Biology-inspired microphysiological systems to advance patient benefit and animal welfare in drug development

The first microfluidic microphysiological systems (MPS) entered the academic scene more than 15 years ago and were considered an enabling technology to human (patho)biology in vitro and, therefore, provide alternative approaches to laboratory animals in pharmaceutical drug development and academic research. Nowadays, the field generates more than a thousand scientific publications per year. Despite the MPS hype in academia and by platform providers, which says this technology is about to reshape the entire in vitro culture landscape in basic and applied research, MPS approaches have neither been widely adopted by the pharmaceutical industry yet nor reached regulated drug authorization processes at all. Here, 46 leading experts from all stakeholders - academia, MPS supplier industry, pharmaceutical and consumer products industries, and leading regulatory agencies - worldwide have analyzed existing challenges and hurdles along the MPS-based assay life cycle in a second workshop of this kind in June 2019. They identified that the level of qualification of MPS-based assays for a given context of use and a communication gap between stakeholders are the major challenges for industrial adoption by end-users. Finally, a regulatory acceptance dilemma exists against that background. This t4 report elaborates on these findings in detail and summarizes solutions how to overcome the roadblocks. It provides recommendations and a roadmap towards regulatory accepted MPS-based models and assays for patients' benefit and further laboratory animal reduction in drug development. Finally, experts highlighted the potential of MPS-based human disease models to feedback into laboratory animal replacement in basic life science research.

Intersubject and Intrasubject Variability of Potential Plasma and Urine Metabolite and Protein Biomarkers in Healthy Human Volunteers.

A limited understanding of intersubject and intrasubject variability hampers effective biomarker translation from in vitro/in vivo studies to clinical trials and clinical decision support. Specifically, variability of biomolecule concentration can play an important role in interpretation, power analysis, and sampling time designation. In the present study, a wide range of 749 plasma metabolites, 62 urine biogenic amines, and 1,263 plasma proteins were analyzed in 10 healthy male volunteers measured repeatedly during 12 hours under tightly controlled conditions. Three variability components in relative concentration data are determined using linear mixed models: between (intersubject), time (intrasubject), and noise (intrasubject). Biomolecules such as 3-carboxy-4-methyl-5-propyl-2-furanpropanoate, platelet-derived growth factor C, and cathepsin D with low noise potentially detect changing conditions within a person. If also the between component is low, biomolecules can easier differentiate conditions between persons, for example cathepsin D, CD27 antigen, and prolylglycine. Variability over time does not necessarily inhibit translatability, but requires choosing sampling times carefully.

Synthesis and structure activity relationships of glycine amide derivatives as novel Vascular Adhesion Protein-1 inhibitors.

Vascular Adhesion Protein-1 (VAP-1) is a promising therapeutic target for the treatment of several inflammatory-related diseases including diabetic microvascular complication. We identified glycine amide derivative 3 as a novel structure with moderate VAP-1 inhibitory activity. Structure-activity relationship studies of glycine amide derivatives revealed that the tertiary amide moiety is important for stability in rat blood and that the position of substituents on the left phenyl ring plays an important role in VAP-1 inhibitory activity. We also found that low TPSA values and weak basicity are both important for high PAMPA values for glycine amide derivatives. These findings led to the identification of a series of orally active compounds with enhanced VAP-1 inhibitory activity. Of these compounds, 4g exhibited the most potent ex vivo efficacy, with plasma VAP-1 inhibitory activity of 60% after oral administration at 1mg/kg.

Synergistic antitumor activities of sepantronium bromide (YM155), a survivin suppressant, in combination with microtubule-targeting agents in triple-negative breast cancer cells.

