Cardiovascular (CV) parameters such as blood pressure (BP), electrocardiogram (ECG), and heart rate (HR) are routinely assessed in non-rodent toxicology studies, yet measurement quality varies by methodology—namely telemetry (implanted or jacketed) versus restraint-based techniques. This analysis examined how measurement methods affect baseline CV values and sensitivity to drug-induced effects. Data from three sources were analyzed: 495 toxicology studies from seven pharmaceutical companies (2015–2023), FDA recently approved drugs (47 NCEs, 26 NBEs in 2022–2023), and two major CROs (2020−2023). Studies involved dogs, minipigs, or non-human primates with dosing durations up to 52 weeks. Additional literature, proprietary data, and questionnaire informed baseline comparisons and statistical and pharmacological sensitivity. In our toxicology datasets, ECG was commonly collected, whereas BP was not. The selection of ECG and BP recording methods was mostly sponsor-dependent. Telemetry was predominantly used in short-duration studies; restraint-based snapshot methods were common in longer studies. Restraint consistently increased baseline BP and HR values and variability. QTc correction methods under restraint were inconsistent across sponsors/CROs. Statistical analysis was common with telemetry but infrequent under restraint. Case examples show that telemetry detected clinically-relevant BP or QTc changes; however, restraint methods usually did not. These findings demonstrate that restraint methods increase baseline values and variability, reducing pharmacological and statistical sensitivity to detect CV effects. Current practices appear driven by sponsor convention as well as scientific rationale. As ICH S7A/S7B evolve, re-evaluation of CV monitoring methods is warranted to enhance scientific rigor, regulatory alignment, and 3Rs compliance in nonclinical safety assessment.

Safety pharmacology is concerned with the identification and characterization of adverse effects of drug candidates on vital organ systems. The emergence of artificial intelligence (AI) and machine learning (ML) has prompted growing interest in their potential application to nonclinical drug safety evaluation across the core battery cardiovascular, central nervous, and respiratory systems. This review traces the historical development and technical foundations of AI, from early neural network research and backpropagation algorithms to the emergence of modern frontier large language models, and examines how these technologies are being applied to safety pharmacology study endpoints including proarrhythmic risk assessment consistent with International Council for Harmonisation (ICH) S7B and Comprehensive in vitro Proarrhythmia Assay (CiPA) frameworks, seizure liability detection via microelectrode array analysis, and respiratory function monitoring through whole-body plethysmography. The applications of AI in a broader toxicological assessment, including multi-endpoint toxicity prediction, digital pathology, and federated learning consortia, are also reviewed. To gauge current adoption and attitudes within the discipline, a survey of Safety Pharmacology Society (SPS) members was conducted at the 2024 annual meeting (N = 89). The survey revealed that 57% of respondents were not currently using AI tools, although 44% of non-users planned adoption within the following year; 84% of respondents intended to apply AI in preclinical safety development. The evolving regulatory landscape, including the 2025 United States Food and Drug Administration (FDA) draft guidance on AI credibility and the 2026 FDA/European Medicines Agency (EMA) joint guiding principles, is discussed alongside challenges related to data quality, model interpretability, and validation requirements. The findings indicate that while AI tools show promise for specific applications such as structure-based toxicity prediction and automated signal analysis, the safety pharmacology community appropriately demands rigorous validation before integration into regulated workflows. The challenges of model interpretability, data quality, and the absence of prospective validation studies represent substantive barriers that must be addressed through collaborative effort among industry, academia, regulatory agencies, and scientific societies. AI in safety pharmacology is currently best positioned as a complementary analytical tool that may help avoid investing resources in compounds with predictable safety liabilities, rather than as a replacement for expert scientific judgment.

