Tarek A. Leil

Quantitative Systems Pharmacology can reduce attrition and improve productivity in pharmaceutical research and development

The empirical hypothesis generation and testing approach to pharmaceutical research and development (R&D), and biomedical research has proven very effective over the last half-century; resulting in tremendous increases productivity and the rates of approval for new drug applications at the Food and Drug Administration (FDA). However, as discovery of new therapeutic approaches for diseases with unmet medical need becomes more challenging, the productivity and efficiency of the traditional approach to drug discovery and development is diminishing. Innovative approaches are needed, such as those offered by Quantitative Systems Pharmacology (QSP) modeling and simulation. This “systems” approach to modeling and simulation can be used to guide the hypothesis generation and testing process in pharmaceutical R&D, in a manner similar to its adoption in other industries in the past. Embedding QSP into the existing processes within pharmaceutical discovery and development will be required in order to realize the full beneficial impact of this innovative approach.

Evaluation of 4β-Hydroxycholesterol as a Clinical Biomarker of CYP3A4 Drug Interactions Using a Bayesian Mechanism–Based Pharmacometric Model

A Bayesian mechanism–based pharmacokinetic/pharmacodynamic model of cytochrome P450 3A4 (CYP3A4) activity was developed based on a clinical study of the effects of ketoconazole and rifampin on midazolam exposure and plasma 4β‐hydroxycholesterol (4βHC) concentrations. Simulations from the model demonstrated that the dynamic range of 4βHC as a biomarker of CYP3A4 induction or inhibition was narrower than that of midazolam; an inhibitor that increases midazolam area under the curve by 20‐fold may only result in a 20% decrease in 4βHC after 14 days of dosing. Likewise, an inducer that elevates CYP3A4 activity by 1.2‐fold would reduce the area under the curve of midazolam by 50% but would only increase 4βHC levels by 20% after 14 days of dosing. Elevation in 4βHC could be reliably detected with a twofold induction in CYP3A4 activity with study sample sizes (N ~ 6–20) typically used in early clinical development. Only a strong CYP3A4 inhibitor (e.g., ketoconazole) could be detected with similar sample sizes.

Use of systems pharmacology modeling to elucidate the operating characteristics of SGLT1 and SGLT2 in renal glucose reabsorption in humans.

In the kidney, glucose in glomerular filtrate is reabsorbed primarily by sodium-glucose cotransporters 1 (SGLT1) and 2 (SGLT2) along the proximal tubules. SGLT2 has been characterized as a high capacity, low affinity pathway responsible for reabsorption of the majority of filtered glucose in the early part of proximal tubules, and SGLT1 reabsorbs the residual glucose in the distal part. Inhibition of SGLT2 is a viable mechanism for removing glucose from the body and improving glycemic control in patients with diabetes. Despite demonstrating high levels (in excess of 80%) of inhibition of glucose transport by SGLT2 in vitro, potent SGLT2 inhibitors, e.g., dapagliflozin and canagliflozin, inhibit renal glucose reabsorption by only 30–50% in clinical studies. Hypotheses for this apparent paradox are mostly focused on the compensatory effect of SGLT1. The paradox has been explained and the role of SGLT1 demonstrated in the mouse, but direct data in humans are lacking. To further explore the roles of SGLT1/2 in renal glucose reabsorption in humans, we developed a systems pharmacology model with emphasis on SGLT1/2 mediated glucose reabsorption and the effects of SGLT2 inhibition. The model was calibrated using robust clinical data in the absence or presence of dapagliflozin (DeFronzo et al., 2013), and evaluated against clinical data from the literature (Mogensen, 1971; Wolf et al., 2009; Polidori et al., 2013). The model adequately described all four data sets. Simulations using the model clarified the operating characteristics of SGLT1/2 in humans in the healthy and diabetic state with or without SGLT2 inhibition. The modeling and simulations support our proposition that the apparent moderate, 30–50% inhibition of renal glucose reabsorption observed with potent SGLT2 inhibitors is a combined result of two physiological determinants: SGLT1 compensation and residual SGLT2 activity. This model will enable in silico inferences and predictions related to SGLT1/2 modulation.

