is far more complicated than a single number. The study tested different target levels for the second drug, ranging from a low threshold of 1 milligram per liter to a high one of 8 milligrams per liter. The choice of which target to use completely changed the outcome of the simulations. If doctors aimed for the lowest target, the treatment appeared successful in almost all cases. But if they aimed for the higher, more conservative target, the success rate dropped dramatically, with many patients failing to reach the necessary levels. This means that two doctors looking at the exact same patient data could reach opposite conclusions about whether the treatment is working, simply because they are using different definitions of success.

Furthermore, the study showed that the patient's kidney function plays a massive role in how much drug is needed. The amount of medicine required to hit the target varied by more than five times between patients with poor kidney function and those with very high kidney function. In the most extreme cases, the dose needed to ensure the drug worked was so high that it approached safety limits, suggesting that a standard dose for everyone is not a safe or effective strategy. The researchers also looked at how well the drugs reached the lungs, a common site for these severe infections, and found that the concentration in the lung tissue was significantly lower than in the blood, further complicating the picture of whether the infection is truly being fought.

Ultimately, this work does not declare the current methods broken, but it does expose a hidden weakness in how doctors currently monitor these powerful antibiotics. The study concludes that the old assumption of perfect synchronization between the two drugs is likely too optimistic. Because the drugs do not clear the body at exactly the same rate, measuring just one leaves a gap in knowledge that could be dangerous for critically ill patients. The author suggests that in the future, when precise treatment is vital, doctors should consider measuring both drugs directly rather than guessing based on one. They also emphasize that the medical community needs to agree on what the right target level actually is, as the current lack of consensus makes it difficult to know if a treatment plan is truly adequate. By bringing these uncertainties into the light, the study provides a clearer, more cautious path forward for treating some of the most difficult infections in modern medicine.

Technical Summary: Implications of Ceftazidime-Avibactam Clearance Correlation for Target Selection and Single-Analyte Therapeutic Drug Monitoring

Problem Statement
Continuous infusion of ceftazidime-avibactam is increasingly utilized to maximize pharmacokinetic/pharmacodynamic (PK/PD) target attainment in critically ill adults with carbapenem-resistant Gram-negative infections. However, the implementation of therapeutic drug monitoring (TDM) for this combination faces a critical uncertainty: whether measuring ceftazidime concentration alone is sufficient to infer avibactam target attainment. This inference relies on the physiological correlation between the clearances of the two components. Current literature and models have assumed a high clearance correlation of 0.94, a value derived from non-parametric bootstrapping in registrational analyses rather than direct estimation from paired individual patient data. Furthermore, there is a lack of consensus regarding the optimal pharmacodynamic target for avibactam, with guidelines alternating between fixed thresholds of 1 mg/L, 4 mg/L, and higher values. This ambiguity fundamentally alters the perceived adequacy of dosing regimens.

Methodology
The study employed a dual approach combining pharmacometric simulation with individual patient data (IPD) analysis:

Joint Two-Analyte Model (IPD Analysis): A joint non-linear mixed-effects model was fitted to IPD from 21 critically ill adults (238 paired observations) receiving intermittent infusion during continuous renal replacement therapy. This model estimated the correlation between ceftazidime and avibactam clearance random effects directly, utilizing a first-order conditional estimation method with the Laplace approximation. The correlation was allowed to vary, and confidence intervals were derived via profile likelihood.
Pharmacometric Simulation: A Monte Carlo simulation of 100,000 virtual critically ill adults was conducted across five European Kidney Function Consortium (EKFC) renal-function classes.
- Inputs: The simulation utilized published clearance relationships and variability estimates (with an assumed correlation of 0.94 for the primary scenario).
- Regimens: 11 continuous-infusion regimens were evaluated.
- Targets: Primary targets included a free ceftazidime/MIC ratio $\ge$ 4 and a free avibactam steady-state concentration $\ge$ 4 mg/L. Sensitivity analyses tested avibactam targets ranging from 1 to 8 mg/L.
- Outcomes: The study calculated the joint probability of target attainment (PTA), cumulative fraction of response (CFR), and the limiting component (ceftazidime vs. avibactam).
- TDM Accuracy: The study evaluated the accuracy of predicting avibactam attainment based solely on ceftazidime measurements under different assumed correlations (0.94, 0.703, and bounds) and assay error levels (0% and 20% CV).