Attrition in drug development due to central nervous system (CNS)-related adverse events is costly and can pose risks to volunteer safety during clinical development, underscoring the need to refine preclinical CNS screening strategies. The primary preclinical method to assess CNS risk is the Functional Observation Battery (FOB)/Irwin; however, these tests are limited by subjective assessment and restricted temporal resolution. Automated neurobehavioral assessment, such as the Home Cage Analyzer (HCA), may provide a solution to these challenges. Moreover, home cage assessments are in alignment with the 3Rs principles of refinement and reduction of animal use. Here, we detail our efforts to validate the HCA as part of GSK’s CNS safety assessment strategy. To this end, we evaluated the HCA using tool compounds-amphetamine, diazepam, and chlorpromazine-each with differing pharmacology and pharmacokinetics. Additionally, we employed a 4 × 4 Latin-square crossover study design allowing each animal to serve as its own control. Our results demonstrate that the HCA captured the expected pharmacological effects of stimulant and sedative compounds across differing light phases. Notably, at the highest amphetamine dose, we observed a distinct shift in behavior: horizontal locomotor activity decreased, vertical rearing activity markedly increased, a nuanced effect that could be missed by traditional observational methods and highlights the value of multimodal assessment. Together, these data strongly support the use of the HCA as a first-line tool in preclinical CNS safety assessment.

Drug-induced hypertension or hypotension can have significant clinical consequences, yet BP regulation involves complex, multi-level mechanisms, making translational assessment of pressor effects more challenging than for endpoints such as cardiac repolarization (hERG/QT). Current nonclinical BP studies primarily use telemetry in conscious, freely moving non-rodents, enabling continuous high-fidelity hemodynamic monitoring. These models show reasonable concordance with human data, but some translational challenges exist.The Health and Environmental Sciences Institute (HESI) Global Integrative Strategies Working Group (WG) was established to improve nonclinical cardiovascular (CV) safety assessment through more informative, translational preclinical models. After earlier success characterizing drug-induced effects on cardiac contractility, the WG focused on drug-associated blood pressure (BP) changes. Provided here is the importance of nonclinical models in predicting BP changes and associated challenges. A HESI Global multi-site study focused on addressing these gaps is also introduced. The study aimed to evaluate the reproducibility and translatability of the standard telemetered dog model using compounds with well-characterized mechanisms (midodrine, nifedipine, hydralazine, prazosin, milrinone). Early results show consistent hemodynamic profiles across laboratories, with final results published separately, showing utility of this model for BP liability assessment.

Attrition within clinical development due to central nervous system (CNS)-related adverse events continues to highlight the need to refine CNS screening strategies. The primary preclinical method to assess CNS risk is the Irwin/Functional Observation Battery (FOB). However, these tests are significantly limited by subjective assessment and temporal resolution. Automated neurobehavioral assessment, such as the Home Cage Analyzer (HCA), can overcome these issues. In addition, in vitro methods, such as microelectrode array (MEA), provide greater throughput and improved translation when using human iPSCs. Ideally, the combination of neurobehavioral and electrophysiological endpoints would converge to provide a clearer picture of CNS perturbance. Critically, a robust CNS screening strategy hinges upon understanding the limitations of such methods and identifying potential gaps in sensitivity between them. This study aimed to validate the Home Cage Analyzer (HCA) and test for convergence or difference in sensitivity between in vivo and in vitro screening using potent CNS active compounds from two distinct drug classes: psychostimulants and sedatives. For HCA experiments, eight male Wistar Han rats (Charles River) were implanted with temperature sensitive RFID transponders (Biomark USA) and pair housed in individually ventilated cages (Tecniplast). Rats were maintained under a 12-h light/dark cycle and habituated to the testing room and HCA rack for 1 week prior to initiation of dosing. Using a crossover Latin-square design, each rat received a single dose of d-amphetamine (0.25, 1, or 3 mg/kg), diazepam (0.5, 1.5, 5 mg/kg), or saline/vehicle. For MEA experiments, hiPSC-derived cortical neuron/astrocyte network (Fujifilm CDI) or rat cortex (E18.5; QBM Biosciences) was cultured in a Maestro Multi-electrode Array System (Axion BioSystems; 48-well plate). l-amphetamine in saline (0.12, 0.37, 1.11, 3.33, 10, 30 μM) or diazepam in DMSO (0.1, 0.3, 1, 3, 10 μM). As expected, we found a dose-dependent increase in activity after treatment with amphetamine in the HCA. Conversely, rats treated with diazepam showed decreased activity levels within the HCA. In line with neurobehavioral effects, diazepam caused a significant decrease in spiking and burst frequency in the MEA, resulting in a well-defined sedative phenotype. Surprisingly, amphetamine application in iPSCs exhibited no phenotypic response within the MEA, including no change in burst rate. Importantly, this lack of an effect in the MEA could be interpreted as a “clean” result, thus representing a missed hazard. In contrast, diazepam demonstrated consistent effects in the HCA and MEA. Together, these results indicate complementarity between in vitro MEA assay and in vivo automated HCA when screening sedatives while revealing a gap between these assays for amphetamine. Thus, a multi-modal approach is critical to a robust CNS screening strategy.

