Longitudinal metabolite profiling of Streptococcus pneumoniae-associated community-acquired pneumonia.

INTRODUCTION: Longitudinal biomarkers in patients with community-acquired pneumonia (CAP) may help in monitoring of disease progression and treatment response. The metabolic host response could be a potential source of such biomarkers since it closely associates with the current health status of the patient. OBJECTIVES: In this study we performed longitudinal metabolite profiling in patients with CAP for a comprehensive range of metabolites to identify potential host response biomarkers. METHODS: Previously collected serum samples from CAP patients with confirmed Streptococcus pneumoniae infection (n = 25) were used. Samples were collected at multiple time points, up to 30 days after admission. A wide range of metabolites was measured, including amines, acylcarnitines, organic acids, and lipids. The associations between metabolites and C-reactive protein (CRP), procalcitonin, CURB disease severity score at admission, and total length of stay were evaluated. RESULTS: Distinct longitudinal profiles of metabolite profiles were identified, including cholesteryl esters, diacyl-phosphatidylethanolamine, diacylglycerols, lysophosphatidylcholines, sphingomyelin, and triglycerides. Positive correlations were found between CRP and phosphatidylcholine (34:1) (cor = 0.63) and negative correlations were found for CRP and nine lysophosphocholines (cor = - 0.57 to - 0.74). The CURB disease severity score was negatively associated with six metabolites, including acylcarnitines (tau = - 0.64 to - 0.58). Negative correlations were found between the length of stay and six triglycerides (TGs), especially TGs (60:3) and (58:2) (cor = - 0.63 and - 0.61). CONCLUSION: The identified metabolites may provide insight into biological mechanisms underlying disease severity and may be of interest for exploration as potential treatment response monitoring biomarker.

Model-based dose optimization framework for bedaquiline, pretomanid and linezolid for the treatment of drug-resistant tuberculosis.

AIMS: Bedaquiline, pretomanid and linezolid (BPaL) combination treatment against Mycobacterium tuberculosis is promising, yet safety and adherence concerns exist that motivate exploration of alternative dosing regimens. We developed a mechanistic modelling framework to compare the efficacy of the current and alternative BPaL treatment strategies. METHODS: Pharmacodynamic models for each drug in the BPaL combination treatment were developed using in vitro time-kill data. These models were combined with pharmacokinetic models, incorporating body weight, lesion volume, site-of-action distribution, bacterial susceptibility and pharmacodynamic interactions to assemble the framework. The model was qualified by comparing the simulations against the observed clinical data. Simulations were performed evaluating bedaquiline and linezolid approved (bedaquiline 400 mg once daily [QD] for 14 days followed by 200 mg three times a week, linezolid 1200 mg QD) and alternative dosing regimens (bedaquiline 200 mg QD, linezolid 600 mg QD). RESULTS: The framework adequately described the observed antibacterial activity data in patients following monotherapy for each drug and approved BPaL dosing. The simulations suggested a minor difference in median time to colony forming unit (CFU)-clearance state with the bedaquiline alternative compared to the approved dosing and the linezolid alternative compared to the approved dosing. Median time to non-replicating-clearance state was predicted to be 15 days from the CFU-clearance state. CONCLUSIONS: The model-based simulations suggested that comparable efficacy can be achieved using alternative bedaquiline and linezolid dosing, which may improve safety and adherence in drug-resistant tuberculosis patients. The framework can be utilized to evaluate treatment optimization approaches, including dosing regimen and duration of treatment predictions to eradicate both replicating- and non-replicating bacteria from lung and lesions.

Generation of realistic virtual adult populations using a model-based copula approach.

