Human Dose and Pharmacokinetic Predictions for Biologics at Boehringer Ingelheim: A Retrospective Analysis.

INTRODUCTION: Accurate predictions of pharmacokinetics and efficacious doses for biologics in humans are critical for selecting appropriate first-in-human starting doses and dose ranges and for estimating clinical material needs and cost of goods. This also impacts clinical feasibility, particularly for subcutaneously administered biologics. METHODS: We performed a comprehensive comparison between predicted and observed clearances and doses in humans for a set of 22 biologic drugs developed at Boehringer Ingelheim (BI) over the last 2 decades. The analysis included biologics across three therapeutic areas comprising a wide variety of modalities: mono- and bispecific monoclonal antibodies (mAbs) and nanobodies and a Fab fragment. RESULTS: Our analysis showed that observed clearances in humans were within twofold of predicted clearances for 17 out of 20 biologics (85%). Six biologics had uncharacteristically high observed human clearances (range 32-280 mL/h) for their respective molecular classes, impacting their clinical developability. For three molecules, molecular characteristics contributed to the high clearance. Clinically selected doses were within twofold of predicted for 58% of projects. With 42% and 25% of projects selecting clinical doses higher than two- or threefold the predicted value, respectively, the importance of better understanding not only the pharmacokinetic (PK) but also the predictivity of pharmacodynamic models is highlighted. CONCLUSIONS: We provide a clinical pharmacology perspective on the commonly accepted twofold range of human clearance predictions as well as the implications of higher than predicted targeted efficacious plasma concentration on clinical development. Finally, an analysis of key success factors for biologics at BI was conducted, which may be relevant for the entire pharmaceutical industry. This is one of the largest retrospective analyses for biologics and provides further evidence that successful predictions of human PK and efficacious dose will be further facilitated by gathering key translational data early in research.

Systemically Administered Anti-uPAR Antibody Plasma and Lung ELF Pharmacokinetics Characterized by Minimal Lung PBPK Model.

Antibody-based therapeutics have recently gained keen attention for the treatment of pulmonary indications. However, systemically administered antibody exposure in the lungs needs to be better understood and remains a topic of interest. In this study, we evaluated the exposure of two different uPAR (urokinase-type plasminogen activator receptor) targeting full-length monoclonal IgGs in plasma and lung epithelial lining fluid (ELF) of mice after IP and IV administration. Antibody AK17 exhibited linear pharmacokinetics (PK) in plasma and ELF at 3 and 30 mg/kg single IV dose. The average plasma and ELF half-lives for AK17 and AK21 ranged between ~321-411 h and ~230-345 h, respectively, indicating sustained systemic and lung exposure of antibodies. The average ELF to the plasma concentration ratio of antibodies was ~0.01 and ~0.03 with IP and IV dosing, respectively, over 2 weeks post single dose. We simultaneously characterized plasma and ELF PK of antibody in mice by developing a minimal lung PBPK model for antibody. This model reasonably captured the plasma and ELF PK data while estimating three parameters. The model accounts for the convective transport of antibody into the tissues via blood and lymph flow. FcRn-mediated transcytosis was incorporated into the model for antibody distribution across the lung epithelial barrier. This model serves as a platform to predict the pulmonary PK of systemically administered antibodies and to support optimal dose selection for desired exposure in the lungs as the site of action.

Using quantitative systems pharmacology modeling to optimize combination therapy of anti-PD-L1 checkpoint inhibitor and T cell engager.

Although immune checkpoint blockade therapies have shown evidence of clinical effectiveness in many types of cancer, the outcome of clinical trials shows that very few patients with colorectal cancer benefit from treatments with checkpoint inhibitors. Bispecific T cell engagers (TCEs) are gaining popularity because they can improve patients' immunological responses by promoting T cell activation. The possibility of combining TCEs with checkpoint inhibitors to increase tumor response and patient survival has been highlighted by preclinical and clinical outcomes. However, identifying predictive biomarkers and optimal dose regimens for individual patients to benefit from combination therapy remains one of the main challenges. In this article, we describe a modular quantitative systems pharmacology (QSP) platform for immuno-oncology that includes specific processes of immune-cancer cell interactions and was created based on published data on colorectal cancer. We generated a virtual patient cohort with the model to conduct in silico virtual clinical trials for combination therapy of a PD-L1 checkpoint inhibitor (atezolizumab) and a bispecific T cell engager (cibisatamab). Using the model calibrated against the clinical trials, we conducted several virtual clinical trials to compare various doses and schedules of administration for two drugs with the goal of therapy optimization. Moreover, we quantified the score of drug synergy for these two drugs to further study the role of the combination therapy.

