Blood Never Lies: The PopPK-Based Lie Detector

Area of Science:

• Pharmacokinetics
• Drug Metabolism
• Population Modeling

Background:

• Accurate drug adherence assessment is crucial for unsupervised oral therapies.
• Conventional adherence monitoring tools often lack objectivity and miss non-adherence.
• Population pharmacokinetic (PopPK) modeling offers improved objectivity but may miss hidden non-adherence with single analytes.

Purpose of the Study:

• To develop and validate a bi-analyte PopPK approach for enhanced detection of drug non-adherence.
• To utilize both parent drug and metabolite concentrations to reconstruct patient dosing histories.
• To improve the accuracy of adherence assessment in long-term oral therapies.

Main Methods:

• Identified spironolactone and canrenone as suitable candidates due to distinct pharmacokinetic profiles.
• Simulated virtual patients under various adherence scenarios.
• Employed a modified Metropolis-Hastings algorithm to jointly estimate pharmacokinetic parameters and dosing patterns from simulated concentrations.
• Derived posterior adherence probability and evaluated diagnostic performance using receiver operating characteristic analysis.

Main Results:

• The bi-analyte approach demonstrated superior non-adherence detection compared to single-analyte methods, especially for recent missed doses.
• Trough concentrations proved more effective than peak concentrations for adherence detection, particularly when patients did not overcompensate.
• The method successfully inferred probabilities of daily tablet intake.

Conclusions:

• The probabilistic bi-analyte PopPK approach significantly enhances non-adherence detection.
• This method is adaptable to other drugs with distinct metabolite profiles and low model variability.
• Improved adherence monitoring can lead to better therapeutic outcomes in long-term oral treatments.