Methods and strategies for assessing uncontrolled drug-drug interactions in population pharmacokinetic analyses: results from the International Society of Pharmacometrics (ISOP) Working Group

Therapeutic Protein Drug Interactions: A White Paper From the International Consortium for Innovation and Quality in Pharmaceutical Development

Typically, therapeutic proteins (TPs) have a low risk for eliciting meaningful drug interactions (DIs). However, there are select instances where TP drug interactions (TP-DIs) of clinical concern can occur. This white paper discusses the various types of TP-DIs involving mechanisms such as changes in disease state, target-mediated drug disposition, neonatal Fc receptor (FcRn), or antidrug antibodies formation. The nature of TP drug interaction being investigated should determine whether the examination is conducted as a standalone TP-DI study in healthy participants, in patients, or assessed via population pharmacokinetic analysis. DIs involving antibody-drug conjugates are discussed briefly, but the primary focus here will be DIs involving cytokine modulation. Cytokine modulation can occur directly by certain TPs, or indirectly due to moderate to severe inflammation, infection, or injury. Disease states that have been shown to result in indirect disease-DIs that are clinically meaningful have been listed (i.e., typically a twofold change in the systemic exposure of a coadministered sensitive cytochrome P450 substrate drug). Type of disease and severity of inflammation should be the primary drivers for risk assessment for disease-DIs. While more clinical inflammatory marker data needs to be collected, the use of two or more clinical inflammatory markers (such as C-reactive protein, albumin, or interleukin 6) may help broadly categorize whether the predicted magnitude of inflammatory disease-DI risk is negligible, weak, or moderate to strong. Based on current knowledge, clinical DI studies are not necessary for all TPs, and should no longer be conducted in certain disease patient populations such as psoriasis, which do not have sufficient systemic inflammation to cause a meaningful indirect disease-DI.

A Survey of Population Pharmacokinetic Reports Submitted to the USFDA: An Analysis of Common Issues in NDA and BLA from 2012 to 2021

BACKGROUND AND OBJECTIVE

Population pharmacokinetic (PopPK) analysis is one of the important components of regulatory submission. The purpose of this study was to survey PopPK analysis reports in new drug applications (NDAs) and biological licensing applications (BLAs) submitted to the US Food and Drug Administration (FDA) and to retrieve the information regarding PopPK-related issues in the US FDA review reports.

METHODS

We surveyed NDAs and BLAs over the period from 2012 to 2022 from Drug@FDA databases and extracted the review reports from the website. We explored the issues identified in these reports and sorted them into four categories: (i) data used for PopPK analysis, (ii) base model development, (iii) final model development, and (iv) model evaluation. The percentage of the issues in each category and the total issues were calculated.

RESULTS

Data from a total of 317 applications were analyzed. Of these, 122 applications had at least one PopPK-related issue, and the count went up to 168 issues. The proportion of issues for each category was ranked as follows: model evaluation (39%, 65 of 168), final model development (34%, 57 of 168), data used for PopPK analysis (14%, 24 of 168), and base model development (13%, 22 of 168). The most common issues were related to the goodness-of-fit plots, covariate selection, and high shrinkage of ETA, accounting for 17.9%, 15.5%, and 11.3% of the total issues, respectively.

CONCLUSION

The findings from this study may help the applicant understand the FDA's thinking on the PopPK analysis.

Leveraging a Previously Published Population Pharmacokinetic Model to Predict Rivaroxaban Exposure in Real-World Patients