Increases in arterial blood pressure (BP) contribute to adverse cardiovascular (CV) outcomes in patients; preclinical effects of a drug on BP are routinely evaluated during the safety pharmacology assessments as outlined in the ICH S7A guidance. A Health and Environmental Sciences Institute (HESI) Consortium initiated a multi-site study with the objective to assess the ability of the standard conscious telemetry instrumented CV dog model to detect drug-induced changes in BP and evaluate translation to human data. The goal of these studies is also to determine the reproducibility and consistency of BP assessment when measured across different laboratories using the same study protocol and recording methodology to detect drug-induced changes in hemodynamics using drugs known to clinically elevate and reduce BP. Animals were chronically instrumented with a BP catheter and ECG electrodes for telemetric collection of hemodynamic and ECG endpoints, respectively. Study endpoints include systolic, diastolic, and mean BP, heart rate, electrocardiogram (ECG), body temperature, and locomotor activity. Drugs evaluated include midodrine (alpha-1 agonist), nifedipine (calcium channel blocker), hydralazine (direct-acting smooth muscle relaxant), prazosin (alpha-1 blocker) and milrinone (phosphodiesterase-3 inhibitor). Drugs were selected based on known pharmacological mechanisms of action, primary cardiovascular effects as well as availability of clinical effect and exposure data. Drugs were evaluated in beagle dogs using a double (8 × 4) Latin square design and administered orally at 3 doses selected to match clinical exposure data with a vehicle control. A full pharmacokinetic profile for each drug was conducted in dogs at doses selected using automated blood sampling (ABS). Initial analysis shows that all 5 positive control drugs show consistent hemodynamic profiles (e.g., BP elevation or reduction) in the dog as seen in humans. These data sets with additional testing at multiple sites will be amenable to further statistical analysis, super-interval analysis and follow-up study endpoint evaluation such as pressure waveform analysis. The results from this chronically instrumented conscious dog model will provide essential information about accuracy and consistency in blood pressure measurement across multiple sites and translation of preclinical BP data to clinical outcomes.

Core battery cardiovascular (CV) parameters [e.g., arterial blood pressure (BP), electrocardiogram (ECG), heart rate (HR)] are recorded in non-rodent safety pharmacology and toxicology studies. Data is used to support drug development in accordance with regulatory guidelines (e.g., ICH S7A. S7B S6, S9 and M3(R2)). However, the quality of measurements varies based on the methods used, ranging from restraint (unacclimated) to telemetry (implanted or jacketed)-based recordings. This analysis reviewed and compared current practice regarding methodological approaches applied to the acquisition of CV data in repeat-dose non-rodent toxicology studies. The data was sourced from 7 Sponsors, 2 major CROs and FDA regulatory drug approvals. Studies evaluated data acquired from dogs or non-human primates (NHP) with study durations up to 52 weeks. A literature-based search and proprietary published examples were used to establish baseline CV values and to ascertain the ability of the different methodologies used to detect drug-induced CV effects. The literature search provided evidence that baseline CV (BP and HR) values are consistently higher in restrained versus non-invasive telemetry methods. Moxifloxacin and proprietary drugs showed that hERG-mediated QTc prolongation was detected in the clinic and in NHP or dog studies using telemetry methods but not in restrained surface‑lead snapshot methods. Data showed that the ECG is collected in most repeat-dose toxicology studies, but BP is usually not collected. Overall, telemetry-based methods account for 61 % of ECG recordings; while much less frequent, restraint-based methods are used for 44 % of BP recordings, when collected. The choice and usage of collection methods is highly sponsor-dependent, with restraint-based methods for individual sponsors ranging from 0 to 100 %. Data showed that telemetry-based methods are predominately used in short-duration studies for drug safety assessments, but restraint-based snapshot methods are used in longer-duration studies. The data compiled thus far serves as a basis to consider CV collection methods used in toxicology studies and to develop recommendations on the measurement of ECG and BP in non-rodent species in support of regulatory safety assessment studies.