Incorporating realistic sets of patient-associated covariates, i.e., virtual populations, in pharmacometric simulation workflows is essential to obtain realistic model predictions. Current covariate simulation strategies often omit or simplify dependency structures between covariates. Copula models are multivariate distribution functions suitable to capture dependency structures between covariates with improved performance compared to standard approaches. We aimed to develop and evaluate a copula model for generation of adult virtual populations for 12 patient-associated covariates commonly used in pharmacometric simulations, using the publicly available NHANES database, including sex, race-ethnicity, body weight, albumin, and several biochemical variables related to organ function. A multivariate (vine) copula was constructed from bivariate relationships in a stepwise fashion. Covariate distributions were well captured for the overall and subgroup populations. Based on the developed copula model, a web application was developed. The developed copula model and associated web application can be used to generate realistic adult virtual populations, ultimately to support model-based clinical trial design or dose optimization strategies.

Protein structure-based gene expression signatures.

Gene expression signatures (GES) connect phenotypes to differential messenger RNA (mRNA) expression of genes, providing a powerful approach to define cellular identity, function, and the effects of perturbations. The use of GES has suffered from vague assessment criteria and limited reproducibility. Because the structure of proteins defines the functional capability of genes, we hypothesized that enrichment of structural features could be a generalizable representation of gene sets. We derive structural gene expression signatures (sGES) using features from multiple levels of protein structure (e.g., domain and fold) encoded by the mRNAs in GES. Comprehensive analyses of data from the Genotype-Tissue Expression Project (GTEx), the all RNA-seq and ChIP-seq sample and signature search (ARCHS4) database, and mRNA expression of drug effects on cardiomyocytes show that sGES are useful for characterizing biological phenomena. sGES enable phenotypic characterization across experimental platforms, facilitates interoperability of expression datasets, and describe drug action on cells.

Semi-mechanistic modeling of resistance development to ß-lactam and ß-lactamase-inhibitor combinations.

The use of ß-lactam (BL) and ß-lactamase inhibitor (BLI) combinations, such as piperacillin-tazobactam (PIP-TAZ) is an effective strategy to combat infections by extended-spectrum ß-lactamase-producing bacteria. However, in Gram-negative bacteria, resistance (both mutational and adaptive) to BL-BLI combination can still develop through multiple mechanisms. These mechanisms may include increased ß-lactamase activity, reduced drug influx, and increased drug efflux. Understanding the relative contribution of these mechanisms during resistance development helps identify the most impactful mechanism to target in designing a treatment to counter BL-BLI resistance. This study used semi-mechanistic mathematical modeling in combination with antibiotic sensitivity assays to assess the potential impact of different resistance mechanisms during the development of PIP-TAZ resistance in a Klebsiella pneumoniae isolate expressing CTX-M-15 and SHV-1 ß-lactamases. The mathematical models were used to evaluate the potential impact of several cellular changes as a sole mediator of PIP-TAZ resistance. Our semi-mechanistic model identified 2 out of the 13 inspected mechanisms as key resistance mechanisms that may independently support the observed magnitude of PIP-TAZ resistance, namely porin loss and efflux pump up-regulation. Simulation using the resulting models also suggested the possible adjustment of PIP-TAZ dose outside its commonly used 8:1 dosing ratio. The current study demonstrated how theory-based mechanistic models informed by experimental data can be used to support hypothesis generation regarding potential resistance mechanisms, which may guide subsequent experimental studies.

Systems Pharmacology: Defining the Interactions of Drug Combinations.

The majority of diseases are associated with alterations in multiple molecular pathways and complex interactions at the cellular and organ levels. Single-target monotherapies therefore have intrinsic limitations with respect to their maximum therapeutic benefits. The potential of combination drug therapies has received interest for the treatment of many diseases and is well established in some areas, such as oncology. Combination drug treatments may allow us to identify synergistic drug effects, reduce adverse drug reactions, and address variability in disease characteristics between patients. Identification of combination therapies remains challenging. We discuss current state-of-the-art systems pharmacology approaches to enable rational identification of combination therapies. These approaches, which include characterization of mechanisms of disease and drug action at a systems level, can enable understanding of drug interactions at the molecular, cellular, physiological, and organismal levels. Such multiscale understanding can enable precision medicine by promoting the rational development of combination therapy at the level of individual patients for many diseases.

