Trastuzumab-MMAU Antibody-Auristatin Conjugates: Valine-Glucoserine Linker with Stabilized Maleimide Conjugation Improves In Vivo Efficacy and Tolerability.

PURPOSE: Antibody-drug conjugates (ADCs) have shown impressive clinical activity with approval of many agents in hematological and solid tumors. However, challenges remain with both efficacy and safety of ADCs. This study describes novel trastuzumab-auristatin conjugates with the hydrophilic MMAE prodrug MMAU, and optimization of a glycopeptide linker leading to a wider therapeutic window. EXPERIMENTAL DESIGN: Trastuzumab was conjugated with auristatin payloads via a series of linkers using a stabilized maleimide handle. The ADCs were characterized in vitro and their relative in vivo anti-tumor efficacies were assessed in HER2+ xenograft models. Relative linker stabilities and the mechanism of linker cleavage were studied using in vitro assays. Toxicity and toxicokinetics of the best performing ADC were evaluated in cynomolgus monkey (cyno). RESULTS: The trastuzumab-MMAU ADC with stabilized glycopeptide linker showed maleimide stabilization and higher resistance to cleavage by serum and lysosomal enzymes compared to a valine-citrulline conjugated trastuzumab ADC (trastuzumab-vc-MMAE). A single dose of 1 or 2 mg/kg of trastuzumab-MMAU at drug-to-antibody ratios (DAR) of 8 and 4 respectively resulted in xenograft tumor growth inhibition, with superior efficacy to trastuzumab-vc-MMAE. Trastuzumab-MMAU DAR4 was tolerated at doses up to 12 mg/kg in cyno, which represents 2- to 4-fold higher dose than that observed with vedotin ADCs, and had increased terminal half-life and exposure. CONCLUSIONS: The optimized trastuzumab-MMAU ADC showed potent antitumor activity and was well tolerated with excellent pharmacokinetics in non-human primates, leading to a superior preclinical therapeutic window. The data supports potential utility of trastuzumab-MMAU for treatment of HER2+ tumors.

Considerations from an International Regulatory and Pharmaceutical Industry (IQ MPS Affiliate) Workshop on the Standardization of Complex In Vitro Models in Drug Development.

In May 2022, there is an International Regulatory and Pharmaceutical Industry (Innovation and Quality [IQ] Microphysiological Systems [MPS] Affiliate) Workshop on the standardization of complex in vitro models (CIVMs) in drug development. This manuscript summarizes the discussions and conclusions of this joint workshop organized and executed by the IQ MPS Affiliate and the United States Food and Drug Administration (FDA). A key objective of the workshop is to facilitate discussions around opportunities and/or needs for standardization of MPS and chart potential pathways to increase model utilization in the context of regulatory decision making. Participation in the workshop included 200 attendees from the FDA, IQ MPS Affiliate, and 26 global regulatory organizations and affiliated parties representing Europe, Japan, and Canada. It is agreed that understanding global perspectives regarding the readiness of CIVM/MPS models for regulatory decision making and potential pathways to gaining acceptance is useful to align on globally. The obstacles are currently too great to develop standards for every context of use (COU). Instead, it is suggested that a more tractable approach may be to think of broadly applicable standards that can be applied regardless of COU and/or organ system. Considerations and next steps for this effort are described.

AI/ML Models to Predict the Severity of Drug-Induced Liver Injury for Small Molecules.