Drug-induced changes in cardiac contractility can lead to cardiotoxicity, a major cause of drug development discontinuation. Current routine in vitro strategies to assess changes in contractility for both small and large molecules are limited to low/medium throughput assays, not always correctly translating in clinical trials. The goal of this study is to present a computational framework for predicting drug-induced changes in human cardiac contractility and to compare in silico predictions with in vitro measurements of sarcomere shortening from adult human primary cardiomyocytes and clinical data available in the literature. In addition to simulating the behavior of pure specific ion channels blockers, a set of 28 reference compounds was simulated for this study. For each compound, half maximal inhibitory concentration (IC50) values were provided for the inhibition of four ionic currents: the fast Na + current, the L-type Ca2+ current (ICaL), the rapid and slow K+ currents. Wide ranges of drug concentrations, from 0.1× up to 100× of the free therapeutic plasma concentrations, were simulated in an experimentally-calibrated population of more than 300 human ventricular electromechanical in silico models constituting a healthy control population, with different ion channel expressions represented to account for cell-to-cell variability. Fifteen electrophysiological and contractility biomarkers were computed and analyzed. Active tension (AT) peak was identified as the most informative one and used to compute AT IC50. Pure ICaL blockers showed a strong reduction of AT peak and max/min AT velocity, whereas minor changes were observed for AT time-to-peak and AT relaxation. In silico predictions were qualitatively in agreement with in vitro recordings from adult human primary cardiomyocytes from literature for 27 out of 28 compounds. Cisapride was the only exception, showing no effects on simulated AT, whereas a negative inotropic effect was observed in vitro, despite there is no evidence of cisapride affecting contractility in clinical studies. Our study highlights the potential of in silico drug trials for early detection of drug-induced changes on cardiac contractility as a promising tool for use during drug development.

Elevated arterial blood pressure (BP) is highly correlated with adverse cardiovascular (CV) outcomes in patients. The effects of a drug on arterial pressure are routinely evaluated during preclinical safety assessment as outlined in the ICH S7A guidance document. A Health and Environmental Sciences Institute (HESI) Consortium initiated a multi-site study with the objective to assess the ability of the standard conscious telemetered CV beagle dog model to detect drug-induced changes in BP and evaluate translation to human data. Animals will be chronically instrumented with a BP catheter and ECG electrodes for telemetric collection of hemodynamic endpoints. Study endpoints include systolic, diastolic, and mean BP, heart rate, electrocardiogram (ECG), body temperature, and locomotor activity. Drugs evaluated include midodrine (alpha-1 agonist), nifedipine (calcium channel blocker), hydralazine (direct-acting smooth muscle relaxant), prazosin (alpha-1 blocker) and milrinone (phosphodiesterase-3 inhibitor) were selected based on known mechanisms of action as well as availability of clinical exposure data. Drugs will be evaluated in beagle dogs using a single (4 × 4) or double (8 × 4) Latin square design. Drugs will be administered orally at 3 doses selected to match clinical exposure data and a vehicle control. A full pharmacokinetic profile for each drug will be conducted at the doses selected at a single site (Abbvie). Blood samples at participating sites will be drawn to confirm drug exposures predicted from independent pharmacokinetic studies. The goal of these studies is to determine whether the assessment of BP, when measured across different laboratories using the same protocol, can consistently detect drug-induced changes in hemodynamics using drugs known to clinically increase and decrease BP.

Reliable prediction and prevention of adverse drug reactions (ADRs) remains a key challenge in the development of new medicines. Advanced mathematical and computational modelling approaches, which incorporate cutting-edge mechanistic understanding of ADRs in concert with systematically collected data addressing knowledge gaps, are integral components of model-informed drug discovery and development (MID3). These approaches provide a precise, quantitative framework for predicting and mitigating safety risks in the earliest phases of drug development. Here, we highlight recent developments in the burgeoning field of quantitative systems toxicology (QST), including insights into the current state-of-the-art, as well as outcomes from the Innovative Medicines Initiative (IMI) 2 TransQST project. QST models that describe the disruption of cardiovascular, gastrointestinal, hepatic and renal physiological functions following drug exposure are presented, along with recommendations for their application in drug discovery and development.