Trends in industrialization of biotherapeutics: a survey of product characteristics of 89 antibody-based biotherapeutics.

There is considerable interest in the pharmaceutical industry toward development of antibody-based biotherapeutics because they can selectively bind diverse receptors and often possess desirable pharmacology. Here, we studied product characteristics of 89 marketed antibody-based biotherapeutics that were approved from 1986 to mid-2020 by gathering publicly available information. Our analyses revealed major trends in their emergence as the best-selling class of pharmaceuticals. Early on, most therapeutic monoclonal antibodies were developed to treat cancer, with CD20 being the most common target. Thanks to industrialization of antibody manufacturing technologies, their use has now blossomed to include 15 different therapeutic areas and nearly 60 targets, and the field is still growing! Drug manufacturers are solidifying their choices regarding types of antibodies and their molecular formats. IgG1 kappa continues to be the most common molecular format among marketed antibody-based biotherapeutics. Most antibody-based biotherapeutics approved since 2015 are either humanized or fully human, but the data we collected do not show a direct correlation between humanness and reported incidence of anti-drug antibodies. Furthermore, there have also been improvements in terms of drug product stability and high concentration liquid formulations suitable for subcutaneous route of administration, which are being approved more often in recent years. These improvements, however, have not been uniformly adopted across all therapeutic areas, suggesting that multiple options for drug product development are being used to serve diverse therapeutic purposes. Insights gained from this analysis may help us devise better end-to-end antibody-based biotherapeutic drug discovery and development strategies.

Development of novel ultra-sensitive IL-11 target engagement assays to support mechanistic PK/PD modeling for an anti-IL-11 antibody therapeutic.

An in-house antibody generation campaign identified a diverse, high affinity set of anti-interleukin-11 (IL-11) monoclonal antibodies (mAbs) to enable successful development of novel, custom ultra-sensitive target engagement assays for detection of "free" (unbound to the dosed anti-IL-11 therapeutic mAb) and "total" (free and mAb-IL-11 complexed form) IL-11 in preclinical species and human. Antibody hits from distinct epitope communities were screened on various platforms, including enzyme-linked immunosorbent assay, Meso Scale Discovery, Simoa HD-1 and Simoa Planar Array (SP-X), and used for assay development and sensitivity optimization. The ultra-sensitive SP-X format achieved a lower limit of quantitation of 0.006 pg/mL, enabling the first reported baseline levels of IL-11 in healthy control plasma determined by custom bioanalytical assays. These newly established baseline levels supported mechanistic pharmacokinetic/pharmacodynamic modeling in mouse, cynomolgus monkey, and human for a greater understanding of preclinical study design and in vivo dynamic interaction of soluble IL-11 with an anti-IL-11 antibody therapeutic candidate. Modeling and simulation also helped refine the utility of assays with respect to their potential use as target engagement biomarkers in the clinic.Abbreviations IL-11: Interleukin-11, TE: Target engagement, PK/PD: Pharmacokinetic/pharmacodynamic, mAb: Monoclonal antibody, NHP: Non-human primate, IgG: Immunoglobulin G, Cyno: Cynomolgulus monkey, GFR: Glomerular filtration rate, BQL: Below quantitation levels, DRM: Disease relevant model, kDa: kilodaltons, SPR: Surface plasmon resonance, pSTAT3: phosphorylated STAT3, IL-11R: Interleukin-11 receptor, TPP: Target product protein, LLOQ: Lower limit of quantitation, RLU: Relative light units.