Drug-induced liver injury (DILI), believed to be a multifactorial toxicity, has been a leading cause of attrition of small molecules during discovery, clinical development, and postmarketing. Identification of DILI risk early reduces the costs and cycle times associated with drug development. In recent years, several groups have reported predictive models that use physicochemical properties or in vitro and in vivo assay endpoints; however, these approaches have not accounted for liver-expressed proteins and drug molecules. To address this gap, we have developed an integrated artificial intelligence/machine learning (AI/ML) model to predict DILI severity for small molecules using a combination of physicochemical properties and off-target interactions predicted in silico. We compiled a data set of 603 diverse compounds from public databases. Among them, 164 were categorized as Most DILI (M-DILI), 245 as Less DILI (L-DILI), and 194 as No DILI (N-DILI) by the FDA. Six machine learning methods were used to create a consensus model for predicting the DILI potential. These methods include k-nearest neighbor (k-NN), support vector machine (SVM), random forest (RF), Naïve Bayes (NB), artificial neural network (ANN), logistic regression (LR), weighted average ensemble learning (WA) and penalized logistic regression (PLR). Among the analyzed ML methods, SVM, RF, LR, WA, and PLR identified M-DILI and N-DILI compounds, achieving a receiver operating characteristic area under the curve of 0.88, sensitivity of 0.73, and specificity of 0.9. Approximately 43 off-targets, along with physicochemical properties (fsp3, log S, basicity, reactive functional groups, and predicted metabolites), were identified as significant factors in distinguishing between M-DILI and N-DILI compounds. The key off-targets that we identified include: PTGS1, PTGS2, SLC22A12, PPARγ, RXRA, CYP2C9, AKR1C3, MGLL, RET, AR, and ABCC4. The present AI/ML computational approach therefore demonstrates that the integration of physicochemical properties and predicted on- and off-target biological interactions can significantly improve DILI predictivity compared to chemical properties alone.

Applications of microphysiological systems to disease models in the biopharmaceutical industry: Opportunities and challenges.

Disease models enable researchers to investigate, test, and identify therapeutic targets that would alter the patients' disease condition and improve quality of life. Advances in genetic alteration and analytical techniques have enabled rapid devel­opment of disease models using preclinical animals and cell cultures. However, success rates of drug development remain low due to limited recapitulation of clinical pathophysiology by these models. To resolve this challenge, the pharmaceutical industry has explored microphysiological system (MPS) disease models, which are complex in vitro systems that include but are not limited to organ-on-a-chip, organoids, spheroids, and 3D bioengineered tissues (e.g., 3D printing, hydrogels). Capable of integrating key in vivo properties, such as disease-relevant human cells, multi-cellularity/dimensionality of organs, and/or well-controlled physical and molecular cues, MPS disease models are being developed for a variety of indications. With on-going qualifications or validations for wide adoption within the pharmaceutical industry, MPS disease models hold exciting potential to enable in-depth investigation of in vivo pathophysiology and enhance drug discovery and development processes. To introduce the present status of MPS disease models, this paper describes notable examples in six disease areas: cancer, liver/kidney diseases, respiratory diseases/COVID-19, neurodegenerative diseases, gastrointestinal diseases, and select rare diseases. Additionally, we describe current technical limitations and provide recommendations for future development that would expand application opportunities within the pharmaceutical industry.

Performance assessment and economic analysis of a human Liver-Chip for predictive toxicology.

BACKGROUND: Conventional preclinical models often miss drug toxicities, meaning the harm these drugs pose to humans is only realized in clinical trials or when they make it to market. This has caused the pharmaceutical industry to waste considerable time and resources developing drugs destined to fail. Organ-on-a-Chip technology has the potential improve success in drug development pipelines, as it can recapitulate organ-level pathophysiology and clinical responses; however, systematic and quantitative evaluations of Organ-Chips' predictive value have not yet been reported. METHODS: 870 Liver-Chips were analyzed to determine their ability to predict drug-induced liver injury caused by small molecules identified as benchmarks by the Innovation and Quality consortium, who has published guidelines defining criteria for qualifying preclinical models. An economic analysis was also performed to measure the value Liver-Chips could offer if they were broadly adopted in supporting toxicity-related decisions as part of preclinical development workflows. RESULTS: Here, we show that the Liver-Chip met the qualification guidelines across a blinded set of 27 known hepatotoxic and non-toxic drugs with a sensitivity of 87% and a specificity of 100%. We also show that this level of performance could generate over $3 billion annually for the pharmaceutical industry through increased small-molecule R&D productivity. CONCLUSIONS: The results of this study show how incorporating predictive Organ-Chips into drug development workflows could substantially improve drug discovery and development, allowing manufacturers to bring safer, more effective medicines to market in less time and at lower costs.