Key Results

Clearance Correlation: The estimated correlation between ceftazidime and avibactam clearance in the IPD cohort was 0.703 (95% profile-likelihood CI: 0.380–0.874). The previously assumed value of 0.94 lies outside this interval ( $p = 0.0006$ ), indicating the assumption is not supported by the observed paired data.
Impact of Avibactam Target Selection: The choice of avibactam target drastically altered the perceived efficacy of regimens:
- At a target of 1 mg/L, joint CFR ranged from 93.4% to 98.6%.
- At the primary target of 4 mg/L, joint CFR ranged from 73.5% to 88.6%.
- At a target of 8 mg/L, joint CFR dropped significantly to 44.0%–60.7%.
Limiting Component: The component limiting target attainment shifted based on the MIC and the avibactam target. At lower MICs (e.g., 2 mg/L), avibactam was the limiting component. At the EUCAST clinical breakpoint (8 mg/L), ceftazidime became the limiting component in all regimens.
TDM Reliability: When using the estimated correlation (0.703) instead of the assumed 0.94, the reliability of single-analyte TDM decreased. The rate of "false reassurance" (predicting avibactam attainment when it was not achieved) increased from 3.6% (at $\rho=0.94$ ) to 9.2% (at $\rho=0.703$ ). This misclassification rate rose further as the correlation decreased toward the lower bound of the confidence interval.
Dose Requirements: Model-derived daily doses required to attain targets varied more than five-fold across renal-function classes. In patients with augmented clearance (EKFC 121–150 mL/min/1.73 m²), the median dose required to meet targets at an MIC of 8 mg/L was 8.94 g/day, with only 57.2% of patients remaining below a prespecified 10 g/day cap.

Significance and Claims
The paper claims that the clearance correlation between ceftazidime and avibactam is lower than previously assumed, which has direct implications for clinical decision-making:

Limitations of Single-Analyte Monitoring: Assuming a correlation of 0.94 leads to an overestimation of the reliability of inferring avibactam exposure from ceftazidime levels. With the lower estimated correlation, measuring ceftazidime alone misclassifies avibactam target attainment in nearly 1 in 10 critically ill patients.
Target Sensitivity: The study demonstrates that the selection of the avibactam target (1 vs. 4 vs. 8 mg/L) is a more significant determinant of perceived regimen adequacy than the pharmacokinetic assumptions examined in sensitivity analyses. The 4 mg/L target is presented as a conservative, clearly defined assumption for scenario analysis but is not established as an in vivo exposure-response threshold.
Clinical Recommendation: The author concludes that direct measurement of avibactam warrants evaluation in critically ill patients when precise target attainment is required, particularly given the lower-than-assumed correlation. Alternatively, a more conservative avibactam target should be considered.
Methodological Contribution: This work provides the first clearance correlation for this combination estimated by fitting both analytes jointly with an uncertainty interval, moving the field from inheriting an unverified assumption to having a value that can be justified or challenged by data.

The study maintains modesty regarding its scope, noting that the correlation was estimated in a specific cohort (continuous renal replacement therapy) and that the primary simulation retained the source model's assumptions for comparison. It emphasizes that the results apply to steady-state continuous infusion in critically ill adults not receiving renal replacement therapy and do not account for loading phases or time-varying concentrations.

Technical Summary: Implications of Ceftazidime-Avibactam Clearance Correlation for Target Selection and Single-Analyte Therapeutic Drug Monitoring

More like this