Physiologically-based pharmacokinetics (PBPK) is a valuable resource to support decisions throughout drug development for sponsors and regulators. PBPK is used extensively to predict drug-drug interactions (DDIs), to inform dosing and clinical study design, to predict drug exposure, to predict variation in drug clearance, and to understand mechanisms of drug disposition.
Virtual Twin technology, based on PBPK simulations, creates a computer model of each patient, replicating their attributes that affect drug exposure. These attributes include the patient’s age, weight, height, gender, ethnicity, and genetics/activities of drug metabolizing enzymes and drug transporters (DMETs). The Virtual Twin model can also incorporate the patient’s current drug dosage, fed or fasted state, co-medications that affect the activity of DMETs, and organ function. Virtual Twin technology can therefore be used for model-informed precision dosing (MIPD). It allows clinicians to optimize drug exposure for an individual patient—one that maximizes the chances of therapeutic benefits while minimizing side effects—by evaluating the impact of different drug doses, schedules, and combinations in the patient’s in silico “virtual twin.” Clinicians can have as many “dress rehearsals” as they like in the safe in silico world prior to dosing the real patient.
By adapting PBPK modeling and simulation (PBPK M&S) using the Simcyp® Simulator, we created “virtual twins” to accurately predict olanzapine (OLZ) exposure in individual patients.1 This proof of concept study, conducted at Flinders University in South Australia, was recently published in the British Journal of Clinical Pharmacology. OLZ is an anti-psychotic drug used to treat acute psychosis, schizophrenia, and bipolar disorder (as an adjunct to lithium and valproate). We chose OLZ as a proof of concept drug due to the following attributes:
- It is a BCS Class II compound—high permeability, low solubility—thus kinetics are not expected to be rate-limited by transporters and in vitro-in vivo extrapolation of clearance (IVIVE) can be determined
- OLZ is not metabolized by CYP3A4, thus eliminating any issues associated with variability in enzyme abundance
- The therapeutic index of OLZ has been established (ranging from 20 to 80 ng/ml)
- Additionally, this atypical anti-psychotic is strongly associated with metabolic adverse effects, eg, weight gain and poor lipid profile, which could lead to high discontinuation rates; thus, predicting OLZ exposure may have a marked clinical impact
Building and validating the Simcyp model
We revisited the in vitro kinetics of OLZ metabolism and accounted for nonspecific microsomal binding and albumin effects.2 In addition to CYP1A2, we identified more OLZ metabolizing enzymes than previously investigated, including CYP2C8, which has an important role in demethylation. A thorough evaluation of OLZ in vitro enabled us to define properly the in vitro kinetic parameters that are required to do IVIVE of clearance, which is essential in building an OLZ drug file in the Simcyp Simulator. A minimal PBPK model with first order absorption best recaptured the clinical PK data. The OLZ PBPK model was then validated against single-dose clinical PK studies and a therapeutic drug monitoring (TDM) database of approximately 260 patients.
Applying the Simcyp model
A small clinical trial was conducted on patients who were commencing OLZ treatment. The Simcyp healthy volunteer population file was “individualized” for gender, age, height, weight, CYP2C8 abundance based on genotyping, and CYP1A2 enzyme based on caffeine to paraxanthine ratio, thus creating virtual twins of 14 patients. OLZ systemic exposure predicted in the virtual twins was compared with the drug concentrations measured in the corresponding patients. Those predicted exposures were also used to calculate a hypothetical decrease in exposure variability after the OLZ dose was adjusted.
Model predictions and outcome
The Virtual Twin technology accurately predicted OLZ PK parameters in the healthy Caucasian, Chinese, and geriatric patients included in the single-dose clinical studies. The exciting part was that we could accurately predict OLZ plasma concentrations at steady state in real patients. This opens the door to study the use of Virtual Twin PBPK at the bedside and in clinic. We then showed that a two-fold decrease in exposure variability using MIPD-adjusted dosing could hypothetically be achieved—this is a massive improvement on the exposure variability currently seen with OLZ in clinical practice. Thus, using Virtual Twin Technology is a potentially powerful way that prescribing clinicians could reduce drug exposure variability more broadly. Indeed, by reducing inter-patient exposure variability to OLZ, patients should benefit from an improved risk-benefit profile, reduced side effects, and a greater likelihood of medication adherence.
 Polasek TM, Tucker GT, Sorich MJ, Wiese MD, Mohan T, Rostami-Hodjegan A, Korprasertthaworn P, Perera V, & Rowland A. (2018). Prediction of olanzapine exposure in individual patients using physiologically-based pharmacokinetic modelling and simulation. Br J Clin Pharmacol 84(3), 462–476.
 Korprasertthaworn P, Polasek TM, Sorich MJ, McLachlan AJ, Miners JO, Tucker,GT, & Rowland, A. (2015). In vitro characterization of the human liver microsomal kinetics and reaction phenotyping of olanzapine metabolism. Drug Metab Dispos 43, 1806–1814.
To learn more about how model-based approaches can improve precision dosing in clinical care, watch this webinar.