Drug development is lengthy and costly, as it relies on laboratory models that fail to predict human reactions to potential drugs. Because of this, toxic drugs sometimes go on to harm humans when they reach clinical trials or once they are in the marketplace. Organ-on-a-Chip technology involves growing cells on small devices to mimic organs of the body, such as the liver. Organ-Chips could potentially help identify toxicities earlier, but there is limited research into how well they predict these effects compared to conventional models. In this study, we analyzed 870 Liver-Chips to determine how well they predict drug-induced liver injury, a common cause of drug failure, and found that Liver-Chips outperformed conventional models. These results suggest that widespread acceptance of Organ-Chips could decrease drug attrition, help minimize harm to patients, and generate billions in revenue for the pharmaceutical industry.

Liver safety evaluation of endothelin receptor antagonists using HepatoPac® : A single model impact assessment on hepatocellular health, function and bile acid disposition.

Marketed (bosentan, ambrisentan) and discontinued (sitaxsentan, CI-1034) endothelin receptor antagonists were examined in the human micropatterned hepatocyte co-culture (MPCC) model HepatoPac® . Differences across hepatocellular health (cellular adenosine triphosphate/glutathione content), function (urea production/albumin secretion) and taurocholic acid transport (biliary clearance/excretion index) were compared using amiodarone and ciclosporin A as positive controls. Ambrisentan had the weakest potency in all six endpoints, while sitaxsentan, bosentan and CI-1034 had more potent effects on hepatobiliary transport than health/function endpoints. Normalization to clinical Cmax gave the following relative rank order of safety based on margins for each endpoint: ambrisentan ≥ CI-1034 ~ bosentan > sitaxsentan. These data suggested impaired hepatobiliary disposition might contribute to a more prominent role in liver injury associated within sensitive human populations exposed to these compounds than direct hepatocellular toxicity. Rat, dog and monkey MPCCs also showed greater sensitivity potential to disrupted hepatobiliary disposition compared with hepatocellular health/functional endpoints. Drug metabolism competency was exhibited across all species. In vivo, rats and dogs appear more resistant to transaminase elevations and/or histological evidence of liver injury caused by these mechanisms even at exceedingly high systemic exposures relative to sensitive humans. Rats and dogs are resistant to hepatobiliary toxicants due to physiological differences in bile composition/handling. Although traditional animal testing provides adequate safety coverage for advancement of novel pharmaceuticals into clinical trials, supplemental assays employing human MPCCs may strengthen weight-of-evidence predictions for sensitive human populations. Proving the predictive value of this single impact assessment model in advance of clinical trial information for human liver injury risk is needed across more pharmaceuticals.

Establishment of a hepatocyte-kupffer cell coculture model for assessment of proinflammatory cytokine effects on metabolizing enzymes and drug transporters.

Elevated levels of proinflammatory cytokines associated with infection and inflammation can modulate cytochrome P450 enzymes, leading to potential disease-drug interactions and altered small-molecule drug disposition. We established a human-derived hepatocyte-Kupffer cell (Hep:KC) coculture model to assess the indirect cytokine impact on hepatocytes through stimulation of KC-mediated cytokine release and compared this model with hepatocytes alone. Characterization of Hep:KC cocultures showed an inflammation response after treatment with lipopolysaccharide and interleukin (IL)-6 (indicated by secretion of various cytokines). Additionally, IL-6 exposure upregulated acute-phase proteins (C-reactive protein, alpha-1-acid glycoprotein, and serum amyloid A2) and downregulated CYP3A4. Compared with hepatocytes alone, Hep:KC cocultures showed enhanced IL-1ß-mediated effects but less impact from both IL-2 and IL-23. Hep:KC cocultures treated with IL-1ß exhibited a higher release of proinflammatory cytokines, an increased upregulation of acute-phase proteins, and a larger extent of metabolic enzyme and transporter suppression. IC50 values for IL-1ß-mediated CYP3A4 suppression were lower in Hep:KC cocultures (98.0-144 pg/ml) compared with hepatocytes alone (IC50 > 5000 pg/ml). Cytochrome suppression was preventable by blocking IL-1ß interaction with IL-1R1 using an antagonist cytokine or an anti-IL-1ß antibody. Unlike IL-1ß, IL-6-mediated effects were comparable between hepatocyte monocultures and Hep:KC cocultures. IL-2 and IL-23 caused a negligible inflammation response and a minimal inhibition of CYP3A4. In both hepatocyte monocultures and Hep:KC cocultures, IL-2RB and IL-23R were undetectable, whereas IL-6R and IL-1R1 levels were higher in Hep:KC cocultures. In summary, compared with hepatocyte monocultures, the Hep:KC coculture system is a more robust in vitro model for studying the impact of proinflammatory cytokines on metabolic enzymes.