Population pharmacokinetic (PK)/pharmacodynamic models are commonly used to inform drug dosing; however, often real-world patients are not well represented in the clinical trial population. We sought to determine how well dosing recommended in the rivaroxaban drug label results in exposure for real-world patients within a reference area under the concentration-time curve (AUC) range. To accomplish this, we assessed the utility of a prior published rivaroxaban population PK model to predict exposure in real-world patients. We used the model to predict rivaroxaban exposure for 230 real-world patients using 3 methods: (1) using patient phenotype information only, (2) using individual post hoc estimates of clearance from the prior model based on single PK samples of rivaroxaban collected at steady state without refitting of the prior model, and (3) using individual post hoc estimates of clearance from the prior model based on PK samples of rivaroxaban collected at steady state after refitting of the prior model. We compared the results across 3 software packages (NONMEM, Phoenix NLME, and Monolix). We found that while the average patient-assigned dosing per the drug label will likely result in the AUC falling within the reference range, AUC for most individual patients will be outside the reference range. When comparing post hoc estimates, the average pairwise percentage differences were all <10% when comparing the software packages, but individual pairwise estimates varied as much as 50%. This study demonstrates the use of a prior published rivaroxaban population PK model to predict exposure in real-world patients.

Comparing the applications of machine learning, PBPK, and population pharmacokinetic models in pharmacokinetic drug-drug interaction prediction

The gold-standard approach for modeling pharmacokinetic mediated drug-drug interactions is the use of physiologically-based pharmacokinetic modeling and population pharmacokinetics. However, these models require extensive amounts of drug-specific data generated from a wide variety of in vitro and in vivo models, which are later refined with clinical data and system-specific parameters. Machine learning has the potential to be utilized for the prediction of drug-drug interactions much earlier in the drug discovery cycle, using inputs derived from, among others, chemical structure. This could lead to refined chemical designs in early drug discovery. Machine-learning models have many advantages, such as the capacity to automate learning (increasing the speed and scalability of predictions), improved generalizability by learning from multicase historical data, and highlighting statistical and potentially clinically significant relationships between input variables. In contrast, the routinely used mechanistic models (physiologically-based pharmacokinetic models and population pharmacokinetics) are currently considered more interpretable, reliable, and require a smaller sample size of data, although insights differ on a case-by-case basis. Therefore, they may be appropriate for later stages of drug-drug interaction assessment when more in vivo and clinical data are available. A combined approach of using mechanistic models to highlight features that can be used for training machine-learning models may also be exploitable in the future to improve the performance of machine learning. In this review, we provide concepts, strategic considerations, and compare machine learning to mechanistic modeling for drug-drug interaction risk assessment across the stages of drug discovery and development.

Creation of an inventory of quality markers used to evaluate pharmacokinetic literature: A systematic review

WHAT IS KNOWN AND OBJECTIVE

Robust critical appraisal tools for clinical pharmacokinetic studies are limited. Before development of such a tool is possible, quality markers (items deemed important for credibility of study results) must be identified. We aim to create an inventory of quality markers intended for the appraisal of clinical pharmacokinetic studies and to categorize identified markers into associated domains of study quality.

METHODS

Medline via ProQuest central (1946-Sep 2020, EMBASE (1974-Sep 2020), Cochrane database of systematic reviews, Google and Google Scholar were searched using the following search categories: pharmacokinetics, reporting guidelines and quality markers. Reference lists of the identified articles were searched manually. Any article (review, study or guideline) reporting quality markers related to the appraisal of pharmacokinetic literature was eligible for inclusion. Articles were further screened and limited to those reported in English on human subjects only. Cell-based and animal-based pharmacokinetic studies were excluded. Extracted data from included articles included identified or perceived markers of quality and baseline article data. Identified quality markers were then categorized according to manuscript reporting domains (abstract, introduction/background, methodology, results, discussion and conclusion).

RESULTS AND DISCUSSION

Of 789 studies identified, 17 articles were included for extraction of quality markers. A total of 35 quality markers were identified across eight categories. The most frequently reported quality markers were related to method (13/35) and result sections (6/35). Quality markers encompassed all aspects of study design and reporting and were both similar and different to established reporting checklists for clinical pharmacokinetic studies.

WHAT IS NEW AND CONCLUSION

The inventory of quality markers is now suitable to undergo further testing for inclusion in a tool designed for the appraisal of clinical pharmacokinetic studies.

