Physiologically-based pharmacokinetics (PBPK) is a mechanistic modeling approach used to inform many critical R&D decisions including clinical trial design, first-in-human dosing, formulation design, treating special populations, and predicting drug-drug interactions (DDIs). This approach can be used for both small molecule and biologic drug development as evidence of its increased use in protein-based therapeutic drug development.
Cancer is the second leading cause of death worldwide. By 2022, the oncology therapeutic drug market is estimated to be at $200B, which reflects a significant R&D spend in pharma in this area of drug development. Traditionally, drug development is focused on determining exposure in healthy volunteers. However, the demographics and physiology of cancer patients differs from healthy individuals, and consequently, the PK of drugs may be altered. Cancer patients face an increased risk of DDIs due to the co-administration of multiple therapies. Thus, PBPK modeling using a virtual oncology population can help accelerate developing safe and efficacious oncology drugs.1
This blog focuses on the benefits and challenges of developing PBPK oncology populations and key case studies of approved oncology drugs, which have used virtual cancer populations to facilitate drug development. I’ll examine the physiological considerations in cancer patients and review the application of PBPK to explain differences in drug disposition. In a future blog post, I will introduce the use of a permeability-limited tumor model compartment to predict intercellular drug concentrations.
The Benefits of Developing PBPK Oncology Populations
There are numerous advantages for developing PBPK oncology populations as a tool for gaining better understanding and, ultimately, developing novel drugs. First, getting new cancer therapies to patients quickly and safely is a challenge since many of these agents are often toxic with narrow therapeutic windows. This poses a safety concern with dosing healthy volunteers in early stage clinical trials. Secondly, using PBPK cancer populations can aid in developing anti-cancer therapeutics as quickly as possible through accelerated drug approval programs. Thirdly, older patients face an increased risk for developing DDIs from polypharmacy for multiple reasons including multiple comorbidities and altered drug metabolism. Looking to the future, the application of model-informed precision dosing can address all these concerns and optimize efficacy by ensuring that individuals are getting the right dose.
Several large pharma groups who belong to Certara’s Simcyp Consortium have developed in-house PBPK oncology populations.2,3,4 The development of these populations provided an impetus for Certara to develop a virtual cancer population in the Simcyp Simulator. The Simcyp cancer population is a useful platform for investigating tumor disposition, impact on treatment regimens, and for conducting virtual DDI trials to assess the potential for safety concerns.
The Challenges of Developing a PBPK Oncology Population
One major challenge in developing a PBPK oncology population is determining what physiological changes in cancer patients needed to be accounted for. An oncology population represents a diverse population spanning a large array of therapeutic targets as illustrated in the following chart from a recent report by IQVIA Institute on global oncology trends.5
New Active Substance Approval in Oncology by Tumor Type, 2014-2018
Key Parameters for Developing a Simcyp Simulator Population
In general, cancer is a disease of older people. There are physiological differences between young, healthy individuals and older individuals. In developing our oncology population, we incorporated systems data (which represents the population database), the physiological data, and drug data, which is then included in a trial design. All these factors interact and result in predictions using in vitro in vivo extrapolation and PBPK along with predictions of differences of exposure in the population. The North European Caucasian population, which had the majority of data from Caucasian and cancer patients, was used as a baseline to look for physiological differences between healthy and cancer patients. We used the parameters of cancer age distribution, height-age-weight relationships, renal function, prediction of glomerular filtration rate (GFR), and changes in plasma protein concentrations, e.g., alpha-1-acid glycoprotein (AAG), human serum albumin (HSA), and serum creatinine, to predict exposure.
The following Hive plot, where each strand represents an individual, provides a visualization of different parameter differences from a simulation of 1,000 individuals using the Northern European Caucasian (red) and Cancer (blue) populations.
Cancer and Northern European Caucasian Population Simulations
BSA: Body surface area
HSA: Human serum albumin
AAG: α1-acid glycoprotein
SCr: Serum creatinine
GFR: Glomerular filtration rate
Verifying the Model Predictions Using Approved Drugs
Our initial performance verification simulated AUC of six different drugs – Midazolam, Caffeine, Rosiglitazone, S-warfarin, Tolbutamide, and Digoxin. Clearance of Midazolam, S-Warfarin, and Tolbutamide in the cancer population were within 1.4 fold of the observed value in cancer patients. The predicted PK data of three anti-cancer drugs (Docetaxel, Methotrexate, and Paclitaxel) using the Simcyp cancer population was within the 5th and 95th percentiles of the simulated concentration-time profiles of the population. This demonstrated that demographics, physiology, biochemistry, and drug metabolizing enzyme/transporter levels can be used to predict the PK of drugs in cancer patients.
Best Practices in PBPK-led Drug Development
Ibrutinib, an anticancer drug that targets B-cell malignancies by blocking Bruton’s tyrosine kinase (BTK), is an example of a drug development program, which utilized PBPK simulations to mitigate some clinical studies for this CYP3A4 substrate, accelerating drug approval. This best practice approach for using PBPK to determine DDIs was based on building the initial model using in vitro and clinical data, model verification and refinement clinical DDI studies using a strong CYP3A inhibitor and inducer, ketoconazole and rifampin, respectively. Finally, this model was applied to untested clinical DDI scenarios using moderate CYP3A inhibitors and inducers and extrapolating to dose recommendations. Using this approach, the investigators were able to (1) verify the model for ketoconazole and rifampin interactions and fill in the gaps for CYP3A4 inhibitors and inducers and (2) inform the label for dose adjustments needed when the drug was co-administered with different inhibitors.
Other examples where model-informed oncology and drug labeling has been used is in bridging healthy DDIs to cancer patients (Sonidegib), determining the impact of absorption and food effects (Alectinib), and pediatric dose bridging studies (Docetaxel).
In a future blog, I will introduce the tumor distribution model that is available within Simcyp for minimal PBPK modeling. To learn more detail on the development of the PBPK cancer population for predicting exposure, please watch my webinar.
参照文献
- Yoshida K., Budha N., and Jin YK. (2017)。Impact of Physiologically Based Pharmacokinetic Models on Regulatory Reviews and Product Labels: Frequent Utilization in the Field of Oncology. Pharmacol. Ther., 101(5), 597-602.
- Cheeti, S, et al. (2013)。A Physiologically Based Pharmacokinetic (PBPK) Approach to Evaluate Pharmacokinetics in Patients with Cancer. Drug Dispos., 34(3), 141-154.
- Thai, HT, et al. (2015)。Optimizing Pharmacokinetic Bridging Studies in Paediatric Oncology using Physiologically-based Pharmacokinetic Modelling: Application to Docetaxel. J. Clin. Pharmacol., 80(3), 534-547.
- Schwenger, E, et al. (2018)。Harnessing Meta-analysis to Refine an Oncology Patient Population for Physiology-Based Pharmacokinetic Modeling of Drugs. Pharmacol. Ther., 103(2): 271-280.
- Global Oncology Trends 2019: Therapeutics, Clinical Development and Health System Implications. IQVIA Institute for Human Data Science. https://www.iqvia.com/-/media/iqvia/pdfs/institute-reports/global-oncology-trends-2019.pdf?_=1585164822883