Quantitative Systems Pharmacology Modeling Framework of Autophagy in Tuberculosis: Application to Adjunctive Metformin Host-Directed Therapy.

Quantitative systems pharmacology (QSP) modeling of the host immune response against Mycobacterium tuberculosis can inform the rational design of host-directed therapies (HDTs). We aimed to develop a QSP framework to evaluate the effects of metformin-associated autophagy induction in combination with antibiotics. A QSP framework for autophagy was developed by extending a model for host immune response to include adenosine monophosphate-activated protein kinase (AMPK)-mTOR-autophagy signaling. This model was combined with pharmacokinetic-pharmacodynamic models for metformin and antibiotics against M. tuberculosis. We compared the model predictions to mice infection experiments and derived predictions for the pathogen- and host-associated dynamics in humans treated with metformin in combination with antibiotics. The model adequately captured the observed bacterial load dynamics in mice M. tuberculosis infection models treated with metformin. Simulations for adjunctive metformin therapy in newly diagnosed patients suggested a limited yet dose-dependent effect of metformin on reduction of the intracellular bacterial load when the overall bacterial load is low, late during antibiotic treatment. We present the first QSP framework for HDTs against M. tuberculosis, linking cellular-level autophagy effects to disease progression and adjunctive HDT treatment response. This framework may be extended to guide the design of HDTs against M. tuberculosis.

Characterizing the kinetics of presepsin and associated inflammatory biomarkers in human endotoxemia.

In this study, we describe the kinetics of a new potential inflammatory biomarker, presepsin, together with a panel of well-established biomarkers in a human endotoxemia study. We evaluated biomarker correlations and identified combinations that could hold valuable insights regarding the state of infection.

Host-response biomarkers for the diagnosis of bacterial respiratory tract infections.

Appropriate antibiotic treatment for respiratory tract infections (RTIs) necessitates rapid and accurate diagnosis of microbial etiology, which remains challenging despite recent innovations. Several host response-based biomarkers due to infection have been suggested to allow discrimination of bacterial and non-bacterial microbial RTI etiology. This review provides an overview of clinical studies that investigated the diagnostic performance of host-response proteomic biomarkers to identify RTI microbial etiology. Procalcitonin and C-reactive protein have been studied most extensively; whereof procalcitonin has demonstrated the strongest diagnostic performance compared to other biomarkers. Proadrenomedullin, soluble triggering receptor expressed on myeloid cells-1, neopterin and pentraxin-3 need more studies to confirm their diagnostic value. For syndecan-4 and lipocalin-2 currently insufficient evidence exists. Common limitations in several of the studies were the relatively small scale setting, heterogeneous patient population and the absence of statistical power calculation.

Pharmacodynamic modeling of cardiac biomarkers in breast cancer patients treated with anthracycline and trastuzumab regimens.

Trastuzumab is associated with cardiotoxicity, manifesting as a decrease of the left-ventricular ejection fraction (LVEF). Administration of anthracyclines prior to trastuzumab increases risk of cardiotoxicity. High-sensitive troponin T and N-terminal-pro-brain natriuretic peptide (NT-proBNP) are molecular markers that may allow earlier detection of drug-induced cardiotoxicity. In this analysis we aimed to quantify the kinetics and exposure-response relationships of LVEF, troponin T and NT-proBNP measurements, in patients receiving anthracycline and trastuzumab. Repeated measurements of LVEF, troponin T and NT-proBNP and dosing records of anthracyclines and trastuzumab were available from a previously published clinical trial. This trial included 206 evaluable patients with early breast cancer. Exposure to anthracycline and trastuzumab was simulated based on available dosing records and by using a kinetic-pharmacodynamic (K-PD) and a fixed pharmacokinetic (PK) model from literature, respectively. The change from baseline troponin T was described with a direct effect model, affected by simulated anthracycline concentrations, representing myocyte damage. The relationship between trastuzumab and LVEF was described by an indirect effect compartment model. The EC50 for LVEF decline was significantly affected by the maximum troponin T concentration after anthracycline treatment, explaining 15.1% of inter-individual variability. In this cohort, NT-proBNP changes could not be demonstrated to be related to anthracycline or trastuzumab treatment. Pharmacodynamic models for troponin T and LVEF were successfully developed, identifying maximum troponin T concentration after anthracycline treatment as a significant determinant for trastuzumab-induced LVEF decline. These models can help identify patients at risk of drug-induced cardiotoxicity and optimize cardiac monitoring strategies.

