Combining data on the bioavailability of midazolam and physiologically-based pharmacokinetic modeling to investigate intestinal CYP3A4 ontogeny.

Pediatric physiologically-based modeling in drug development has grown in the past decade and optimizing the underlying systems parameters is important in relation to overall performance. In this study, variation of clinical oral bioavailability of midazolam as a function of age is used to assess the underlying ontogeny models for intestinal CYP3A4. Data on midazolam bioavailability in adults and children and different ontogeny patterns for intestinal CYP3A4 were first collected from the literature. A pediatric PBPK model was then used to assess six different ontogeny models in predicting bioavailability from preterm neonates to adults. The average fold error ranged from 0.7 to 1.38, with the rank order of least to most biased model being No Ontogeny < Upreti = Johnson < Goelen < Chen < Kiss. The absolute average fold error ranged from 1.17 to 1.64 with the rank order of most to least precise being Johnson > Upreti > No Ontogeny > Goelen > Kiss > Chen. The optimal ontogeny model is difficult to discern when considering the possible influence of CYP3A5 and other population variability; however, this study suggests that from term neonates and older a faster onset Johnson model with a lower fraction at birth may be close to this. For inclusion in other PBPK models, independent verification will be needed to confirm these results. Further research is needed in this area both in terms of age-related changes in midazolam and similar drug bioavailability and intestinal CYP3A4 ontogeny.

Prediction of Pediatric Pharmacokinetics for CYP3A4 Metabolized Drugs: Comparison of the Performance of Two Hepatic Ontogeny Within a Physiologically Based Pharmacokinetic Model.

The rapid growth in the use of pediatric physiologically based pharmacokinetic (PBPK) models, particularly for regulatory applications, has focused emphasis on model verification and ensuring system parameters are robust, including how these change with age. Uncertainty remains regarding the ontogeny of some enzymes and transporters, in this study 2 published ontogeny profiles for hepatic CYP3A4 were compared. Clinical pharmacokinetic data on 4 intravenously administered CYP3A4 substrates (alfentanil, fentanyl, midazolam, and sildenafil) used across the pediatric age range was collected from the literature. The PBPK models were verified in the adult population and then used to compare the Salem and a modified Upreti ontogeny profiles for CYP3A4 in terms of parent drug clearance and area under the curve from birth onward. Overall, the modified Upreti ontogeny profile resulted in 15 out of 17 age-related predictions within 2-fold and 12 out of 17 predictions within 1.5-fold ranges of observed values, for the Salem ontogeny these values were 12 out of 17 and 8 out of 17, respectively. The Upreti ontogeny profile performed better than Salem, average fold error and absolute average fold error were 1.14 and 1.35 compared to 1.56 and 1.90, respectively. Identifying the optimal CYP3A4 ontogeny is important for regulatory use of PBPK especially given the number of drugs cleared by this enzyme. This study broadens the evidence from previous studies that Upreti is more favorable than Salem, but further work is needed especially in the neonatal and early infant age range.

Physiologically Based Pharmacokinetics Modeling in the Neonatal Population-Current Advances, Challenges, and Opportunities.

Physiologically based pharmacokinetic (PBPK) modeling is an approach to predicting drug pharmacokinetics, using knowledge of the human physiology involved and drug physiochemical properties. This approach is useful when predicting drug pharmacokinetics in under-studied populations, such as pediatrics. PBPK modeling is a particularly important tool for dose optimization for the neonatal population, given that clinical trials rarely include this patient population. However, important knowledge gaps exist for neonates, resulting in uncertainty with the model predictions. This review aims to outline the sources of variability that should be considered with developing a neonatal PBPK model, the data that are currently available for the neonatal ontogeny, and lastly to highlight the data gaps where further research would be needed.

Impact of CYP2C:TG Haplotype on CYP2C19 Substrates Clearance In Vivo, Protein Content, and In Vitro Activity.

