Accelerating the Progress of Kinase Inhibitors in Oncology

Applied Clinical TrialsApplied Clinical Trials-05-01-2024
Volume 33
Issue 5

The potential of PBPK modeling in answering key questions around these drugs.

Hannah Jones, PhD, Senior Vice President, PBPK Consultancy Services, Certara

Hannah Jones, PhD, Senior Vice President, PBPK Consultancy Services, Certara

Masoud Jamei, PhD, Senior Vice President, Research and Development, Certara

Masoud Jamei, PhD, Senior Vice President, Research and Development, Certara

More than 50 kinase inhibitors (KIs) have been approved by the FDA to treat 20 types of cancer. KIs have been shown to slow cancer proliferation, metastasis, and angiogenesis while avoiding many of the cytotoxic effects of chemotherapy. As the approved KIs target only 20 of the 538 kinases encoded by the human genome, they also hold enormous potential for future advances.

Physiologically-based pharmacokinetic (PBPK) modeling is used extensively in the development of oncology drugs. With KIs, PBPK modeling is primarily used for predicting drug-drug interactions (DDIs) during clinical development, especially if a sponsor is considering combining its drug candidate with one that is already on the market, and for informing dosing for special populations, such as pediatric patients or individuals with liver or kidney impairment, for the new drug label.

However, it is also employed for first-in-human PK predictions early in the discovery process, and for informing inclusion/exclusion criteria for the clinical trial protocol, which could expand the pool of potential participants and help to accelerate the drug development process.

Combining drugs

If a sponsor is considering combining its KI drug candidate with another small molecule, the first step in creating a PBPK model is to study each compound’s drug metabolism and pharmacokinetic properties, their physicochemical parameters, and the in vitro data on metabolism, inhibition and absorption. Solubility and permeability data are vital.

The model is often created using one clinical dataset. Then, the model is verified using separate clinical datasets. If the sponsor has already performed studies to determine how the drug candidate interacts with other treatments, these will also be used to verify the model performance. It is important to know how the compound is metabolized and cleared and whether it has any perpetrator interaction properties. That information helps to determine how sensitive it will be to inhibitors or inducers of those enzymes and how it might affect the metabolism of other drugs. A model can also be built based on information from the scientific literature or the FDA clinical pharmacology review. These models can be fit for purpose and designed to answer specific questions. They can be combined to determine how other drugs affect their drug candidate.

Managing comedications

As most patients undergoing cancer therapy receive multiple drugs, and may be facing an infection, it is especially important to investigate possible DDIs to set doses for patients. Once the drug is approved, these results are also used to inform the drug label because clinicians need to know, if they are planning to administer the KI with another drug, whether they should increase, decrease, or not change the dose.

As cancer drugs cause nausea, they may be given with acid-reducing agents. If a patient is taking one of those medications, PBPK modeling can help to determine whether they need to adjust the dose of the drug candidate or avoid it.

KIs, in some cases, have a low bioavailability in the gastrointestinal tract—meaning if the drug is taken orally, only a small amount of it will get into the circulation. If it is taken with food, that could increase or decrease its bioavailability significantly, further impacting the amount of drug that reaches the site of action. As a result, this could lead to reduced efficacy or increased toxicity.

Creating virtual patients

In early clinical trials, drug candidates are usually given to a group of healthy volunteers. But with an oncology treatment, due to safety concerns, it is typically given to the patient population first. However, PBPK modeling allows the drug’s characteristics to be studied in a computer-generated, virtual cancer patient population. This virtual patient population is designed to reflect specific elements of the disease. For example, some of the enzyme, transporter, and plasma protein levels are different in patients with cancer. Other physiological parameters may change, too, including blood flow.

Many oncology drugs preferentially bind to a plasma protein, usually alpha-1-acid glycoprotein (AGP) or albumin. AGP levels are also quite different in cancer patients. So, if the drug preferentially binds to AGP, there is often higher exposure in cancer patients. Once the drug model is ready, the compound can be evaluated in virtual populations of different ages, sexes, body weight, ethnicity, and with preexisting conditions.