Evaluating renal function and age as predictors of amikacin clearance in neonates: model-based analysis and optimal dosing strategies.

AIMS: We aimed to compare the performance of renal function and age as predictors of inter-individual variability (IIV) in clearance of amikacin in neonates through parallel development of population pharmacokinetic (PK) models and their associated impact on optimal dosing regimens. METHODS: Amikacin concentrations were retrospectively collected for 149 neonates receiving amikacin (post-natal age (PNA) between 4-89 days). Two population PK models were developed in parallel, considering at least as predictors current body weight (WT), in combination with either creatinine clearance (CLcr ) or age descriptors. Using stochastic simulations for both renal function or age-based dosing, we identified optimal dosing strategies that were based on attainment of optimal peak- (PCC) and trough target concentration coverage (TCC) windows associated with efficacy and toxicity. RESULTS: The CLcr and age-based population PK models both included current body weight (WT) on CL, central distribution volume and intercompartmental clearance, in combination with either CLcr or PNA as predictors for IIV of clearance (CL). The WT-CLcr model explained 6.9% more IIV in CL compared with the WT-PNA model. Both models successfully described an external dataset (n = 53) of amikacin PK. The simulation analysis of optimal dose regimens suggested similar performance of either CLcr or PNA based dosing. CONCLUSION: CLcr predicted more IIV in CL, but did not translate into clinically relevant improvements of target concentrations. Our optimized dose regimens can be considered for further evaluation to optimize initial treatment with amikacin.

Pooled population pharmacokinetic model of imipenem in plasma and the lung epithelial lining fluid.

AIMS: Several clinical trials have confirmed the therapeutic benefit of imipenem for treatment of lung infections. There is however no knowledge of the penetration of imipenem into the lung epithelial lining fluid (ELF), the site of action relevant for lung infections. Furthermore, although the plasma pharmacokinetics (PK) of imipenem has been widely studied, most studies have been based on selected patient groups. The aim of this analysis was to characterize imipenem plasma PK across populations and to quantify imipenem ELF penetration. METHODS: A population model for imipenem plasma PK was developed using data obtained from healthy volunteers, elderly subjects and subjects with renal impairment, in order to identify predictors for inter-individual variability (IIV) of imipenem PK. Subsequently, a clinical study which measured plasma and ELF concentrations of imipenem was included in order to quantify lung penetration. RESULTS: A two compartmental model best described the plasma PK of imipenem. Creatinine clearance and body weight were included as subject characteristics predictive for IIV on clearance. Typical estimates for clearance, central and peripheral volume, and inter-compartmental clearance were 11.5 l h(-1) , 9.37 l, 6.41 l, 13.7 l h(-1) , respectively (relative standard error (RSE) <8%). The distribution of imipenem into ELF was described using a time-independent penetration coefficient of 0.44 (RSE 14%). CONCLUSION: The identified lung penetration coefficient confirms the clinical relevance of imipenem for treatment of lung infections, while the population PK model provided insights into predictors of IIV for imipenem PK and may be of relevance to support dose optimization in various subject groups.