Model‐Based Exposure–Response Analysis of Apixaban to Quantify Bleeding Risk in Special Populations of Subjects Undergoing Orthopedic Surgery

Population pharmacokinetic (PK) and exposure–response analyses of apixaban were performed using data from phase I–III studies to predict bleeding risks for patients receiving apixaban 2.5 mg b.i.d. after total knee or hip replacement (TKR, THR) surgery (N = 5,510). Renal function, age, gender, and body weight impacted apixaban exposure. Bleeding risk increased as a function of exposure. Predicted bleeding frequencies for TKR and THR populations at risk for high apixaban exposure (female, age > 75 years, calculated creatinine clearance (cCrCL) < 30 ml/min, body weight < 50 kg) (6.85 and 10.3%, respectively) were comparable to the reference population (male/female, age 65−75 years, cCrCL ≥ 80 ml/min, body weight 65−85 kg) (6.18 and 9.32%, respectively). A 100% increase in apixaban exposure is expected to raise bleeding frequencies to 7.25% (TKR) and 10.9% (THR), whereas a 200% increase would raise them to 8.49 and 12.7%. Coexistence of combined patient risk factors or doubling of exposure is not likely to result in a substantial, clinically relevant increase in bleeding risk with 2.5 mg b.i.d. apixaban.

Virtual Systems Pharmacology (ViSP) software for simulation from mechanistic systems-level models

Multiple software programs are available for designing and running large scale system-level pharmacology models used in the drug development process. Depending on the problem, scientists may be forced to use several modeling tools that could increase model development time, IT costs and so on. Therefore, it is desirable to have a single platform that allows setting up and running large-scale simulations for the models that have been developed with different modeling tools. We developed a workflow and a software platform in which a model file is compiled into a self-contained executable that is no longer dependent on the software that was used to create the model. At the same time the full model specifics is preserved by presenting all model parameters as input parameters for the executable. This platform was implemented as a model agnostic, therapeutic area agnostic and web-based application with a database back-end that can be used to configure, manage and execute large-scale simulations for multiple models by multiple users. The user interface is designed to be easily configurable to reflect the specifics of the model and the users particular needs and the back-end database has been implemented to store and manage all aspects of the systems, such as Models, Virtual Patients, User Interface Settings, and Results. The platform can be adapted and deployed on an existing cluster or cloud computing environment. Its use was demonstrated with a metabolic disease systems pharmacology model that simulates the effects of two antidiabetic drugs, metformin and fasiglifam, in type 2 diabetes mellitus patients.

