Clinical Researcher—November 2020 (Volume 34, Issue 9)

REGULATORY COMPLIANCE

Karen Rowland Yeo, PhD

 

Patients often take more than one drug at a time, especially elderly patients and those with complex diseases, such as cancer and neurological disorders. Therefore, it is crucial to determine what the potential risk might be of a new drug candidate interacting with existing marketed medications.

Drug-drug interactions (DDIs) occur when two or more drugs interact with each other. Together the drugs might produce a different pharmacological or clinical response from that seen when they each act independently. DDIs can increase, decrease, or delay drug absorption or metabolism. DDIs can also increase or decrease drug action and cause adverse events. As a result, DDIs can have a significant impact on a drug’s benefit-risk profile.

The U.S. Food and Drug Administration (FDA) requires that an investigational drug’s clinically relevant DDIs are identified during the drug development process as part of the sponsor’s assessment of the drug’s benefits and risks. Those DDIs need to be defined by nonclinical and clinical methods at drug approval, monitored after approval, and communicated in the product labeling.

FDA’s Approach

Underscoring the importance of this practice, the FDA states, “Unanticipated, unrecognized, or mismanaged DDIs are an important cause of morbidity and mortality associated with prescription drug use and have occasionally been the basis for withdrawal of approved drugs from the market. In some instances, understanding how to safely manage a DDI can allow approval of a drug that would otherwise have an unacceptable level of risk.”{1}

Further emphasizing the regulatory significance of DDIs, the FDA published two guidance documents in January 2020—one each focusing on in vitro{1} and clinical{2} cytochrome P450 (CYP) enzyme- and transporter-mediated drug interactions. While these guidance documents do not cover all types of DDIs, CYP 450 enzymes contribute to about 70% of the overall metabolism of marketed drugs.

Studies to investigate CYP enzyme- and transporter-mediated DDIs need to determine:

• Whether the investigational drug alters the pharmacokinetics (PK) of other drugs (a DDI “perpetrator”)
• Whether other drugs alter the PK of the investigational drug (a DDI “victim”)
• Magnitude of changes in PK parameters
• Clinical significance of the observed or expected DDIs
• Appropriate management and prevention strategies for clinically significant DDIs

Other global regulatory agencies, such as the European Medicines Agency and Japan’s Pharmaceuticals and Medical Devices Agency, follow a similar approach to the FDA regarding DDI guidance. Additionally, in September 2020, China’s National Medical Products Administration issued its technical guidelines for drug interaction research.

Optimizing DDI Risk Management

There are several characteristics that make drugs more susceptible to clinically significant DDIs, including a narrow therapeutic index, nonlinear PK, steep dose response curves, and enzyme- or transporter-inhibiting or -inducing properties.{3}

An enormous number of drug combinations could occur in practice, so it is impractical and unethical to test for all possible DDIs in clinical studies. However, physiologically based PK (PBPK) modeling allows drug combinations to be tested using computer-generated, virtual patient populations without involving any real patients. As these models can incorporate genetic, physiological, and epidemiological data, they can also simulate patient populations with different demographics and ethnicities, and can be used to evaluate both the investigational drug’s potential to be a DDI perpetrator or victim.

Regulatory Acceptance

In its aforementioned in vitro DDI guidance, the FDA includes more than 20 citations regarding the use of PBPK modeling to help translate in vitro observations into in vivo predictions of potential clinical DDIs. The agency reports that PBPK models can predict the DDI potential of an investigational drug and/or a metabolite as an enzyme substrate or an enzyme perpetrator.{1}

Further, in its clinical DDI guidance, the FDA states that PBPK models can be used in lieu of some prospective DDI studies. It notes that PBPK models have successfully predicted the impact of weak and moderate inhibitors on the substrates of some CYP isoforms (e.g., CYP2D6, CYP3A) and the impact of weak and moderate inducers on CYP3A substrates. Prior to using PBPK modeling, however, FDA recommends that sponsors verify their models using human PK data and information from DDI studies that used strong index perpetrators.{2}

While these final guidance documents address small molecules, roughly half of the new drugs being developed are either therapeutic proteins, combination small/large molecules, or other types of complex biologics. The FDA is addressing those DDI challenges as well, and issued a draft guidance in August 2020 that cites PBPK as an emerging approach for evaluating DDI potential in therapeutic proteins.{4}