Analyze impact of tumor-associated kinetics on antibody delivery in solid tumors with a physiologically based pharmacokinetics/pharmacodynamics model.

Monoclonal antibody (mAb)-based drugs are critical anti-cancer therapies. Unfortunately, therapeutic efficacy can be compromised by spatially heterogeneous intratumoral Ab deposition. Binding-site barriers arising from Ab and tumor-associated kinetics often underlie this phenomenon. Quantitative insight into these issues may lead to more efficient drug delivery. Difficulties in addressing this issue include (1) lack of techniques to quantify critical kinetic events, (2) lack of a pharmacokinetic/pharmacodynamic (PK/PD) model to assess important parameters for specific tumor types, and (3) uncertainty or variability of critical kinetic factors even within a single tumor type. This study developed a mechanism-based PK/PD model to profile heterogeneous distribution of Ab within tumors and tested this model using real-life experimental data. Model simulations incorporating several uncertainties were used to determine how mAb and tumor-associated kinetics influence receptor occupancy. Simulations were also used to predict the potential impact of these findings in preclinical tumor models and human tumors. We found significant differences in tumor-associated kinetics between groups in which mAb therapy was effective versus groups in which it was ineffective. These kinetic differences included rates of tumor-associated antigen (TAA) degradation, TAA expression, apparent flow rates of interstitial fluid, and ratios of Ab-TAA complex internalization to TAA degradation. We found less significant differences in mAb kinetics, including rates of clearance or affinity for target antigens. In conclusion, our mechanism-based PK/PD model suggests that TAA-associated kinetic factors participate more significantly than those associated with the Ab in generating barriers to mAb delivery and distribution in tumors.

Retrospective analysis of model-based predictivity of human pharmacokinetics for anti-IL-36R monoclonal antibody MAB92 using a rat anti-mouse IL-36R monoclonal antibody and RNA expression data (FANTOM5).

Accurate prediction of the human pharmacokinetics (PK) of a candidate monoclonal antibody from nonclinical data is critical to maximize the success of clinical trials. However, for monoclonal antibodies exhibiting nonlinear clearance due to target-mediated drug disposition, PK predictions are particularly challenging. That challenge is further compounded for molecules lacking cross-reactivity in a nonhuman primate, in which case a surrogate antibody selective for the target in rodent may be required. For these cases, prediction of human PK must account for any interspecies differences in binding kinetics, target expression, target turnover, and potentially epitope. We present here a model-based method for predicting the human PK of MAB92 (also known as BI 655130), a humanized IgG1 κ monoclonal antibody directed against human IL-36R. Preclinical PK was generated in the mouse with a chimeric rat anti-mouse IgG2a surrogate antibody cross-reactive against mouse IL-36R. Target-specific parameters such as antibody binding affinity (KD), internalization rate of the drug target complex (kint), target degradation rate (kdeg), and target abundance (R0) were integrated into the model. Two different methods of assigning human R0 were evaluated: the first assumed comparable expression between human and mouse and the second used high-resolution mRNA transcriptome data (FANTOM5) as a surrogate for expression. Utilizing the mouse R0 to predict human PK, AUC0-∞ was substantially underpredicted for nonsaturating doses; however, after correcting for differences in RNA transcriptome between species, AUC0-∞ was predicted largely within 1.5-fold of observations in first-in-human studies, demonstrating the validity of the modeling approach. Our results suggest that semi-mechanistic models incorporating RNA transcriptome data and target-specific parameters may improve the predictivity of first-in-human PK.

Quantitative Conjugated Payload Measurement Using Enzymatic Release of Antibody-Drug Conjugate with Cleavable Linker.