Changes in cardiac contractility can lead to hemodynamic alterations and abnormal structural changes within the heart, which over time can cause reduced function and failure. Recent advancements in technology, allowing the assessment of cardiac contractility to become easier and more efficient, have resulted in a greater focus on contractility measures within safety pharmacology studies. In particular, the development of in vitro assays such as the FLEXcyte assay, which utilizes human stem cell derived cardiomyocytes (hiPSC-CMs), allows for the ability to screen for potential contractility changes earlier in development. The objective of this study was to explore the concordance of changes in contractility between FLEXcyte measurements and left ventricular pressure measurements (LV + dP/dtmax) in the telemetered rat model. Verapamil and milrinone were selected as tool compounds. For the FLEXcyte, hiPSC-CMs (Cardiosight-S; NEXCEL Co., Ltd.) were maintained in a serum-free culture medium for 7 days with media changes every 48 h. Initially, 30 s of baseline was recorded every 15 min for 1 h. Following baseline, a dose response curve was attained using 5 concentrations. Spontaneous beat rate and contraction amplitude were analyzed to determine the concordance with heart rate and LV + dP/dtmax in the telemetered rat. For the telemetered study, CRL WI(Han) rats were instrumented with HD-S21 (DSI) implants monitoring systemic and LV pressures and placed into a 4 × 4 latin square design for each tool compound. For verapamil, the FLEXcyte showed an increase in beat rate (22.7 % at 33 nM) and a substantial decrease in amplitude (66.5 % at 33 nM), which concords well with the increase in heart rate (25.2 %) and decrease in LV + dP/dtmax (−33.6 %) in the telemetered rat. However, for milrinone, the FLEXcyte did not show any significant changes in beat rate or amplitude while the telemetered rat showed increases in both heart rate (up to 57 %) and LV + dP/dtmax (up to 87.4 %). While more work is needed, this data highlights the importance of conducting both in vitro and in vivo studies to assess the potential effects on cardiac contractility in drug development, allowing for the identification of molecules with diverse mechanisms of action.

INTRODUCTION:Cardiovascular (CV) parameters1 such as blood pressure (BP), electrocardiogram (ECG), and heart rate (HR) are recorded in non-rodent non-clinical safety studies to support drug development. However, measurement quality varies depending on the methodology used, including restraint-based or telemetry (implanted or jacketed) techniques. Measurement quality, in this context, refers to the sensitivity and reliability of CV measurements in affecting baseline values of measured CV parameters and in detecting pharmacological effects. This retrospective multifaceted analysis evaluated the impact of recording methods on baseline CV parameters and their statistical and pharmacological sensitivities in detecting drug-induced CV effects. METHODS:Data were collected from three sources: (i) 495 studies from seven pharmaceutical sponsors (2015-2023), (ii) FDA-approved drugs (47 NCEs, 26 NBEs from 2022 to 2023), and (iii) two major CROs (2020-2023). Studies were conducted in dogs, non-human primates (NHP), or minipigs, with treatment durations of up to 52 weeks. Additionally, literature-based and proprietary data were analyzed to assess baseline CV values and methodology sensitivity. A survey was conducted to evaluate statistical analysis practices in these studies. RESULTS:All 3 datasets showed that the ECG is collected in most repeat dose toxicology studies, but not BP; the findings were largely independent on the modality (i.e., NCE versus NBE) or the indication (i.e., oncology vs. non-oncology). The choice and usage of ECG and BP methods is highly sponsor-dependent, with restraint-based methods for individual sponsors ranging from 0 to 100 %. FDA data showed that telemetry-based methods are predominantly used in short, single dose toxicology/safety pharmacology studies for NCEs. Subsequent studies of longer duration employ predominantly restraint-based snapshot methods. CRO data showed that approximately 30 % of toxicology studies do not include ECGs; however, when an ECG is recorded it is primarily collected in restrained animals using a snapshot approach. BP is infrequently recorded, regardless of methodology, in repeat dose toxicology studies. The de novo analysis and literature-based search showed that baseline BP/HR values were highly variable with consistently higher means under restraint compared to telemetry methods. The root mean square errors for BP/HR were larger under restrained conditions, in both species. Under restrained conditions, the use of fixed formulae for HR-corrected QT resulted in inconsistent QTc values across sponsor and CROs. The survey showed that statistical analysis of ECG/BP data was infrequently performed under restrained conditions in contrast to telemetry-based methods. Proprietary and published case studies showed that drug-induced BP elevation or QTc prolongation observed clinically and in NHP or dog using telemetry were not reliably detected under restraint-based conditions, highlighting limitations of RB methods in CV safety evaluation. The data illustrate that animal restraint reduces the pharmacological and statistical sensitivities to detect CV effects. CONCLUSIONS:ECG and BP recording methods vary based on sponsor preference, experience and/or institutional practices in addition to scientific rationale. Literature and case studies confirm the limitations of restraint-based methods. As ICH S7A/S7B evolve, revisiting CV assessment practices is necessary to align with scientific, technological, regulatory, and 3Rs considerations, ultimately improving best practices for regulatory safety assessment.