Extrinsic factors that can affect a drug candidate’s metabolism and PK, such as smoking, drinking alcohol, and diet, can also be modeled.

Determining pediatric dosing

In addition to investigating potential DDIs and informing adult drug doses, PBPK modeling can be used to predict what KI dose a sponsor will need to use in pediatric patients to support their pediatric plan. Once the exposure in pediatric patients has been confirmed using the model, possible DDIs may also be explored because children can be on multiple drugs, too. As conducting DDI studies in children is not possible, modeling is really the only option to help inform those decisions for clinicians.

Case study: Ibrutinib

Ibrutinib (Imbruvica) was being investigated as a treatment for mantle cell lymphoma. As most KIs are metabolized by the enzyme cytochrome P450 (CYP) 3A4, ibrutinib’s sponsor conducted a DDI study with ketoconazole, which is a strong CYP3A4 inhibitor, and saw a more than 20-fold increase in exposure. It also conducted a strong CYP3A4 inducer study with rifampicin.

Then, PBPK models were built using both in vitro and in vivo data and verified using clinical data from the strong inhibitor and strong inducer studies. Modeling analysis determined that about 90% of the ibrutinib was metabolized by CYP3A4. Using data from the models, it was possible to infer what impact a moderate or a weak inhibitor or a moderate inducer would have on ibrutinib exposure. The models were also used to study the effect of dose staggering and dose adjustment.

When ibrutinib received its first approval from the FDA in 2013, Simcyp PBPK Simulator predictions informed 24 claims on its label, including DDI scenarios and a dosing optimization strategy for comedications, all without the need for clinical studies. Regulators now cite this use of PBPK modeling as a best practice.

Case study: Adagrasib

Adagrasib (Krazati) was being developed as a therapy for previously treated KRAS G12C-mutated non-small cell lung cancer. But adagrasib is a complex molecule; it is a substrate for CYP3A4 and a very strong inactivator of CYP3A4. Adagrasib is initially metabolized by CYP3A4, but as the inactivation mechanism kicks in, clearance is reduced because the drug inactivates the enzyme that is metabolizing it, and the exposure increases significantly.

A PBPK model was developed to predict the impact of CYP3A4 inhibitors and inducers on adagrasib exposure following the recommended dosing regimen. Clinical studies could not be conducted in healthy volunteers because of emetic issues.

As adagrasib’s metabolism changes over time, it was not sufficient to study its interaction with a strong CYP3A4 inhibitor (itraconazole); after only a single dose, the PBPK model needed to extrapolate to multiple dosing. The model was also used to study CYP3A4 induction with adagrasib and predict potential DDIs with CYP2D6, CYP2C9, and P-gp (a multi-drug transporter).

As adagrasib’s PK can change over time, with dose, and with population, it would be impossible to run all the possible scenarios in real life, and some would not be permitted for ethical reasons. But what would happen if the patient not only had cancer but also poor hepatic function because they were elderly? The Simcyp PBPK Simulator allows potential DDIs to be explored for that patient by combining two models.

FDA accepted Simcyp PBPK Simulator modeling predictions in lieu of clinical studies for 10 label claims for adagrasib, including transporter DDIs. The modeling also resulted in a label expansion to include multi-dosing recommendations for patients with mild, moderate, and severe hepatic impairment.

Model momentum

During the past 10 years, Simcyp PBPK Simulator models have been used to inform, reduce, and/or replace clinical studies for more than 40 KI drugs. Their predictions were accepted by regulators in lieu of clinical studies for more than 100 label claims for KI drugs.

Hannah Jones, PhD, is Senior Vice President, PBPK Consultancy Services, Certara; Masoud Jamei, PhD, is Senior Vice President, Research and Development, Certara

Related Content
© 2024 MJH Life Sciences

All rights reserved.