As antibody-drug conjugate (ADC) design is evolving with novel payload, linker, and conjugation chemistry, the need for sensitive and precise quantitative measurement of conjugated payload to support pharmacokinetics (PK) is in high demand. Compared to ADCs containing noncleavable linkers, a strategy specific to linkers which are liable to pH, chemical reduction, or enzymatic cleavage has gained popularity in recent years. One bioanalytical approach to take advantage of this type of linker design is the development of a PK assay measuring released conjugated payload. For the ADC utilizing a dipeptide ValCit linker studied in this report, the release of payload PF-06380101 was achieved with high efficiency using a purified cathepsin B enzyme. The subsequent liquid chromatography mass spectrometry (LC/MS) quantitation leads to the PK profile of the conjugated payload. For this particular linker using a maleimide-based conjugation chemistry, one potential route of payload loss would result in an albumin adduct of the linker-payload. While this adduct's formation has been previously reported, here, for the first time, we have shown that payload from a source other than ADC contributes only up to 4% of total conjugated payload while it accounts for approximately 35% of payload lost from the ADC at 48 h after dosing to rats.

Where Did the Linker-Payload Go? A Quantitative Investigation on the Destination of the Released Linker-Payload from an Antibody-Drug Conjugate with a Maleimide Linker in Plasma.

The reactive thiol of cysteine is often used for coupling maleimide-containing linker-payloads to antibodies resulting in the generation of antibody drug conjugates (ADCs). Currently, a numbers of ADCs in drug development are made by coupling a linker-payload to native or engineered cysteine residues on the antibody. An ADC conjugated via hinge-cysteines to an auristatin payload was used as a model in this study to understand the impact of the maleimide linkers on ADC stability. The payload was conjugated to trastuzumab by a protease-cleavable linker, maleimido-caproyl-valine-citruline-p-amino-benzyloxy carbonyl (mcVC-PABC). In plasma stability assays, when the ADC (Trastuzumab-mcVC-PABC-Auristatin-0101) was incubated with plasma over a 144-h time-course, a discrepancy was observed between the measured released free payload concentration and the measured loss of drug-to-antibody ratio (DAR), as measured by liquid chromatography-mass spectrometry (LC-MS). We found that an enzymatic release of payload from ADC-depleted human plasma at 144 h was able to account for almost 100% of the DAR loss. Intact protein mass analysis showed that at the 144 h time point, the mass of the major protein in ADC-depleted human plasma had an additional 1347 Da over the native albumin extracted from human plasma, exactly matching the mass of the linker-payload. In addition, protein gel electrophoresis showed that there was only one enriched protein in the 144 h ADC-depleted and antipayload immunoprecipitated plasma sample, as compared to the 0 h plasma immunoprecipitated sample, and the mass of this enriched protein was slightly heavier than the mass of serum albumin. Furthermore, the albumin adduct was also identified in 96 h and 168 h postdose in vivo cynomolgus monkey plasma. These results strongly suggest that the majority of the deconjugated mc-VC-PABC-auristatin ultimately is transferred to serum albumin, forming a long-lived albumin-linker-payload adduct. To our knowledge, this is the first report quantitatively characterizing the extent of linker-payload transfer to serum albumin and the first clear example of in vivo formation of an albumin-linker-payload adduct.

Inhibitors of HIV-1 attachment. Part 9: an assessment of oral prodrug approaches to improve the plasma exposure of a tetrazole-containing derivative.

7-(2H-Tetrazol-5-yl)-1H-indole 3 was found to be a potent inhibitor of HIV-1 attachment but the compound lacked oral bioavailability in rats. The cause of the low exposure was believed to be poor absorption attributed to the acidic nature of the tetrazole moiety and, in an effort to address this liability, three more lipohilic tetrazole analogs, N-acetoxymethyl 4, N-pivaloyloxymethyl 5, and N-methyl 6, were evaluated as potential oral prodrugs in rats. Prodrug 5 was ineffective in improving the plasma concentration of 3 in vivo but compound 4 provided a 15-fold enhancement of the plasma concentration of 3. Most interestingly, oral dosing of analog 6 afforded a substantial increase in the plasma concentration of the parent in rats when compared to dosing of parent. This represents a novel example of a methyl tetrazole that acts as a prodrug for a free NH tetrazole-containing compound.

Inhibitors of HIV-1 attachment. Part 8: the effect of C7-heteroaryl substitution on the potency, and in vitro and in vivo profiles of indole-based inhibitors.