Pronounced between-subject and circadian variability in thymidylate synthase and dihydropyrimidine dehydrogenase enzyme activity in human volunteers.

AIMS: The enzymatic activity of dihydropyrimidine dehydrogenase (DPD) and thymidylate synthase (TS) are important for the tolerability and efficacy of the fluoropyrimidine drugs. In the present study, we explored between-subject variability (BSV) and circadian rhythmicity in DPD and TS activity in human volunteers. METHODS: The BSVs in DPD activity (n = 20) in peripheral blood mononuclear cells (PBMCs) and in plasma, measured by means of the dihydrouracil (DHU) and uracil (U) plasma levels and DHU : U ratio (n = 40), and TS activity in PBMCs (n = 19), were examined. Samples were collected every 4 h throughout 1 day for assessment of circadian rhythmicity in DPD and TS activity in PBMCs (n = 12) and DHU : U plasma ratios (n = 23). In addition, the effects of genetic polymorphisms and gene expression on DPD and TS activity were explored. RESULTS: Population mean (± standard deviation) DPD activity in PBMCs and DHU : U plasma ratio were 9.2 (±2.1) nmol mg(-1) h(-1) and 10.6 (±2.4), respectively. Individual TS activity in PBMCs ranged from 0.024 nmol mg(-1) h(-1) to 0.596 nmol mg(-1) h(-1) . Circadian rhythmicity was demonstrated for all phenotype markers. Between 00:30 h and 02:00 h, DPD activity in PBMCs peaked, while the DHU : U plasma ratio and TS activity in PBMCs showed trough activity. Peak-to-trough ratios for DPD and TS activity in PBMCs were 1.69 and 1.62, respectively. For the DHU : U plasma ratio, the peak-to-trough ratio was 1.43. CONCLUSIONS: BSV and circadian variability in DPD and TS activity were demonstrated. Circadian rhythmicity in DPD might be tissue dependent. The results suggested an influence of circadian rhythms on phenotype-guided fluoropyrimidine dosing and supported implications for chronotherapy with high-dose fluoropyrimidine administration during the night.

Structure-Based Prediction of Anti-infective Drug Concentrations in the Human Lung Epithelial Lining Fluid.

PURPOSE: Obtaining pharmacologically relevant exposure levels of antibiotics in the epithelial lining fluid (ELF) is of critical importance to ensure optimal treatment of lung infections. Our objectives were to develop a model for the prediction of the ELF-plasma concentration ratio (EPR) of antibiotics based on their chemical structure descriptors (CSDs). METHODS: EPR data was obtained by aggregating ELF and plasma concentrations from historical clinical studies investigating antibiotics and associated agents. An elastic net regularized regression model was used to predict EPRs based on a large number of CSDs. The model was tuned using leave-one-drug-out cross validation, and the predictions were further evaluated using a test dataset. RESULTS: EPR data of 56 unique compounds was included. A high degree of variability in EPRs both between- and within drugs was apparent. No trends related to study design or pharmacokinetic factors could be identified. The model predicted 80% of the within-drug variability (R(2) WDV) and 78.6% of drugs were within 3-fold difference from the observations. Key CSDs were related to molecular size and lipophilicity. When predicting EPRs for a test dataset the R(2) WDV was 75%. CONCLUSIONS: This model is of relevance to inform dose selection and optimization during antibiotic drug development of agents targeting lung infections.

Integrated semi-physiological pharmacokinetic model for both sunitinib and its active metabolite SU12662.