Triple-negative breast cancer (TNBC) has a poor prognosis compared to other subtypes, and effective treatment options are limited to cytotoxic agents, including microtubule-targeting agents, due to the lack of molecular targets. Here, we examined the combined effect of sepantronium bromide (YM155) and microtubule-targeting agents in TNBC models. The combination of YM155 with docetaxel showed synergistic antiproliferative and caspase 3/7-inducing effects in MRK-nu-1 and MDA-MB-453 human TNBC cell lines in vitro. YM155 also synergistically enhanced the efficacies of other microtubule-targeting agents, including paclitaxel and vinorelbine, which induced accumulation of survivin at the G2/M phase, whereas it did not affect the efficacy of doxorubicin. Combination treatment with YM155 and microtubule-targeting agents decreased the accumulation of survivin at the G2/M phase and induced greater apoptosis than either single agent alone. Further, combination treatment with YM155 and docetaxel also had a synergistic antitumor effect, achieving complete regression without exacerbation of body weight loss in all mice, in a MRK-nu-1 human TNBC xenograft model. These results suggest that survivin inhibition synergistically sensitize human TNBC cells to microtubule-targeting agents.

A quantitative approach to hepatic clearance prediction of metabolism by aldehyde oxidase using custom pooled hepatocytes.

We describe the usability of human pooled hepatocytes for non-CYP metabolism evaluation and an in vivo-in vitro correlation analysis for aldehyde oxidase (AO) substrate compounds using pooled hepatocytes. By comparing intrinsic clearance values of 18 compounds primarily metabolized by AO, UDP-glucuronosyltransferase, carbonyl/aldo-keto reductase, flavin-containing monooxygenase, and monoamineoxidase in individual hepatocytes and pooled hepatocytes from the same individual donors, intrinsic clearance in the pooled hepatocytes was ± 30% of the average clearance value in individuals for 15 of 18 compounds, suggesting that pooled hepatocytes maintained the average activity of the individual hepatocytes. Although the results of an in vivo-in vitro correlation analysis for AO substrate compounds showed a trend toward under-prediction, the underestimation ratios for all AO substrates were nevertheless comparable (7.2- to 14.9-fold), suggesting that hepatic clearance prediction for these compounds can be quantified using empirical scaling. These observations enabled us to obtain specific pooled hepatocytes that showed the expected non-CYP enzyme activities by pre-characterization and to quantify hepatic clearance prediction for AO compounds using an empirical scaling factor.

A practical and direct comparison of intrinsic metabolic clearance of several non-CYP enzyme substrates in freshly isolated and cryopreserved hepatocytes.

Human hepatocytes are a physiologically relevant tool useful in evaluating liver-related pharmacokinetics, including non-cytochrome P-450 (CYP) metabolism, due to their broad spectrum of metabolic enzyme activity. To verify the usefulness of human hepatocytes in evaluating non-CYP metabolism for drug discovery, we compared intrinsic clearance values (CL(int)) in freshly isolated and cryopreserved hepatocytes using 14 compounds primarily metabolized by non-CYP enzymes, including UDP-glucuronosyltransferase, carbonyl/aldo-keto reductase, aldehyde oxidase, flavin-containing monooxygenase, and monoamineoxidase. Cryopreservation resulted in a >20% reduction (maximum: 50%) in CL(int) in 7/14 compounds (statistically significant for 5 compounds) on comparing CL(int) values in freshly isolated and cryopreserved hepatocytes from the same donors (n = 4). However, the number of compounds with >20% CL(int) reduction decreased to 3 on comparing average of CL(int) values including un-matched donors (dolasetron: -27%, naltorexone: -32%, and phthalazine: -48%; statistically significant for phthalazine, n = 6-11). These findings suggest that fresh hepatocytes are useful in evaluating intact non-CYP enzyme activities. However, we must note that the reduction in CL(int) by cryopreservation could be rendered negligible if high-activity lots are selected for assay. We therefore recommend using cryopreserved hepatocytes for large-scale screening for non-CYP metabolism in drug discovery research considering the advantages in usability with cryopreserved hepatocytes.

Case report of extensive metabolism by aldehyde oxidase in humans: pharmacokinetics and metabolite profile of FK3453 in rats, dogs, and humans.