As part of the SAR profiling of the indole-oxoacetic piperazinyl benzamide class of HIV-1 attachment inhibitors, substitution at the C7 position of the lead 4-fluoroindole 2 with various 5- and 6-membered heteroaryl moieties was explored. Highly potent (picomolar) inhibitors of pseudotyped HIV-1 in a primary, cell-based assay were identified and select examples were shown to possess nanomolar inhibitory activity against M- and T-tropic viruses in cell culture. These C7-heteroaryl-indole analogs maintained the ligand efficiency (LE) of 2 and were also lipophilic efficient as measured by LLE and LELP. Pharmacokinetic studies of this class of inhibitor in rats showed that several possessed substantially improved IV clearance and half-lives compared to 2. Oral exposure in the rat correlated with membrane permeability as measured in a Caco-2 assay where the highly permeable 1,2,4-oxadiazole analog 13 exhibited the highest exposure.

Inhibitors of HIV-1 attachment. Part 7: indole-7-carboxamides as potent and orally bioavailable antiviral agents.

A series of substituted carboxamides at the indole C7 position of the previously described 4-fluoro-substituted indole HIV-1 attachment inhibitor 1 was synthesized and the SAR delineated. Heteroaryl carboxamide inhibitors that exhibited pM potency in the primary cell-based assay against a pseudotype virus expressing a JRFL envelope were identified. The simple methyl amide analog 4 displayed a promising in vitro profile, with its favorable HLM stability and membrane permeability translating into favorable pharmacokinetic properties in preclinical species.

Immune-mediated agranulocytosis caused by the cocaine adulterant levamisole: a case for reactive metabolite(s) involvement.

The United States Public Health Service Administration is alerting medical professionals that a substantial percentage of cocaine imported into the United States is adulterated with levamisole, a veterinary pharmaceutical that can cause blood cell disorders such as severe neutropenia and agranulocytosis. Levamisole was previously approved in combination with fluorouracil for the treatment of colon cancer; however, the drug was withdrawn from the U.S. market in 2000 because of the frequent occurrence of agranulocytosis. The detection of autoantibodies such as antithrombin (lupus anticoagulant) and an increased risk of agranulocytosis in patients carrying the human leukocyte antigen B27 genotype suggest that toxicity is immune-mediated. In this perspective, we provide an historical account of the levamisole/cocaine story as it first surfaced in 2008, including a succinct review of levamisole pharmacology, pharmacokinetics, and preclinical/clinical evidence for levamisole-induced agranulocytosis. Based on the available information on levamisole metabolism in humans, we propose that reactive metabolite formation is the rate-limiting step in the etiology of agranulocytosis associated with levamisole, in a manner similar to other drugs (e.g., propylthiouracil, methimazole, captopril, etc.) associated with blood dyscrasias. Finally, considering the toxicity associated with levamisole, we propose that the 2,3,5,6-tetrahydroimidazo[2,1-b]thiazole scaffold found in levamisole be categorized as a new structural alert, which is to be avoided in drug design.

LC-MS/MS method development and validation for the determination of polymyxins and vancomycin in rat plasma.

Simple, sensitive and robust liquid chromatography-tandem mass spectrometer (LC-MS/MS) methods were developed and validated for the determination of lipopeptide polymyxins and glycopeptide vancomycin in rat plasma. The effect of trichloroacetic acid (TCA) concentration on sample recoveries (peak area of sample recovered from plasma/peak area of sample from neat solvent solutions) was studied and an optimized concentration of 30% TCA were determined that gives the best sample recovery for the peptides from rat plasma. The effect of the TCA concentration on the chromatographic behavior of peptides was studied on a Phenomenex Jupiter C18 5µ 300Å 50mm×2mm column using a mobile phase with a pH of 2.8. Other than protein precipitation, TCA also acted as ion pairing reagent and was only present in the samples but not in the mobile phases. The data demonstrated that by increasing the TCA concentration, the analyte retention and sensitivity were improved. The absence of TCA in mobile phase helped to reduce the ion source contamination and to achieve good reproducibility. The plasma method was linearly calibrated from 5 to 5000ng/mL for polymyxins with precisions to be of 2.3-10.8%, and accuracies to be 91.7-107.4% for polymyxin B1, B2, E1, E2, respectively. For vancomycin the calibration is from 1 to 5000ng/mL with precisions to be of 7.8-10.3 and accuracies to be 96.2-102.0%. The LLOQs corresponding with a coefficient of variation less than 20% were 7.5, 18.1, 7.3, 5.0 and 1.0ng/mL for polymyxin B1, B2, E1, E2 and vancomycin, respectively.