Translational aspects of cytochrome P450-mediated drug-drug interactions: A case study with clopidogrel

Multimorbidity, polypharmacotherapy and drug interactions are increasingly common in the ageing population. Many drug-drug interactions (DDIs) are caused by perpetrator drugs inhibiting or inducing cytochrome P450 (CYP) enzymes, resulting in alterations of the plasma concentrations of a victim drug. DDIs can have a major negative health impact, and in the past, unrecognized DDIs have resulted in drug withdrawals from the market. Signals to investigate DDIs may emerge from a variety of sources. Nowadays, standard methods are widely available to identify and characterize the mechanisms of CYP-mediated DDIs in vitro. Clinical pharmacokinetic studies, in turn, provide experimental data on pharmacokinetic outcomes of DDIs. Physiologically based pharmacokinetic (PBPK) modelling utilizing both in vitro and in vivo data is a powerful tool to predict different DDI scenarios. Finally, epidemiological studies can provide estimates on the health outcomes of DDIs. Thus, to fully characterize the mechanisms, clinical effects and implications of CYP-mediated DDIs, translational research approaches are required. This minireview provides an overview of translational approaches to study CYP-mediated DDIs, going beyond regulatory DDI guidelines, and an illustrative case study of how the DDI potential of clopidogrel was unveiled by combining these different methods.

Pediatric Drug-Drug Interaction Evaluation: Drug, Patient Population, and Methodological Considerations

Hospitalized pediatric patients and those with complex or chronic conditions treated on an outpatient basis are commonly prescribed multiple drugs, resulting in increased risk for drug-drug interactions (DDIs). Although dedicated DDI evaluations are routinely performed in healthy adult volunteers during drug development, they are rarely performed in pediatric patients because of ethical, logistical, and methodological challenges. In the absence of pediatric DDI evaluations, adult DDI data are often extrapolated to pediatric patients. However, the magnitude of a DDI in pediatric patients may differ from adults because of age-dependent physiological changes that can impact drug disposition or response and because of other factors related to the drug (eg, dose, formulation) and the patient population (eg, disease state, obesity). Therefore, the DDI magnitude needs to be assessed in children separately from adults, although a lack of clinical DDI data in pediatric populations makes this evaluation challenging. As a result, pediatric DDI assessment relies on the predictive performance of the pharmacometric approaches used, such as population and physiologically based pharmacokinetic modeling. Therefore, careful consideration needs to be given to adequately account for the age-dependent physiological changes in these models to build high confidence for such untested DDI scenarios. This review article summarizes the key considerations related to the drug, patient population, and methodology, and how they can impact DDI evaluation in the pediatric population.

Capturing the Magic Bullet: Pharmacokinetic Principles and Modeling of Antibody-Drug Conjugates

Over the past two decades, antibody-drug conjugates (ADCs) have emerged as a promising class of drugs for cancer therapy and have expanded to nononcology fields such as inflammatory diseases, atherosclerosis, and bacteremia. Eight ADCs are currently approved by FDA for clinical applications, with more novel ADCs under clinical development. Compared with traditional chemotherapy, ADCs combine the target specificity of antibodies with chemotherapeutic capabilities of cytotoxic drugs. The benefits include reduced systemic toxicity and enhanced therapeutic index for patients. However, the heterogeneous structures of ADCs and their dynamic changes following administration create challenges in their development. The understanding of ADC pharmacokinetics (PK) is crucial for the optimization of clinical dosing regimens when translating from animal to human. In addition, it contributes to the optimization of dose selection and clinical monitoring with regard to safety and efficacy. This manuscript reviews the PK characteristics of ADCs and summarizes the diverse approaches for PK modeling that can be used to evaluate an ADC at the preclinical and clinical stages to support their successful development. Despite the numerous available options, fit-for-purpose modeling approaches for the PK and PD of ADCs should be critically planned and well-thought-out to adequately support the development of an ADC.