AIMS: Previously published pharmacokinetic (PK) models for sunitinib and its active metabolite SU12662 were based on a limited dataset or lacked important elements such as correlations between sunitinib and its metabolite. The current study aimed to develop an improved PK model that circumvented these limitations and to prove the utility of the PK model in treatment optimization in clinical practice. METHODS: One thousand two hundred and five plasma samples from 70 cancer patients were collected from three PK studies with sunitinib and SU12662. A semi-physiological PK model for sunitinib and SU12662 was developed incorporating pre-systemic metabolism using non-linear mixed effects modelling (nonmem). Allometric scaling based on body weight was applied. The final model was used for simulation of the PK of different treatment regimens. RESULTS: Sunitinib and SU12662 PK were best described by a one and two compartment model, respectively. Introduction of pre-systemic formation of SU12662 strongly improved model fit, compared with solely systemic metabolism. The clearance of sunitinib and SU12662 was estimated at 35.7 (relative standard error (RSE) 5.7%) l h(-1) and 17.1 (RSE 7.4%) l h(-1), respectively for 70 kg patients. Correlation coefficients were estimated between inter-individual variability of both clearances, both volumes of distribution and between clearance and volume of distribution of SU12662 as 0.53, 0.48 and 0.45, respectively. Simulation of the PK model predicted correctly the ratio of patients who did not reach proposed PK targets for efficacy. CONCLUSIONS: A semi-physiological PK model for sunitinib and SU12662 in cancer patients was presented including pre-systemic metabolism. The model was superior to previous PK models in many aspects.

Towards integrative systems pharmacology models in oncology drug development.

Quantitative systems pharmacology (QSP) modeling represents an emerging area of value to further streamline knowledge integration and to better inform decision making processes in drug development. QSP models reside at the interface between systems biology models and pharmacological models, yet their concrete implementation still needs to be established further. This review outlines key modeling techniques in both of these areas and to subsequently discuss challenges and opportunities for further integration, in oncology drug development.

Design of informative renal impairment studies: evaluation of the impact of design stratification on bias, precision and dose adjustment error.

PURPOSE: Renal impairment (RI) studies are conducted to estimate the impact of RI on pharmacokinetics (PK). In some disease areas, these studies can be difficult to conduct, for instance due to the limited number of eligible patients. The objective of this analysis was to evaluate bias and precision of population PK parameters, and the dose adjustment error (DAE) for RI studies i) with different levels of study design imbalance in the stratification of subjects across RI categories, and ii) that include additional patients in the control arm of RI studies, that may be available from previously conducted PK studies. METHODS: Study designs were simulated and re-estimated using a hypothetical 2-compartmental PK model with varying magnitude of the fraction of renal elimination (FR) and magnitude of between-subject variability (BSV). The DAE was computed based on the difference between the theoretical necessary dose adjustment versus the empirical estimated dose adjustment to reach a similar exposure as controls. RESULTS: Although some design imbalance may still lead to DAEs of acceptable magnitude (DAE < -11.05-14.44 inter-quartile range, IQR), at least some patients are necessary in the more severe RI groups. When 100 additional patients with normal renal function were included in a sub-informative design, the DAE changed from < -7.63-16.64 IQR to < -8.89-8.69 IQR. CONCLUSIONS: We quantified the impact of study design imbalance on bias and precision of PK parameters and DAE, as may occur for RI studies in some indications. Adding additional data from earlier studies to the analysis dataset improves the bias and precision of PK parameters.

Design of a drug-drug interaction study of vincristine with azole antifungals in pediatric cancer patients using clinical trial simulation.

BACKGROUND: The aim of the current work was to perform a clinical trial simulation (CTS) analysis to optimize a drug-drug interaction (DDI) study of vincristine in children who also received azole antifungals, taking into account challenges of conducting clinical trials in this population, and, to provide a motivating example of the application of CTS in the design of pediatric oncology clinical trials. PROCEDURE: A pharmacokinetic (PK) model for vincristine in children was used to simulate concentration-time profiles. A continuous model for body surface area versus age was defined based on pediatric growth curves. Informative sampling time windows were derived using D-optimal design. The CTS framework was used to different magnitudes of clearance inhibition (10%, 25%, or 40%), sample size (30-500), the impact of missing samples or sampling occasions, and the age distribution, on the power to detect a significant inhibition effect, and in addition, the relative estimation error (REE) of the interaction effect. RESULTS: A minimum group specific sample size of 38 patients with a total sample size of 150 patients was required to detect a clearance inhibition effect of 40% with 80% power, while in the case of a lower effect of clearance inhibition, a substantially larger sample size was required. However, for the majority of re-estimated drug effects, the inhibition effect could be estimated precisely (REE < 25%) in even smaller sample sizes and with lower effect sizes. CONCLUSION: This work demonstrated the utility of CTS for the evaluation of PK clinical trial designs in the pediatric oncology population.