Utilization of in vitro Caco-2 permeability and liver microsomal half-life screens in discovering BMS-488043, a novel HIV-1 attachment inhibitor with improved pharmacokinetic properties.

Optimizing pharmacokinetic properties to improve oral exposure is a common theme in modern drug discovery. In the present work, in vitro Caco-2 permeability and microsomal half-life screens were utilized in an effort to guide the structure-activity relationship in order to improve the pharmacokinetic properties of novel HIV-1 attachment inhibitors. The relevance of the in vitro screens to in vivo pharmacokinetic properties was first demonstrated with a number of program compounds at the early stage of lead optimization. The Caco-2 permeability, tested at 200 microM, was quantitatively predictive of in vivo oral absorption, with complete absorption occurring at a Caco-2 permeability of 100 nm/s or higher. The liver microsomal half-life screen, conducted at 1 microM substrate concentration, can readily differentiate low-, intermediate-, and high-clearance compounds in rats, with a nearly 1:1 correlation in 12 out of 13 program compounds tested. Among the >100 compounds evaluated, BMS-488043 emerged as a lead, exhibiting a Caco-2 permeability of 178 nm/s and a microsomal half-life predictive of a low clearance (4 mL/min/kg) in humans. These in vitro characteristics translated well to the in vivo setting. The oral bioavailability of BMS-488043 in rats, dogs, and monkeys was 90%, 57%, and 60%, respectively. The clearance was low in all three species tested, with a terminal half-life ranging from 2.4 to 4.7 h. Furthermore, the oral exposure of BMS-488043 was significantly improved (6- to 12-fold in rats and monkeys) compared to the prototype compound BMS-378806 that had a suboptimal Caco-2 permeability (51 nm/s) and microsomal half-life. More importantly, the improvements in preclinical pharmacokinetics translated well to humans, leading to a >15-fold increase in the human oral exposure of BMS-488043 than BMS-378806 and enabling a clinical proof-of-concept for this novel class of anti-HIV agents. The current studies demonstrated the valuable role of in vitro ADME screens in improving oral pharmacokinetics at the lead optimization stage.

Carbamate-appended N-alkylsulfonamides as inhibitors of gamma-secretase.

The synthesis and gamma-secretase inhibition data for a series of carbamate-appended N-alkylsulfonamides are described. Carbamate 54 was found to significantly reduce brain Abeta in transgenic mice. 54 was also found to possess markedly improved brain levels in transgenic mice compared to previously disclosed 1 and 2.

Nitrogen-appended N-alkylsulfonamides as inhibitors of gamma-secretase.

The synthesis and gamma-secretase inhibition data for a series of nitrogen-appended N-alkylsulfonamides (11-47) are described. Inhibition of brain Abeta in transgenic mice was demonstrated by two of these compounds (23 and 44).

N-(5-chloro-2-(hydroxymethyl)-N-alkyl-arylsulfonamides as gamma-secretase inhibitors.

A series of N-alkylbenzenesulfonamides were developed from a high throughput screening hit. Classic and parallel synthesis strategies were employed to produce compounds with good in vitro and in vivo gamma-secretase activity.

Novel 1H-(benzimidazol-2-yl)-1H-pyridin-2-one inhibitors of insulin-like growth factor I (IGF-1R) kinase.

A novel class of 1H-(benzimidazol-2-yl)-1H-pyridin-2-one inhibitors of insulin-like growth factor I (IGF-1R) kinase is described. This report discusses the SAR of 4-(2-hydroxy-2-phenylethylamino)-substituted pyridones with improved IGF-1R potency.