Pharmacological plasticity-How do you hit a moving target?

Paul Ehrlich's concept of the magic bullet, by which a single drug induces pharmacological effects by interacting with a single receptor has been a strong driving force in pharmacology for a century. It is continually thwarted, though, by the fact that the treated organism is highly dynamic and the target molecule(s) is (are) never static. In this article, we address some of the factors that modify and cause the mobility and plasticity of drug targets and their interactions with ligands and discuss how these can lead to unexpected (lack of) effects of drugs. These factors include genetic, epigenetic, and phenotypic variability, cellular plasticity, chronobiological rhythms, time, age and disease resolution, sex, drug metabolism, and distribution. We emphasize four existing approaches that can be taken, either singly or in combination, to try to minimize effects of pharmacological plasticity. These are firstly, to enhance specificity using target conditions close to those in diseases, secondly, by simultaneously or thirdly, sequentially aiming at multiple targets, and fourthly, in synchronization with concurrent dietary, psychological, training, and biorhythm-synchronizing procedures to optimize drug therapy.

Clinical Studies on Drug-Drug Interactions Involving Metabolism and Transport: Methodology, Pitfalls, and Interpretation

Many drug-drug interactions (DDIs) are based on alterations of the plasma concentrations of a victim drug due to another drug causing inhibition and/or induction of the metabolism or transporter-mediated disposition of the victim drug. In the worst case, such interactions cause more than tenfold increases or decreases in victim drug exposure, with potentially life-threatening consequences. There has been tremendous progress in the predictability and modeling of DDIs. Accordingly, the combination of modeling approaches and clinical studies is the current mainstay in evaluation of the pharmacokinetic DDI risks of drugs. In this paper, we focus on the methodology of clinical studies on DDIs involving drug metabolism or transport. We specifically present considerations related to general DDI study designs, recommended enzyme and transporter index substrates and inhibitors, pharmacogenetic perspectives, index drug cocktails, endogenous substrates, limited sampling strategies, physiologically-based pharmacokinetic modeling, complex DDIs, methodological pitfalls, and interpretation of DDI information.

Population Pharmacokinetic Analysis of Ixazomib, an Oral Proteasome Inhibitor, Including Data from the Phase III TOURMALINE-MM1 Study to Inform Labelling

Ixazomib is an oral proteasome inhibitor, approved in USA, Canada, Australia and Europe in combination with lenalidomide and dexamethasone, for the treatment of patients with multiple myeloma who have received at least one prior therapy. We report a population pharmacokinetic model-based analysis for ixazomib that was pivotal in describing the clinical pharmacokinetics of ixazomib, to inform product labelling. Plasma concentration–time data were collected from 755 patients who received oral or intravenous ixazomib in once- or twice-weekly schedules in ten trials, including the global phase III TOURMALINE-MM1 study. Data were analysed using nonlinear mixed-effects modelling (NONMEM software version 7.2, ICON Development Solutions, Hanover, MD, USA). Ixazomib plasma concentrations from intravenous and oral studies were described by a three-compartment model with linear distribution and elimination kinetics, including first-order linear absorption with a lag time describing the oral dose data. Body surface area on the volume of the second peripheral compartment was the only covariate included in the final model. None of the additional covariates tested including body surface area (1.2–2.7 m²), sex, age (23–91 years), race, mild/moderate renal impairment and mild hepatic impairment were found to impact systemic clearance, suggesting that no dose adjustment is required based on these covariates. The geometric mean terminal disposition phase half-life was 9.5 days, steady-state volume of distribution was 543 L and systemic clearance was 1.86 L/h. The absolute bioavailability of an oral dose was estimated to be 58%.