Non-linear blood-brain barrier transport and dosing strategies influence receptor occupancy ratios of morphine and its metabolites in pain matrix.

BACKGROUND AND PURPOSE: Morphine is important for treatment of acute and chronic pain. However, there is high interpatient variability and often inadequate pain relief and adverse effects. To better understand variability in the dose-effect relationships of morphine, we investigated the effects of its non-linear blood-brain barrier (BBB) transport on µ-receptor occupancy in different CNS locations, in conjunction with its main metabolites that bind to the same receptor. EXPERIMENTAL APPROACH: CNS exposure profiles for morphine, M3G and M6G for clinically relevant dosing regimens based on intravenous, oral immediate- and extended-release formulations were generated using a physiology-based pharmacokinetic model of the CNS, with non-linear BBB transport of morphine. The simulated CNS exposure profiles were then used to derive corresponding µ-receptor occupancies at multiple CNS pain matrix locations. KEY RESULTS: Simulated CNS exposure profiles for morphine, M3G and M6G, associated with non-linear BBB transport of morphine resulted in varying µ-receptor occupancies between different dose regimens, formulations and CNS locations. At lower doses, the µ-receptor occupancy of morphine was relatively higher than at higher doses of morphine, due to the relative contribution of M3G and M6G. At such higher doses, M6G showed higher occupancy than morphine, whereas M3G occupancy was low throughout the dose ranges. CONCLUSION AND IMPLICATIONS: Non-linear BBB transport of morphine affects the µ-receptor occupancy ratios of morphine with its metabolites, depending on dose and route of administration, and CNS location. These predictions need validation in animal or clinical experiments, to understand the clinical implications.

Predictions of Bedaquiline Central Nervous System Exposure in Patients with Tuberculosis Meningitis Using Physiologically based Pharmacokinetic Modeling.

BACKGROUND AND OBJECTIVE: The use of bedaquiline as a treatment option for drug-resistant tuberculosis meningitis (TBM) is of interest to address the increased prevalence of resistance to first-line antibiotics. To this end, we describe a whole-body physiologically based pharmacokinetic (PBPK) model for bedaquiline to predict central nervous system (CNS) exposure. METHODS: A whole-body PBPK model was developed for bedaquiline and its metabolite, M2. The model included compartments for brain and cerebrospinal fluid (CSF). Model predictions were evaluated by comparison to plasma PK time profiles following different dosing regimens and sparse CSF concentrations data from patients. Simulations were then conducted to compare CNS and lung exposures to plasma exposure at clinically relevant dosing schedules. RESULTS: The model appropriately described the observed plasma and CSF bedaquiline and M2 concentrations from patients with pulmonary tuberculosis (TB). The model predicted a high impact of tissue binding on target site drug concentrations in CNS. Predicted unbound exposures within brain interstitial exposures were comparable with unbound vascular plasma and unbound lung exposures. However, unbound brain intracellular exposures were predicted to be 7% of unbound vascular plasma and unbound lung intracellular exposures. CONCLUSIONS: The whole-body PBPK model for bedaquiline and M2 predicted unbound concentrations in brain to be significantly lower than the unbound concentrations in the lung at clinically relevant doses. Our findings suggest that bedaquiline may result in relatively inferior efficacy against drug-resistant TBM when compared with efficacy against drug-resistant pulmonary TB.