QTc is an established biomarker for drug-induced Torsade de Pointes (TdP) but with concerns for a false positive signal. Clinically, JTpc and TpTec have emerged as ECG sub-intervals to differentiate predominant hERG vs. mixed ion channel blocking drugs that prolong QTc. Drugs that prolong QTc with increased TpTe, but without JTpc effects, are likely mixed ion channel blockers with low TdP risk. Drug effects on QTc, JTpc and TpTec were characterized with cynomolgus monkeys using telemetry in a Lead II configuration. Drugs and vehicle were administered orally to group sizes of 4 to 8 animals, in 4 laboratories. In monkeys, dofetilide (0.03–0.3 mg/kg, PO) was associated with exposure dependent QTc and JTpc increases but no significant TpTec effect. Quinidine (2 to 50 mg/kg, PO) increased QTc and JTpc but did not change TpTec. Mexiletine (1–15 mg/kg, PO) and verapamil (50 mg/kg, PO) did not induce any significant effect on QTc, JTpc or TpTec. Clinically, predominant hERG blockers (dofetilide and quinidine) prolong QTc, JTpc and TpTec and are associated with an increased risk for TdP. Results from the current study demonstrate that ECG changes after dofetilide and quinidine administration to telemetered monkeys differ from the clinical response, lacking the expected effects on TpTec. Potential explanations for the lack of translation include physio-pharmacology species differences or ECG recording and analysis methodology variations. Mixed ion channel blockers verapamil and mexiletine administered to monkeys showed no significant QTc, JTpc or TpTec prolongation as expected based on the similar clinical response for these agents.

Life supportive cardiovascular (CV) parameters [e.g., arterial blood pressure (BP), electrocardiogram (ECG), heart rate (HR)] are recorded in non-rodent safety pharmacology and toxicology studies and data are used to support pharmaceutical drug development in accordance with several guidelines (e.g., ICH M3, S6, S7, S9). Various methodological approaches are used to collect CV parameters on toxicology studies, including fully implantable telemetry, non-invasive jacket telemetry, and short-duration restraint-based measurement from surface ECG leads and a blood pressure cuff, but their robustness and applicability domain differ widely. The challenge of using conscious animals, especially non-human primates, is that arousal behavior, locomotion, body temperature, and sympathetic activation are significant sources of variability that impact CV parameters, thus telemetry methods are the preferred approach. Historically, toxicology studies have incorporated short-duration restraint methods (manual; chemical sedation) to collect brief periods (“a snapshot at one timepoint”) of CV endpoints in a large number of non-rodents (N ≥ 24) after repeat dosing. Non-rodent species, however, demonstrate increased BP and HR in response to room entry and/or manual restraint, which confounds the interpretation of drug-related effects. Published case examples and company experiences indicate that CV waveforms collected under restraint are of low quality, can vary based on body position and sensor placement (skin electrode; pressure cuff), as well as the type of sedative and dose used. In addition, the data captured can be misleading or incongruent with CV safety pharmacology findings in the same animal species. As a result, restraint-induced stress lowers the sensitivity to detect CV changes as measured over a short-duration and has a higher probability for false negative findings. Although functional CV evaluations in restrained animals have been included in toxicology studies for decades, there is no industry consensus on methods, assay performance, or value (translation) for risk assessment. This communication will review the available literature and leverage pharmaceutical and Contract Research Organization (CRO) experiences, and propose recommendations, with an emphasis on short-duration techniques, from the ICH E14/S7B Industry Support Group with the aim to identify how, when, and if short term restraint-based CV data are valuable for nonclinical safety assessment.