A novel haplotype composed of two non-coding variants, CYP2C18 NM_000772.3:c.*31T (rs2860840) and NM_000772.2:c.819+2182G (rs11188059), referred to as "CYP2C:TG," was recently associated with ultrarapid metabolism of various CYP2C19 substrates. As the underlying mechanism and clinical relevance of this effect remain uncertain, we analyzed existing in vivo and in vitro data to determine the magnitude of the CYP2C:TG haplotype effect. We assessed variability in pharmacokinetics of CYP2C19 substrates, including citalopram, sertraline, voriconazole, omeprazole, pantoprazole, and rabeprazole in 222 healthy volunteers receiving one of these six drugs. We also determined its impact on CYP2C8, CYP2C9, CYP2C18, and CYP2C19 protein abundance in 135 human liver tissue samples, and on CYP2C18/CYP2C19 activity in vitro using N-desmethyl atomoxetine formation. No effects were observed according to CYP2C:TG haplotype or to CYP2C19*1+TG alleles (i.e., CYP2C19 alleles containing the CYP2C:TG haplotype). In contrast, CYP2C19 intermediate (e.g., CYP2C19*1/*2) and poor metabolizers (e.g., CYP2C19*2/*2) showed significantly higher exposure in vivo, lower CYP2C19 protein abundance in human liver microsomes, and lower activity in vitro compared with normal, rapid (i.e., CYP2C19*1/*17), and ultrarapid metabolizers (i.e., CYP2C19*17/*17). Moreover, a tendency toward lower exposure was observed in ultrarapid metabolizers compared with rapid metabolizers and normal metabolizers. Furthermore, when the CYP2C19*17 allele was present, CYP2C18 protein abundance was increased suggesting that genetic variation in CYP2C19 may be relevant to the overall metabolism of certain drugs by regulating not only its expression levels, but also those of CYP2C18. Considering all available data, we conclude that there is insufficient evidence supporting clinical CYP2C:TG testing to inform drug therapy.

Cardiac Disease Alters Myocardial Tissue Levels of Epoxyeicosatrienoic Acids and Key Proteins Involved in Their Biosynthesis and Degradation.

CYP2J2 is the main epoxygenase in the heart that is responsible for oxidizing arachidonic acid to cis-epoxyeicosatrienoic acids (EETs). Once formed, EETs can then be hydrolyzed by soluble epoxide hydrolase (sEH, encoded by EPHX2) or re-esterified back to the membrane. EETs have several cardioprotective properties and higher levels are usually associated with better cardiac outcomes/prognosis. This study investigates how cardiovascular disease (CVD) can influence total EET levels by altering protein expression and activity of enzymes involved in their biosynthesis and degradation. Diseased ventricular cardiac tissues were collected from patients receiving Left Ventricular Assist Device (LVAD) or heart transplants and compared to ventricular tissue from controls free of CVD. EETs, and enzymes involved in EETs biosynthesis and degradation, were measured using mass spectrometric assays. Terfenadine hydroxylation was used to probe CYP2J2 activity. Significantly higher cis- and trans-EET levels were observed in control cardiac tissue (n = 17) relative to diseased tissue (n = 24). Control cardiac tissue had higher CYP2J2 protein levels, which resulted in higher rate of terfenadine hydroxylation, compared to diseased cardiac tissues. In addition, levels of both NADPH-Cytochrome P450 oxidoreductase (POR) and sEH proteins were significantly higher in control versus diseased cardiac tissue. Overall, alterations in protein and activity of enzymes involved in the biosynthesis and degradation of EETs provide a mechanistic understanding for decreased EET levels in diseased tissues.

Selective deuteration of bupropion slows epimerization and reduces metabolism.

Strategically replacing hydrogen with deuterium at sites of metabolism in small molecule drugs can significantly alter clearance and potentially enhance clinical safety. Bupropion is an antidepressant and smoking cessation medication with the potential to cause seizures. We hypothesized that incorporating deuterium at specific sites in bupropion may greatly reduce epimerization, potentially slow metabolism, and reduce the formation of toxic metabolites, namely hydroxybupropion which has been associated with bupropion's toxicity. Four deuterated analogues were synthesized incorporating deuterium at sites of metabolism and epimerization with the aim of altering the metabolic profile of bupropion. Spectroscopic binding and metabolism studies with bupropion and R-or S-d4 and R-or S-d10 analogs were performed with recombinant CYP2B6, human liver microsomes, and human hepatocytes. Results demonstrate that deuterated bupropion analogues exhibited 20-25% decrease in racemization and displayed a significant decrease in the formation of CYP2B6-mediated R,R - or S,S-hydroxybupropion with recombinant protein and human liver microsomes. In primary human hepatocytes, metabolism of deuterated analogs to R,R - and S,S-hydroxybupropion and threo- and erythro-hydrobupropion was significantly less than R/S-d0 bupropion. Selective deuterium substitution at metabolic soft spots in bupropion has the potential to provide a drug with a simplified pharmacokinetic profile, reduced toxicity and improved tolerability in patients.