Editorial: The emerging discipline of quantitative systems pharmacology

Quantitative Systems Pharmacology (QSP) has emerged recently as an approach that integrates knowledge coming from multiple disciplines including drug pharmacology, systems biology, physiology, mathematics and biochemistry. QSP was formally defined as a discipline and endorsed in the NIH White Paper (Sorger et al., 2011) in 2011. It has emerged at a time when the pharmaceutical industry is facing growing challenges in efficiency and productivity in R&D. QSP has the potential to help overcome some of these challenges. QSP models allow researchers to evaluate multiple hypotheses in-silico that would otherwise need to be evaluated experimentally. There is an expectation that the use of QSP will reduce the cost of R&D and the risks associated with uncertainties and gaps in our knowledge while bringing new therapies to patients. QSP models are typically perceived as a research tool for hypothesis generation in drug discovery and exploratory clinical development; however, recently the US FDA used a QSP model to evaluate a proposed drug regimen for a new biologic therapy (Peterson and Riggs, 2015). In their communication with NPS Pharma, the FDAs Clinical Pharmacology division used a published QSP model of the calcium homeostasis system (Peterson and Riggs, 2010) to recommend an alternate dosing regimen for NATPARA, an injectable parathyroid hormone replacement drug used to control low blood calcium in patients with hypoparathyroidism. The use of a QSP model by the FDA to recommend an alternate dosing regimen to a sponsor highlights one of the important future applications of QSP models in regulatory interactions, and also represents an important milestone for the field. It is the first public instance of a QSP model being used by a regulatory agency to make a clinical recommendation to a sponsor. In the future, it is anticipated that it will be sponsors that leverage the utility of QSP models to support their own clinical decision making with regulatory agencies. In the present research topic entitled, “The Emerging Discipline of Quantitative Systems Pharmacology,” we provide an introduction to the developing field of QSP with a series of articles that describe models in different disease areas; showing how these models can be used to evaluate important research questions in pharmaceutical RD for example to translate knowledge between experimental systems (e.g., animal to human), and to predict the effects of multiple therapeutic interventions in combination; a task that would be inefficient using only clinical experimentation. The other articles in the research topic go on to demonstrate applications of QSP models to influence decision making in biomedical research. One of the important applications of QSP modeling in pharmaceutical R&D is optimization of clinical dose and schedule. Oncology is one of the disease areas where the narrow therapeutic window of most therapeutics demands fine tuning of dose and schedule. Utilization of high doses of anti-angiogenesis therapy can result in rapid suppression of angiogenesis and hypoxia leading to tumor shrinkage. Paradoxically, this can subsequently lead to reduced drug delivery to the tumor and resumption in tumor angiogenesis, followed by progression of tumor growth. The paper by Sharan and Woo (2015) discusses how to delay or prevent this from happening by optimizing the dose and regimen of the anti-angiogenesis therapies with a QSP model of angiogenesis. Another important application of QSP models is to provide mechanistic explanations for clinical data that are often counterintuitive to the perceived mechanism of action (MoA) of a drug. Despite the fact that two SGLT2 inhibitors are already approved for use in patients, questions remain regarding their MoA and why efficacy is lower than expected based on their high potency and selectivity for SGLT2. The papers by Demin et al. (2014) and Lu et al. (2014) explored this issue using mechanistic models of renal tubular filtration and transport, incorporating the PK and MoA of SGLT2 inhibitors. Lu et al. (2014) proposed two possible explanations for the lower than expected efficacy; the residual activity of SGLT2 following inhibition in the renal tubules, and the compensatory effect of SGLT1. Demin et al. (2014) supported this hypothesis, but also offered an alternative in which the sites of action of SGLT2 inhibitors are located not in the lumen of the kidneys proximal tubules where the concentration of SGLT2 inhibitor is high, but perhaps in the proximal tubule where the concentration of inhibitor is lower. Complex dose-response dependencies are often encountered in many disease areas, for instance, in treatments of schizophrenia as investigated by Spiros et al (Spiros et al., 2014). With the use of a sophisticated QSP model of cognitive impairment in schizophrenia, the authors predicted an inverse U-shape dose-response with glycine that is a consequence of the shifting balance between excitation and inhibition in the cortical network. The application of mechanistic models for prediction of target dependent or independent toxicity in secondary tissues has been the focus of QSP for many years, as this is the most common reason for termination of the development of otherwise efficacious therapies. In order to predict toxicity using a mechanistic model, it is useful to incorporate a physiologically based PK (PBPK) model to predict drug concentrations in the target organ. Woodhead et al. (2014) described the use of a PBPK/toxicity model of drug induced liver injury (DILI) to evaluate the impact of bile salt export pump (BSEP) inhibition on hepatotoxicity in rats and humans. The DILI model was used to predict the responses to BSEP inhibitors with and without clinical hepatotoxicity. In accordance with the observed clinical data, the model predicted that bosentan, but not telmisartan, will cause mild hepatocellular ATP decline and serum ALT elevation. Similar to the research by Woodhead et al. (2014), Chetty et al. also relied on PBPK to predict drug concentrations in tissue, linking these concentrations to target-mediated pharmacodynamic (PD) effects (Chetty et al., 2014). They did so using the Simcyp PBPK simulator, a tool that has a built-in PBPK model and allows users to input drug specific parameters that have been measured in-vitro to predict in-vivo plasma and tissue PK. Simcyp has become an important tool for pharmaceutical R&D and for communicating with regulatory agencies regarding the PK of investigational drugs and their potential PK-related drug-interactions. Chetty et al. described the use of Simcyp to predict the tissue concentrations of four different drugs, metoprolol, nifedipine, triazolam, and zolpidem (Chetty et al., 2014). They demonstrated how polymorphisms in drug metabolizing enzymes would have an effect on the concentration of drugs in the target tissue and the subsequent impact on pharmacodynamics. One of the technical issues that potentially limit more widespread use of QSP in biomedical research is the lack of an accepted standard modeling tool to facilitate sharing of models between researchers. The tool should permit evaluation of experimental scenarios of interest in a flexible computational environment for conducting efficient high throughput simulation. A potential solution was implemented in the web-based virtual systems pharmacology (ViSP) platform described in the article by Ermakov et al. (2014). The salient feature of ViSP is the use of a model in the form of an executable file. Matched with a full set of model parameters this executable becomes independent of the model structure and the software that were used to develop the model while preserving flexibility in the input parameters. These characteristics could be useful in the future when the utilization and sharing of QSP models becomes more widespread. In conclusion, we would like to emphasize the emerging potential that QSP holds for biomedical research and in particular to improve decision making in pharmaceutical RD examples like those included in this research topic.