Drug-induced changes in cardiac contractility (inotropy) can lead to cardiotoxicity, a major cause of discontinuation in drug development. Preclinical approaches to assess cardiac inotropy are imperfect, with in vitro assays limited to stem cell-derived or adult human primary cardiomyocytes. Human mechanistic in silico modelling and simulations are already successfully applied for proarrhythmia prediction, contributing to cardiac safety assessment strategies in early drug development. In this study, we investigated their ability to predict drug-induced effects on cardiac inotropy. We considered a validation set of 28 neutral/negative inotropic and 13 positive inotropic reference compounds and simulated their effects on cell contractility via ion channel inhibition and perturbation of nine biomechanical modelling parameters, respectively. For each compound, a wide range of drug concentrations was simulated in an experimentally calibrated control population of 323 human ventricular in silico cells. Simulated biomarkers indicating drug-induced inotropic effects were compared with in vitro preclinical data from the literature. Computer simulations predicted drug-induced inotropic changes observed in vitro for 25 neutral/negative inotropes and 10 positive inotropes. Predictions of negative inotropic changes were quantitatively in agreement for 86% of tested drugs. Active tension peak was identified as the biomarker with highest predictive potential. This study describes the validation and application of an in silico cardiac electromechanical model for drug safety evaluation, combining ion channel inhibition data and information on potential inotropic mechanisms to predict inotropic changes. Furthermore, a route for its integration as part of a preclinical drug safety assessment strategy is outlined.

Optimization of ICH safety guideline studies for inclusion into regulatory submissions is critical for resource conservation, animal use reduction, and efficient drug development. The ICH S7A guidance for Safety Pharmacology (SP) studies adopted in 2001 identified the core battery of studies to evaluate the acute safety of putative pharmaceutical molecules prior to First in Human (FIH) trials. To assess the utility of respiratory studies in predicting clinical AE’s, seven pharmaceutical companies pooled preclinical and clinical respiratory findings. A large database of novel molecules included all relevant data from standard S7A respiratory (n = 459) and FIH studies (n = 309). The data were analyzed with respect to the progression of these molecules, clinical adverse event reporting of these same molecules, and achieved exposures. These S7A respiratory assay findings had no impact on compound progression, and only 12 of 309 drug candidates were ‘positive’ preclinically and reported a respiratory-related AE in clinical trials (i.e. cough, dyspnea, etc.), an overall incidence rate of 3.9%. Contingency tables/statistics support a lack of concordance of these preclinical assays. Overall, our extensive analysis clearly indicated that the preclinical respiratory assay fails to provide any prognostic value for detecting clinically relevant respiratory adverse events.

Introduction: Corrected QT interval (QTc)is an established biomarker for drug-induced Torsade de Pointe (TdP), but with concerns for a false positive signal. Clinically, JTpc and TpTec have emerged as ECG sub-intervals to differentiate predominant hERG vs. mixed ion channel blocking drugs that prolong QTc. Methods: In a multicentric, prospective, controlled study, different proarrhythmic drug effects on QTc, JTpc and TpTec were characterized with cynomolgus monkeys using telemetry in a Lead II configuration for internal and external telemetry.Drugs and vehicle were administered orally (PO) to group size of 4 to 8 animals, in 4 laboratories. Results: In monkeys, dofetilide (0.03-0.3 mg/kg) was associated with exposure dependent QTc and JTpc increase, but no significant TpTec effect. Similarly, quinidine (2-50 mg/kg) increased QTc and JTpc but did not change TpTec. Mexiletine (1-15 mg/kg) and verapamil (50 mg/kg) did not induce any significant effect on QTc, JTpc or TpTec. Discussion: Clinically, predominant hERG blockers (dofetilide and quinidine) prolong QTc, JTpc and TpTec and are associated with increased risk for TdP. Results from this study demonstrate that ECG changes after dofetilide and quinidine administration to telemetered monkeys differ from the clinical response, lacking the expected effects on TpTec. Potential explanations for the lack of translation include physio-pharmacology species differences or ECG recording and analysis methodology variations. Mixed ion channel blockers verapamil and mexiletine administered to monkeys showed no significant QTc, JTpc or TpTec prolongation as expected based on the similar clinical response for these agents.