Increased Exposure of Tacrolimus by Co-administered Mycophenolate Mofetil: Population Pharmacokinetic Analysis in Healthy Volunteers

The objective of the study was to investigate the pharmacokinetic drug-drug interactions between tacrolimus (TAC) and mycophenolate mofetil (MMF) in healthy Korean male volunteers. Seventeen volunteers participated in a three-period, single-dose, and fixed sequence study. They sequentially received MMF, TAC, and the combination. Concentrations of TAC, mycophenolic acid (MPA), and its metabolites MPA 7-O-glucuronide and MPA acyl glucuronide were measured. The variants of CYP3A4, CYP3A5, SLCO1B1, SLCO1B3, ABCC2, UGT1A9, and UGT2B7 were genotyped. Drug interaction was evaluated with a non-compartmental analysis and population pharmacokinetic modelling to quantify the interaction effect. A total of 1,082 concentrations of those analytes were analysed. AUC_0-inf of TAC increased by 22.1% (322.4 ± 174.1 to 393.6 ± 121.7 ng·h/mL; P < 0.05) when co-administered with MMF, whereas the pharmacokinetic parameters of MPA and its metabolites were not changed by TAC. Apparent clearance (CL/F) of TAC was 17.8 L/h [relative standard error (RSE) 11%] or 13.8 L/h (RSE 11%) without or with MMF, respectively. Interaction was explained by the exponential model. The CYP3A5 genotype was the only significant covariate. The population estimate of CL/F of TAC was 1.48-fold (RSE 16%) in CYP3A5 expressers when compared to nonexpressers. CL/F of TAC was decreased when co-administered with MMF in these subjects.

Combining "Bottom-up" and "Top-down" Approaches to Assess the Impact of Food and Gastric pH on Pictilisib (GDC-0941) Pharmacokinetics

Pictilisib, a weakly basic compound, is an orally administered, potent, and selective pan-inhibitor of phosphatidylinositol 3-kinases for oncology indications. To investigate the significance of high-fat food and gastric pH on pictilisib pharmacokinetics (PK) and enable label recommendations, a dedicated clinical study was conducted in healthy volunteers, whereby both top-down (population PK, PopPK) and bottom-up (physiologically based PK, PBPK) approaches were applied to enhance confidence of recommendation and facilitate the clinical development through scenario simulations. The PopPK model identified food (for absorption rate constant (Ka )) and proton pump inhibitors (PPI, for relative bioavailability (Frel ) and Ka ) as significant covariates. Food and PPI also impacted the variability of Frel . The PBPK model accounted for the supersaturation tendency of pictilisib, and gastric emptying physiology successfully predicted the food and PPI effect on pictilisib absorption. Our research highlights the importance of applying both quantitative approaches to address critical drug development questions.

The Utility of a Population Approach in Drug-Drug Interaction Assessments: A Simulation Evaluation

This study aims at evaluating the utility of the population pharmacokinetics approach in therapeutic protein drug-drug-interaction (DDI) assessment. Simulations were conducted for 2 representative victim drugs, methotrexate and trastuzumab, using a parallel-group design with and without the interaction drug. The effect of a perpetrator on the exposure of the victim drug is described as the ratio of clearance/apparent clearance of the victim drug given with or without the perpetrator. The power of DDI assessment was calculated as the percentage of runs with 90% confidence interval of the estimated DDI effect within 80% to 125% for the scenarios of no DDI, benchmarked with the noncompartmental approach with intensive sampling. The impact of the number of subjects, the number of sampling points per subject, sampling time error, and model misspecification on the power of DDI determination were evaluated. Results showed that with equal numbers of subjects in each arm, the population pharmacokinetics approach with sparse sampling may need about the same or a higher number of subjects compared to a noncompartmental approach in order to achieve similar power. Increasing the number of subjects, even if only in the study drug alone arm, can increase the power. Sampling or dosing time error had notable impacts on the power for methotrexate but not for trastuzumab. Model misspecification had no notable impacts on the power for trastuzumab. Overall, the population pharmacokinetics approach with sparse sampling built in phase 2/3 studies allows appropriate DDI assessment with adequate study design and analysis and can be considered as an alternative to dedicated DDI studies.