Comparison of the pharmacokinetics and pharmacodynamics of dapagliflozin in patients with type 1 versus type 2 diabetes mellitus.

To compare the pharmacokinetics and pharmacodynamics of dapagliflozin in patients with type 1 diabetes mellitus (T1DM) versus type 2 diabetes mellitus (T2DM) in order to explore the potential of dapagliflozin as add‐on therapy to insulin in patients with T1DM.

Quantitative Extrapolation: An Approach to Validation of Adult Drug Efficacy in Pediatric Subjects

Confirmation of efficacy in pediatric drug development has traditionally required large, fully powered efficacy studies that have proven to have major feasibility and ethical challenges. Extrapolation of efficacy in the framework provided by the US Food and Drug Administration and European Medicines Agency is an appropriate solution when there is similarity of disease. When there is uncertainty regarding the degree of disease similarity, partial extrapolation may be utilized. The authors propose a more quantitative approach to partial extrapolation (ie, quantitative extrapolation), involving (1) integration of adult pharmacokinetic (PK), pharmacodynamic (PD), and clinical outcome data using pharmacometric models, (2) extrapolation using the adult pharmacometric model to predict PD and efficacy outcomes in pediatric subjects, and (3) validation of pediatric predictions with a streamlined plan of pediatric trials (ie, a quantitative extrapolation plan). A case study is presented for quantitative extrapolation using dipeptidyl peptidase 4 (DPP-4) inhibitors. In this example, the authors demonstrate how adult PK, PD, and HbA1c data can be integrated using a pharmacometric model for DPP-4 inhibitors with pediatric dose selection and efficacy validated with relatively few pediatric subjects.

Population Pharmacokinetics of Apixaban in Subjects With Nonvalvular Atrial Fibrillation

This analysis describes the population pharmacokinetics (PPK) of apixaban in nonvalvular atrial fibrillation (NVAF) subjects, and quantifies the impact of intrinsic and extrinsic factors on exposure. The PPK model was developed using data from phase I–III studies. Apixaban exposure was characterized by a two‐compartment PPK model with first‐order absorption and elimination. Predictive covariates on apparent clearance included age, sex, Asian race, renal function, and concomitant strong/moderate cytochrome P450 (CYP)3A4/P‐glycoprotein (P‐gp) inhibitors. Individual covariate effects generally resulted in < 25% change in apixaban exposure vs. the reference NVAF subject (non‐Asian, male, aged 65 years, weighing 70 kg without concomitant CYP3A4/P‐gp inhibitors), except for severe renal impairment, which resulted in 55% higher exposure than the reference subject. The dose‐reduction algorithm resulted in a ~27% lower median exposure, with a large overlap between the 2.5‐mg and 5‐mg groups. The impact of Asian race on apixaban exposure was < 15% and not considered clinically significant.