Ontogeny of Scaling Factors for Pediatric Physiology-Based Pharmacokinetic Modeling and Simulation: Microsomal Protein Per Gram of Liver.

Microsomal protein per gram of liver (MPPGL) is an important scaling factor for bottom-up physiology-based pharmacokinetic modeling and simulation, but data in pediatrics are limited. Therefore, MPPGL was determined in 160 liver samples from pediatric (n = 129) and adult (n = 31) donors obtained from four sources: the University of Maryland Brain and Tissue Bank (UMBTB), tissue retrieval services at the University of Minnesota and University of Pittsburgh, and Sekisui-Xenotech. Tissues were homogenized and subjected to differential centrifugation to prepare microsomes, and cytochrome c reductase activities in tissue homogenates and microsomes were used to estimate cytochrome P450 reductase (POR) activity as a marker of microsomal recovery; microsomal POR content was also assessed by quantitative proteomics. MPPGL values varied 5- to 10-fold within various age groups/developmental stages, and tissue source was identified as a contributing factor. Using a "trimmed" dataset comprised of samples ranging from 3 to 18 years of age common to the four sources, POR protein abundance and activity in microsomes and POR activity in homogenates was lower in UMBTB samples (autopsy) compared with other sources (perfused/flash-frozen). Regression analyses revealed that the UMBTB samples were driving an apparent age effect as no effect of age on log-transformed MPPGL values was observed when the UMBTB samples were excluded. We conclude that a mean±SD MPPGL value of 30.4±1.7 mg/g is representative between one month postnatal age and early adulthood. Potential source effects should be considered for studies involving tissue samples from multiple sources with different procurement and processing procedures. SIGNIFICANCE STATEMENT: Microsomal protein per gram of liver (MPPGL) is an important scaling factor for bottom up PBPK modeling and simulation, but data in pediatrics are limited. Although MPPGL varies 5- to 10-fold at a given developmental stage, a value of 30.4 ± 1.7 mg/g (mean ± SD) is representative between one month postnatal age and early adulthood. However, when tissue samples are obtained from multiple sources, different procurement and processing procedures may influence the results and should be taken into consideration.

The Impact of the CYP2D6 "Enhancer" Single Nucleotide Polymorphism on CYP2D6 Activity.

rs5758550 has been associated with enhanced transcription and suggested to be a useful marker of CYP2D6 activity. As there are limited and inconsistent data regarding the utility of this distant "enhancer" single nucleotide polymorphism (SNP), our goal was to further assess the impact of rs5758550 on CYP2D6 activity toward two probe substrates, atomoxetine (ATX) and dextromethorphan (DM), using in vivo urinary metabolite (DM; n = 188) and pharmacokinetic (ATX; n = 70) and in vitro metabolite formation (ATX and DM; n = 166) data. All subjects and tissues were extensively genotyped, the "enhancer" SNP phased with established CYP2D6 haplotypes either computationally or experimentally, and the impact on CYP2D6 activity investigated using several linear models of varying complexity to determine the proportion of variability in CYP2D6 activity captured by each model. For all datasets and models, the "enhancer" SNP had no or only a modest impact on CYP2D6 activity prediction. An increased effect, when present, was more pronounced for ATX than DM suggesting potential substate-dependency. In addition, CYP2D6*2 alleles with the "enhancer" SNP were associated with modestly higher metabolite formation rates in vitro, but not in vivo; no effect was detected for CYP2D6*1 alleles with "enhancer" SNP. In summary, it remains inconclusive whether the small effects detected in this investigation are indeed caused by the "enhancer" SNP or are rather due to its incomplete linkage with other variants within the gene. Taken together, there does not appear to be sufficient evidence to warrant the "enhancer" SNP be included in clinical CYP2D6 pharmacogenetic testing.

Introduction to Computational Modeling of Multicellular Tissues.