Prediction of cartilage compressive modulus using multiexponential analysis of T(2) relaxation data and support vector regression.

Evaluation of mechanical characteristics of cartilage by magnetic resonance imaging would provide a noninvasive measure of tissue quality both for tissue engineering and when monitoring clinical response to therapeutic interventions for cartilage degradation. We use results from multiexponential transverse relaxation analysis to predict equilibrium and dynamic stiffness of control and degraded bovine nasal cartilage, a biochemical model for articular cartilage. Sulfated glycosaminoglycan concentration/wet weight (ww) and equilibrium and dynamic stiffness decreased with degradation from 103.6 ± 37.0 µg/mg ww, 1.71 ± 1.10 MPa and 15.3 ± 6.7 MPa in controls to 8.25 ± 2.4 µg/mg ww, 0.015 ± 0.006 MPa and 0.89 ± 0.25MPa, respectively, in severely degraded explants. Magnetic resonance measurements were performed on cartilage explants at 4 °C in a 9.4 T wide-bore NMR spectrometer using a Carr-Purcell-Meiboom-Gill sequence. Multiexponential T2 analysis revealed four water compartments with T2 values of approximately 0.14, 3, 40 and 150 ms, with corresponding weight fractions of approximately 3, 2, 4 and 91%. Correlations between weight fractions and stiffness based on conventional univariate and multiple linear regressions exhibited a maximum r(2) of 0.65, while those based on support vector regression (SVR) had a maximum r(2) value of 0.90. These results indicate that (i) compartment weight fractions derived from multiexponential analysis reflect cartilage stiffness and (ii) SVR-based multivariate regression exhibits greatly improved accuracy in predicting mechanical properties as compared with conventional regression.

Improved MR-based characterization of engineered cartilage using multiexponential T2 relaxation and multivariate analysis.

Noninvasive monitoring of tissue quality would be of substantial use in the development of cartilage tissue engineering strategies. Conventional MR parameters provide noninvasive measures of biophysical tissue properties and are sensitive to changes in matrix development, but do not clearly distinguish between groups with different levels of matrix development. Furthermore, MR outcomes are nonspecific, with particular changes in matrix components resulting in changes in multiple MR parameters. To address these limitations, we present two new approaches for the evaluation of tissue engineered constructs using MR, and apply them to immature and mature engineered cartilage after 1 and 5 weeks of development, respectively. First, we applied multiexponential T(2) analysis for the quantification of matrix macromolecule-associated water compartments. Second, we applied multivariate support vector machine analysis using multiple MR parameters to improve detection of degree of matrix development. Monoexponential T(2) values decreased with maturation, but without further specificity. Much more specific information was provided by multiexponential analysis. The T(2) distribution in both immature and mature constructs was qualitatively comparable to that of native cartilage. The analysis showed that proteoglycan-bound water increased significantly during maturation, from a fraction of 0.05 ± 0.01 to 0.07 ± 0.01. Classification of samples based on individual MR parameters, T(1), T(2), k(m) or apparent diffusion coefficient, showed that the best classifiers were T(1) and k(m), with classification accuracies of 85% and 84%, respectively. Support vector machine analysis improved the accuracy to 98% using the combination (k(m), apparent diffusion coefficient). These approaches were validated using biochemical and Fourier transform infrared imaging spectroscopic analyses, which showed increased proteoglycan and collagen with maturation. In summary, multiexponential T(2) and multivariate support vector machine analyses provide improved sensitivity to changes in matrix development and specificity to matrix composition in tissue engineered cartilage. These approaches show substantial potential for the evaluation of engineered cartilage tissue and for extension to other tissue engineering constructs.