We describe the preclinical and clinical pharmacokinetic profiles of FK3453 [6-(2-amino-4-phenylpyrimidin-5-yl)-2-isopropylpyridazin-3(2H)-one] and the mechanism responsible for poor oral exposure of FK3453 in humans. FK3453 showed favourable profiles in preclinical pharmacokinetic studies, including satisfactory absolute bioavailability and total body clearance in animals (30.5%-41.4%, 54.7%-68.2%, and 71.3%-93.4% and 10.8-17.6, 1.9-17.1, and 5.0 mL/min/kg in male rats, female rats, and dogs, respectively), and good metabolic stability in liver microsomes (42.3, 14.5, and 1.1 mL/min/kg in male rats, dogs, and humans, respectively). However, despite these promising preclinical findings, plasma concentrations of FK3453 in humans were extremely low, with the oxidative metabolite of the aminopyrimidine moiety (M4) identified as a major metabolite. Given that aldehyde oxidase (AO) and xanthine oxidase (XO) were presumed to be the enzymes responsible for M4 formation, we investigated the mechanism of M4 formation using human liver subcellular fractions. M4 was detected in the incubation mixture with S9 and cytosol but not with microsomes, and M4 formation was inhibited by AO inhibitors (menadione, isovanillin) but not by cytochrome P-450 inhibitor (1-aminobenzotiazole) or XO inhibitor (allopurinol). These results suggest M4 formation is catalyzed by AO, and therefore, its poor exposure in humans was attributed to extensive AO metabolism.

Quantitative prediction of intestinal metabolism in humans from a simplified intestinal availability model and empirical scaling factor.

This study aimed to establish a practical and convenient method of predicting intestinal availability (F(g)) in humans for highly permeable compounds at the drug discovery stage, with a focus on CYP3A4-mediated metabolism. We constructed a "simplified F(g) model," described using only metabolic parameters, assuming that passive diffusion is dominant when permeability is high and that the effect of transporters in epithelial cells is negligible. Five substrates for CYP3A4 (alprazolam, amlodipine, clonazepam, midazolam, and nifedipine) and four for both CYP3A4 and P-glycoprotein (P-gp) (nicardipine, quinidine, tacrolimus, and verapamil) were used as model compounds. Observed fraction of drug absorbed (F(a)F(g)) values for these compounds were calculated from in vivo pharmacokinetic (PK) parameters, whereas in vitro intestinal intrinsic clearance (CL(int,intestine)) was determined using human intestinal microsomes. The CL(int,intestine) for the model compounds corrected with that of midazolam was defined as CL(m,index) and incorporated into a simplified F(g) model with empirical scaling factor. Regardless of whether the compound was a P-gp substrate, the F(a)F(g) could be reasonably fitted by the simplified F(g) model, and the value of the empirical scaling factor was well estimated. These results suggest that the effects of P-gp on F(a) and F(g) are substantially minor, at least in the case of highly permeable compounds. Furthermore, liver intrinsic clearance (CL(int,liver)) can be used as a surrogate index of intestinal metabolism based on the relationship between CL(int,liver) and CL(m,index). F(g) can be easily predicted using a simplified F(g) model with the empirical scaling factor, enabling more confident selection of drug candidates with desirable PK profiles in humans.

A comparison of pharmacokinetics between humans and monkeys.

To verify the availability of pharmacokinetic parameters in cynomolgus monkeys, hepatic availability (Fh) and the fraction absorbed multiplied by intestinal availability (FaFg) were evaluated to determine their contributions to absolute bioavailability (F) after intravenous and oral administrations. These results were compared with those for humans using 13 commercial drugs for which human pharmacokinetic parameters have been reported. In addition, in vitro studies of these drugs, including membrane permeability, intrinsic clearance, and p-glycoprotein affinity, were performed to classify the drugs on the basis of their pharmacokinetic properties. In the present study, monkeys had a markedly lower F than humans for 8 of 13 drugs. Although there were no obvious differences in Fh between humans and monkeys, a remarkable species difference in FaFg was observed. Subsequently, we compared the FaFg values for monkeys with the in vitro pharmacokinetic properties of each drug. No obvious FaFg differences were observed between humans and monkeys for drugs that undergo almost no in vivo metabolism. In contrast, low FaFg were observed in monkeys for drugs that undergo relatively high metabolism in monkeys. These results suggest that first-pass intestinal metabolism is greater in cynomolgus monkeys than in humans, and that bioavailability in cynomolgus monkeys after oral administration is unsuitable for predicting pharmacokinetics in humans. In addition, a rough correlation was also observed between in vitro metabolic stability and Fg in humans, possibly indicating the potential for Fg prediction in humans using only in vitro parameters after slight modification of the evaluation system for in vitro intestinal metabolism.