The study of biological tissues is extremely complicated, as they comprise mechanisms and properties at many different temporal and spatial scales. For this reason, modeling is becoming one of the most active and important research fields for the analysis and understanding of tissues. However, this is not a simple task, as it requires mathematical and computational skills, as well as the development of software tools for its implementation. Here, we provide an introduction covering some of the most important and basic issues for modeling tissues. In particular, we focus on both the chemical and cellular properties of a tissue. We explain how to represent and couple these properties within a virtual tissue. All our examples were done using Multicell, a Python library that simplifies their reproducibility, even by readers with little experience in biological modeling.

The Respective Roles of CYP3A4 and CYP2D6 in the Metabolism of Pimozide to Established and Novel Metabolites.

Pimozide is a dopamine receptor antagonist indicated for the treatment of Tourette syndrome. Prior in vitro studies characterized N-dealkylation of pimozide to 1,3-dihydro-1-(4-piperidinyl)-2H-benzimidazol-2-one (DHPBI) via CYP3A4 and, to a lesser extent, CYP1A2 as the only notable routes of pimozide biotransformation. However, drug-drug interactions between pimozide and CYP2D6 inhibitors and CYP2D6 genotype-dependent effects have since been observed. To reconcile these incongruities between the prior in vitro and in vivo studies, we characterized two novel pimozide metabolites: 5-hydroxypimozide and 6-hydroxypimozide. Notably, 5-hydroxypimozide was the major metabolite produced by recombinant CYP2D6 (Km ∼82 nM, V max ∼0.78 pmol/min per picomoles), and DHPBI was the major metabolite produced by recombinant CYP3A4 (apparent Km ∼1300 nM, V max ∼2.6 pmol/min per picomoles). Kinetics in pooled human liver microsomes (HLMs) for the 5-hydroxylation (Km ∼2200 nM, V max ∼59 pmol/min per milligram) and N-dealkylation (Km ∼3900 nM, V max ∼600 pmol/min per milligram) reactions were also determined. Collectively, formation of DHPBI, 5-hydroxypimozide, and 6-hydroxypimozide accounted for 90% of pimozide depleted in incubations of NADPH-supplemented pooled HLMs. Studies conducted in HLMs isolated from individual donors with specific cytochrome P450 isoform protein abundances determined via mass spectrometry revealed that 5-hydroxypimozide (r 2 = 0.94) and 6-hydroxypimozide (r 2 = 0.86) formation rates were correlated with CYP2D6 abundance, whereas the DHPBI formation rate (r 2 = 0.98) was correlated with CYP3A4 abundance. Furthermore, the HLMs differed with respect to their capacity to form 5-hydroxypimozide relative to DHPBI. Collectively, these data confirm a role for CYP2D6 in pimozide clearance via 5-hydroxylation and provide an explanation for a lack of involvement when only DHPBI formation was monitored in prior in vitro studies. SIGNIFICANCE STATEMENT: Current CYP2D6 genotype-guided dosing information in the pimozide label is discordant with available knowledge regarding the primary biotransformation pathways. Herein, we characterize the CYP2D6-dependent biotransformation of pimozide to previously unidentified metabolites. In human liver microsomes, formation rates for the novel metabolites and a previously identified metabolite were determined to be a function of CYP2D6 and CYP3A4 content, respectively. These findings provide a mechanistic basis for observations of CYP2D6 genotype-dependent pimozide clearance in vivo.

Pediatric Clinical Endpoint and Pharmacodynamic Biomarkers: Limitations and Opportunities.

Medical research in children typically lags behind that of adult research in both quantity and quality. The conduct of rigorous clinical trials in children can raise ethical concerns because of children's status as a 'vulnerable' population. Moreover, carrying out studies in pediatrics also requires logistical considerations that rarely occur with adult clinical trials. Due to the relatively smaller number of pediatric studies to support evidence-based medicine, the practice of medicine in children is far more reliant upon expert opinion than in adult medicine. Children are at risk of not receiving the same level of benefits from precision medicine research, which has flourished with new technologies capable of generating large amounts of data quickly at an individual level. Although progress has been made in pediatric pharmacokinetics, which has led to safer and more effective dosing, gaps in knowledge still exists when it comes to characterization of pediatric disease and differences in pharmacodynamic response between children and adults. This review highlights three specific therapeutic areas where biomarker development can enhance precision medicine in children: asthma, type 2 diabetes mellitus, and pain. These 'case studies' are meant to update the reader on biomarkers used currently in the diagnosis and treatment of these conditions, and their shortcomings within a pediatric context. Current research on surrogate endpoints and pharmacodynamic biomarkers in the above therapeutic areas will also be described. These cases highlight the current lack in pediatric specific surrogate endpoints and pharmacodynamic biomarkers, as well as the research presently being conducted to address these deficiencies. We finally briefly highlight other therapeutic areas where further research in pediatric surrogate endpoints and pharmacodynamic biomarkers can be impactful to the care of children.