Multivariate analysis of cartilage degradation using the support vector machine algorithm.

An important limitation in MRI studies of early osteoarthritis is that measured MRI parameters exhibit substantial overlap between different degrees of cartilage degradation. We investigated whether multivariate support vector machine analysis would permit improved tissue characterization. Bovine nasal cartilage samples were subjected to pathomimetic degradation and their T(1), T(2), magnetization transfer rate (k(m) ), and apparent diffusion coefficient (ADC) were measured. Support vector machine analysis performed using certain parameter combinations exhibited particularly favorable classification properties. The areas under the receiver operating characteristic (ROC) curve for detection of extensive and mild degradation were 1.00 and 0.94, respectively, using the set (T(1), k(m), ADC), compared with 0.97 and 0.60 using T(1), the best univariate classifier. Furthermore, a degradation probability for each sample, derived from the support vector machine formalism using the parameter set (T(1), k(m), ADC), demonstrated much stronger correlations (r(2) = 0.79-0.88) with direct measurements of tissue biochemical components than did even the best-performing individual MRI parameter, T(1) (r(2) = 0.53-0.64). These results, combined with our previous investigation of Gaussian cluster-based tissue discrimination, indicate that the combinations (T(1), k(m)) and (T(1), k(m), ADC) may emerge as particularly useful for characterization of early cartilage degradation.

Characterization of engineered cartilage constructs using multiexponential T2 relaxation analysis and support vector regression.

Increased sensitivity in the characterization of cartilage matrix status by magnetic resonance (MR) imaging, through the identification of surrogate markers for tissue quality, would be of great use in the noninvasive evaluation of engineered cartilage. Recent advances in MR evaluation of cartilage include multiexponential and multiparametric analysis, which we now extend to engineered cartilage. We studied constructs which developed from chondrocytes seeded in collagen hydrogels. MR measurements of transverse relaxation times were performed on samples after 1, 2, 3, and 4 weeks of development. Corresponding biochemical measurements of sulfated glycosaminoglycan (sGAG) were also performed. sGAG per wet weight increased from 7.74±1.34 µg/mg in week 1 to 21.06±4.14 µg/mg in week 4. Using multiexponential T2 analysis, we detected at least three distinct water compartments, with T2 values and weight fractions of (45 ms, 3%), (200 ms, 4%), and (500 ms, 97%), respectively. These values are consistent with known properties of engineered cartilage and previous studies of native cartilage. Correlations between sGAG and MR measurements were examined using conventional univariate analysis with T2 data from monoexponential fits with individual multiexponential compartment fractions and sums of these fractions, through multiple linear regression based on linear combinations of fractions, and, finally, with multivariate analysis using the support vector regression (SVR) formalism. The phenomenological relationship between T2 from monoexponential fitting and sGAG exhibited a correlation coefficient of r²=0.56, comparable to the more physically motivated correlations between individual fractions or sums of fractions and sGAG; the correlation based on the sum of the two proteoglycan-associated fractions was r²=0.58. Correlations between measured sGAG and those calculated using standard linear regression were more modest, with r² in the range 0.43-0.54. However, correlations using SVR exhibited r² values in the range 0.68-0.93. These results indicate that the SVR-based multivariate approach was able to determine tissue sGAG with substantially higher accuracy than conventional monoexponential T2 measurements or conventional regression modeling based on water fractions. This combined technique, in which the results of multiexponential analysis are examined with multivariate statistical techniques, holds the potential to greatly improve the accuracy of cartilage matrix characterization in engineered constructs using noninvasive MR data.