The FDA’s acceptance of PBPK modelling in lieu of clinical DDI studies has steadily evolved. Initially both inducer and inhibitor studies were needed to verify the PBPK model. Later, only one study was required. Now there are instances in which no clinical DDI studies were conducted with the drug as a victim.{5}

Case Studies

Ibrutinib

Approved for the treatment of mantle cell lymphoma, ibrutinib is susceptible to interactions with a strong inhibitor and inducer of CYP3A4 enzymes. PBPK models built using in vitro data were validated using clinical data on the observed effects of both a strong CYP3A4 inhibitor and a strong inducer on ibrutinib exposure. Simulations then predicted the effects of a moderate CYP3A4 inducer and other CYP3A4 inhibitors (strong, moderate, and weak) on ibrutinib exposure. They also investigated the impact of dose staggering and dose adjustment.

This example is cited by the FDA as a best practice. The final drug label featured 24 DDI claims, which were included without the need for clinical trials. It also included a dose optimization strategy for patients with different metabolic profiles.{5}

Cobimetinib

Approved for the treatment of advanced melanoma, cobimetinib is a kinase inhibitor. This case would traditionally have followed a similar PBPK modeling approach to ibrutinib, with model verification based on CYP3A4 strong inhibitor and inducer clinical data. However, with cobimetinib, which is a CYP3A4/UGT2B7 substrate, the sponsor had only itraconazole (a strong CYP3A4 inhibitor) data available and no rifampin (inducer) data.

To create the model, the itraconazole study data was combined with mass balance, human PK, and in vitro data to predict the inducer effects and inform the final drug label. In this instance, the PBPK simulator’s oncology population file was leveraged to predict the effects of CYP3A4 modulators on cobimetinib PK in healthy volunteers and cancer patients using data from only one clinical study. The resulting inducer recommendations on the final label were informed using PBPK simulations alone.{5}

Voxelotor

Approved for the treatment of sickle cell disease (SCD), voxelotor is the first treatment that directly inhibits sickle hemoglobin polymerization, the principal cause of the condition. In this case, PBPK modeling was initially used to determine dose projections for children aged nine months to 12 years. First, a virtual SCD patient population was developed using in vitro and clinical data from healthy volunteers and SCD clinical studies. The resulting model was verified using voxelotor data from adults and adolescents with the disease, and then successfully employed to predict drug exposure in children.

A follow-on request was received to predict voxelotor DDIs with CYP3A4 enzymes, but there were no data from clinical DDI studies using the drug as a victim upon which to draw for building the model. In that instance, the dose prediction model built for healthy and SCD patients was leveraged, together with in vitro data, to create the DDI predictions. Sensitivity analyses performed under multiple scenarios were then used to inform the final label without the need for clinical studies. Furthermore, there was no post-marketing requirement for DDI studies.{5}

Conclusion

PBPK modeling is an effective, accepted method of informing and replacing DDI studies, thus saving time and money. It is a proven asset, helping to manage potential DDI risk for patients who need to take multiple medications concurrently. We anticipate that the use of PBPK modeling for assessing DDI potential will soon be expanded into other areas, such as transporters, and it will also be employed to answer many other drug development questions.

References

1. U.S. Food and Drug Administration Guidance. 2020. In Vitro Drug Interaction Studies—Cytochrome P450 Enzyme- and Transporter-Mediated Drug Interactions.
2. U.S. Food and Drug Administration Guidance. 2020. Clinical Drug Interaction Studies—Cytochrome P450 Enzyme- and Transporter-Mediated Drug Interactions.
3. Hermann R, et al. 2018. Core Entrustable Professional Activities in Clinical Pharmacology: Pearls for Clinical Practice Drug-Drug and Food-Drug Interactions. The Journal of Clinical Pharmacology 58(6)704–16.
4. U.S. Food and Drug Administration Draft Guidance. 2020. Drug-Drug Interaction Assessment for Therapeutic Proteins Guidance for Industry. https://www.fda.gov/media/140909/download
5. Yeo K. 2020. Simcyp PBPK for Drug-Drug Interactions (DDIs): A Regulatory Imperative. https://www-cdn-assets.certara.com/app/uploads/2020/09/WP_DDI-Optimized.pdf

Karen Rowland Yeo, PhD, is Senior Vice President of Client and Regulatory Strategy at Certara UK Limited’s Simcyp Division. Previously, she was Head of PBPK Consultancy Services at Simcyp, leading a team of scientists applying PBPK modeling in the drug development process.