Description of an Innovative Pediatric Individualized Therapeutics Clinic: Working toward Precision Drug Therapy.

The GOLDILOKs® (Genomic and Ontogeny-Linked Dose Individualization and cLinical Optimization for KidS) Clinic aims to provide families and physicians with data to make more informed decisions with regard to pharmacological therapy by using innovative therapy and genomic technologies. The objectives are two-fold: (1) To describe the utility of the GOLDILOKs® Clinic to referring prescribers by evaluating the type of referrals made to the GOLDILOKs® Clinic and (2) to assess the most often utilized technologies (e.g., genotyping) completed to formulate therapy recommendations. Patient data from July 2010 to June 2016 was retrospectively reviewed following Institutional Review Board (IRB) approval. The GOLDILOKs® Clinic evaluated 306 patients and had increases in annual referrals from 14 in 2010⁻2011 to 84 in 2016⁻2017. The children that were referred were predominately Caucasian (82%) and male (59%) with an average age of 12.4 ± 5.9 years. Subspecialty versus primary care referrals accounted for 82% and 18% of referrals, respectively. Adverse drug reactions (n = 166) and poor medication response (n = 179) were the major reasons for referral. However, it must be noted that patients could have multiple reasons for referral. Pharmacogenetic results were extensively used to provide guidance for future therapy in patients with medication-related problems. Genotyping of drug metabolizing enzymes and drug target receptors was performed in 221 patients (72.2%). Recommendations were fully accepted by 63% and partially accepted by 22% of internal provider referrals.

Clinical Pharmacogenetics Implementation Consortium Guideline for Cytochrome P450 (CYP)2D6 Genotype and Atomoxetine Therapy.

Atomoxetine is a nonstimulant medication used to treat attention-deficit/hyperactivity disorder (ADHD). Cytochrome P450 (CYP)2D6 polymorphisms influence the metabolism of atomoxetine thereby affecting drug efficacy and safety. We summarize evidence from the published literature supporting these associations and provide therapeutic recommendations for atomoxetine based on CYP2D6 genotype (updates at www.cpicpgx.org).

Ten Years' Experience with the CYP2D6 Activity Score: A Perspective on Future Investigations to Improve Clinical Predictions for Precision Therapeutics.

The seminal paper on the CYP2D6 Activity Score (AS) was first published ten years ago and, since its introduction in 2008, it has been widely accepted in the field of pharmacogenetics. This scoring system facilitates the translation of highly complex CYP2D6 diplotype data into a patient’s phenotype to guide drug therapy and is at the core of all CYP2D6 gene/drug pair guidelines issued by the Clinical Pharmacogenetics Implementation Consortium (CPIC). The AS, however, only explains a portion of the variability observed among individuals and ethnicities. In this review, we provide an overview of sources in addition to CYP2D6 genotype that contribute to the variability in CYP2D6-mediated drug metabolism and discuss other factors, genetic and non-genetic, that likely contribute to the observed variability in CYP2D6 enzymatic activity.

The logic of the floral transition: Reverse-engineering the switch controlling the identity of lateral organs.

Much laboratory work has been carried out to determine the gene regulatory network (GRN) that results in plant cells becoming flowers instead of leaves. However, this also involves the spatial distribution of different cell types, and poses the question of whether alternative networks could produce the same set of observed results. This issue has been addressed here through a survey of the published intercellular distribution of expressed regulatory genes and techniques both developed and applied to Boolean network models. This has uncovered a large number of models which are compatible with the currently available data. An exhaustive exploration had some success but proved to be unfeasible due to the massive number of alternative models, so genetic programming algorithms have also been employed. This approach allows exploration on the basis of both data-fitting criteria and parsimony of the regulatory processes, ruling out biologically unrealistic mechanisms. One of the conclusions is that, despite the multiplicity of acceptable models, an overall structure dominates, with differences mostly in alternative fine-grained regulatory interactions. The overall structure confirms the known interactions, including some that were not present in the training set, showing that current data are sufficient to determine the overall structure of the GRN. The model stresses the importance of relative spatial location, through explicit references to this aspect. This approach also provides a quantitative indication of how likely some regulatory interactions might be, and can be applied to the study of other developmental transitions.