Mapping proteoglycan-bound water in cartilage: Improved specificity of matrix assessment using multiexponential transverse relaxation analysis.

Association of MR parameters with cartilage matrix components remains an area of ongoing investigation. Multiexponential analysis of nonlocalized transverse relaxation data has previously been used to quantify water compartments associated with matrix macromolecules in cartilage. We extend this to mapping the proteoglycan (PG)-bound water fraction in cartilage, using mature and young bovine nasal cartilage model systems, toward the goal of matrix component-specific imaging. PG-bound water fraction from mature and young bovine nasal cartilage was 0.31 ± 0.04 and 0.22 ± 0.06, respectively, in agreement with biochemically derived PG content and PG-to-water weight ratios. Fourier transform infrared imaging spectroscopic-derived PG maps normalized by water content (IR-PG(ww) ) showed spatial correspondence with PG-bound water fraction maps. Extensive simulation analysis demonstrated that the accuracy and precision of our determination of PG-bound water fraction was within 2%, which is well-within the observed tissue differences. Our results demonstrate the feasibility of performing imaging-based multiexponential analysis of transverse relaxation data to map PG in cartilage.

Magnetic resonance studies of macromolecular content in engineered cartilage treated with pulsed low-intensity ultrasound.

Noninvasive monitoring of matrix development in tissue-engineered cartilage constructs would permit ongoing assessment with the ability to modify culture conditions during development to optimize tissue characteristics. In this study, chondrocytes seeded in a collagen hydrogel were exposed for 20 min/day to pulsed low-intensity ultrasound (PLIUS) at 30 mWcm(-2) and cultured for up to 5 weeks. Biochemical assays, histology, immunohistochemistry, Fourier transform infrared spectroscopy, and magnetic resonance imaging (MRI) were performed at weeks 3 and 5 after initiation of growth. The noninvasive MRI measurements were correlated with those from the invasive studies. In particular, MRI transverse relaxation time (T2) and magnetization transfer rate (k(m)) correlated with macromolecular content, which was increased by application of PLIUS. This indicates the sensitivity of MR techniques to PLIUS-induced changes in matrix development, and highlights the potential for noninvasive assessment of the efficacy of anabolic interventions for engineered tissue.

Nondestructive assessment of engineered cartilage constructs using near-infrared spectroscopy.

Noninvasive assessment of engineered cartilage properties would enable better control of the developing tissue towards the desired structural and compositional endpoints through optimization of the biochemical environment in real time. The objective of this study is to assess the matrix constituents of cartilage using near-infrared spectroscopy (NIRS), a technique that permits full-depth assessment of developing engineered tissue constructs. Mid-infrared (mid-IR) and NIR data were acquired from full-thickness cartilage constructs that were grown up to 4 weeks with and without mechanical stimulation. Correlations were assessed between established mid-IR peak areas that reflect the relative amount of collagen (amide I, amide II, and 1338 cm(-1)) and proteoglycan (PG), (850 cm(-1)), and the integrated area of the NIR water absorbance at 5190 cm(-1). This analysis was performed to evaluate whether simple assessment of the NIR water absorbance could yield information about matrix development. It was found that an increase in the mid-IR PG absorbance at 850 cm(-1) correlated with the area of the NIR water peak (Spearman's rho = 0.95, p < 0.0001). In the second analysis, a partial least squares method (PLS1) was used to assess whether an extended NIR spectral range (5400-3800 cm(-1)) could be utilized to predict collagen and proteoglycan content of the constructs based on mid-IR absorbances. A subset of spectra was randomly selected as an independent prediction set in this analysis. Average of the normalized root mean square errors of prediction of first-derivative NIR spectral models were 7% for 850 cm(-1) (PG), 11% for 1338 cm(-1) (collagen), 8% for amide II (collagen), and 8% for amide I (collagen). These results demonstrate the ability of NIRS to monitor macromolecular content of cartilage constructs and is the first step towards employing NIR to assess engineered cartilage in situ.