Characterization of Atomoxetine Biotransformation and Implications for Development of PBPK Models for Dose Individualization in Children.

Atomoxetine (ATX) is a second-line nonstimulant medication used to control symptoms of attention deficit hyperactivity disorder (ADHD). Inconsistent therapeutic efficacy has been reported with ATX, which may be related to variable CYP2D6-mediated drug clearance. We characterized ATX metabolism in a panel of human liver samples as a basis for a bottom-up PBPK model to aid in ATX exposure prediction and control. Km, Vmax, and Clint values in pooled human liver microsomes (HLMs) were 2.4 µM, 479 pmol/min/mg protein, and 202 µl/min/mg protein, respectively. Mean population values of kinetic parameters are not adequate to describe variability in a population, given that Km, Vmax, and Clint values from single-donor HLMs ranged from 0.93 to 79.2 µM, 20.0 to 1600 pmol/min/mg protein, and 0.3 to 936 µl/min/mg protein. All kinetic parameters were calculated from 4-hydroxyatomoxetine (4-OH-ATX) formation. CYP2E1 and CYP3A contributed to 4-OH-ATX formation in livers with CYP2D6 intermediate and poor metabolizer status. In HLMs with lower CYP2D6 activity levels, 2-hydroxymethylatomoxetine (2-CH2OH-ATX) formation became a more predominant pathway of metabolism, which appeared to be catalyzed by CYP2B6. ATX biotransformation at clinically relevant plasma concentrations was characterized in a panel of pediatric HLM (n = 116) samples by evaluating primary metabolites. Competing pathways of ATX metabolism [N-desmethylatomoxetine (NDM-ATX) and 2-CH2OH-ATX formation] had increasing importance in livers with lesser CYP2D6 activity, but, overall ATX clearance was still compromised. Modeling ATX exposure to individualize therapy would require comprehensive knowledge of factors that affect CYP2D6 activity as well as an understanding of competing pathways, particularly for individuals with lower CYP2D6 activity.

Potential contribution of cytochrome P450 2B6 to hepatic 4-hydroxycyclophosphamide formation in vitro and in vivo.

Results from retrospective studies on the relationship between cytochrome P450 (P450) 2B6 (CYP2B6) genotype and cyclophosphamide (CY) efficacy and toxicity in adult cancer patients have been conflicting. We evaluated this relationship in children, who have faster CY clearance and receive different CY-based regimens than adults. These factors may influence the P450s metabolizing CY to 4-hydroxycyclophosphamide (4HCY), the principal precursor to CY's cytotoxic metabolite. Therefore, we sought to characterize the in vitro and in vivo roles of hepatic CYP2B6 and its main allelic variants in 4HCY formation. CYP2B6 is the major isozyme responsible for 4HCY formation in recombinant P450 Supersomes. In human liver microsomes (HLM), 4HCY formation correlated with known phenotypic markers of CYP2B6 activity, specifically formation of (S)-2-ethyl-1,5-dimethyl-3,3-diphenyl pyrrolidine and hydroxybupropion. However, in HLM, CYP3A4/5 also contributes to 4HCY formation at the CY concentrations similar to plasma concentrations achieved in children (0.1 mM). 4HCY formation was not associated with CYP2B6 genotype at low (0.1 mM) or high (1 mM) CY concentrations potentially because CYP3A4/5 and other isozymes also form 4HCY. To remove this confounder, 4HCY formation was evaluated in recombinant CYP2B6 enzymes, which demonstrated that 4HCY formation was lower for CYP2B6.4 and CYP2B6.5 compared with CYP2B6.1. In vivo, CYP2B6 genotype was not directly related to CY clearance or ratio of 4HCY/CY areas under the curve in 51 children receiving CY-based regimens. Concomitant chemotherapy agents did not influence 4HCY formation in vitro. We conclude that CYP2B6 genotype is not consistently related to 4HCY formation in vitro or in vivo.