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.

The evolving role of investigative toxicology in the pharmaceutical industry.

For decades, preclinical toxicology was essentially a descriptive discipline in which treatment-related effects were carefully reported and used as a basis to calculate safety margins for drug candidates. In recent years, however, technological advances have increasingly enabled researchers to gain insights into toxicity mechanisms, supporting greater understanding of species relevance and translatability to humans, prediction of safety events, mitigation of side effects and development of safety biomarkers. Consequently, investigative (or mechanistic) toxicology has been gaining momentum and is now a key capability in the pharmaceutical industry. Here, we provide an overview of the current status of the field using case studies and discuss the potential impact of ongoing technological developments, based on a survey of investigative toxicologists from 14 European-based medium-sized to large pharmaceutical companies.

Optimizing drug discovery by Investigative Toxicology: Current and future trends.

Investigative Toxicology describes the de-risking and mechanistic elucidation of toxicities, supporting early safety decisions in the pharmaceutical industry. Recently, Investigative Toxicology has contributed to a shift in pharmaceutical toxicology, from a descriptive to an evidence-based, mechanistic discipline. This was triggered by high costs and low throughput of Good Laboratory Practice in vivo studies, and increasing demands for adhering to the 3R (Replacement, Reduction and Refinement) principles of animal welfare. Outside the boundaries of regulatory toxicology, Investigative Toxicology has the flexibility to embrace new technologies, enhancing translational steps from in silico, in vitro to in vivo mechanistic understanding to eventually predict human response. One major goal of Investigative Toxicology is improving preclinical decisions, which coincides with the concept of animal-free safety testing. Currently, compounds under preclinical development are being discarded due to the use of inappropriate animal models. Progress in Investigative Toxicology could lead to humanized in vitro test systems and the development of medicines less reliant on animal tests. To advance this field a group of 14 European-based leaders from the pharmaceutical industry founded the Investigative Toxicology Leaders Forum (ITLF), an open, non-exclusive and pre-competitive group that shares knowledge and experience. The ITLF collaborated with the Centre for Alternatives to Animal Testing Europe (CAAT-Europe) to organize an "Investigative Toxicology Think-Tank", which aimed to enhance the interaction with experts from academia and regulatory bodies in the field. Summarizing the topics and discussion of the workshop, this article highlights Investigative Toxicology's position by identifying key challenges and perspectives.

Analysis of negative historical control group data from the in vitro micronucleus assay using TK6 cells.

The recent revisions of the Organisation for Economic Co-operation and Development (OECD) genetic toxicology test guidelines emphasize the importance of historical negative controls both for data quality and interpretation. The goal of a HESI Genetic Toxicology Technical Committee (GTTC) workgroup was to collect data from participating laboratories and to conduct a statistical analysis to understand and publish the range of values that are normally seen in experienced laboratories using TK6 cells to conduct the in vitro micronucleus assay. Data from negative control samples from in vitro micronucleus assays using TK6 cells from 13 laboratories were collected using a standard collection form. Although in some cases statistically significant differences can be seen within laboratories for different test conditions, they were very small. The mean incidence of micronucleated cells/1000 cells ranged from 3.2/1000 to 13.8/1000. These almost four-fold differences in micronucleus levels cannot be explained by differences in scoring method, presence or absence of exogenous metabolic activation (S9), length of treatment, presence or absence of cytochalasin B or different solvents used as vehicles. The range of means from the four laboratories using flow cytometry methods (3.7-fold: 3.5-12.9 micronucleated cells/1000 cells) was similar to that from the nine laboratories using other scoring methods (4.3-fold: 3.2-13.8 micronucleated cells/1000 cells). No laboratory could be identified as an outlier or as showing unacceptably high variability. Quality Control (QC) methods applied to analyse the intra-laboratory variability showed that there was evidence of inter-experimental variability greater than would be expected by chance (i.e. over-dispersion). However, in general, this was low. This study demonstrates the value of QC methods in helping to analyse the reproducibility of results, building up a 'normal' range of values, and as an aid to identify variability within a laboratory in order to implement processes to maintain and improve uniformity.

Integrating High-Dimensional Transcriptomics and Image Analysis Tools into Early Safety Screening: Proof of Concept for a New Early Drug Development Strategy.

During drug discovery and development, the early identification of adverse effects is expected to reduce costly late-stage failures of candidate drugs. As risk/safety assessment takes place rather late during the development process and due to the limited ability of animal models to predict the human situation, modern unbiased high-dimensional biology readouts are sought, such as molecular signatures predictive for in vivo response using high-throughput cell-based assays. In this theoretical proof of concept, we provide findings of an in-depth exploration of a single chemical core structure. Via transcriptional profiling, we identified a subset of close analogues that commonly downregulate multiple tubulin genes across cellular contexts, suggesting possible spindle poison effects. Confirmation via a qualified toxicity assay (in vitro micronucleus test) and the identification of a characteristic aggregate-formation phenotype via exploratory high-content imaging validated the initial findings. SAR analysis triggered the synthesis of a new set of compounds and allowed us to extend the series showing the genotoxic effect. We demonstrate the potential to flag toxicity issues by utilizing data from exploratory experiments that are typically generated for target evaluation purposes during early drug discovery. We share our thoughts on how this approach may be incorporated into drug development strategies.

Are zebrafish larvae suitable for assessing the hepatotoxicity potential of drug candidates?

Drug-induced liver injury (DILI) is poorly predicted by single-cell-based assays, probably because of the lack of physiological interactions with other cells within the liver. An intact whole liver system such as one present in zebrafish larvae could provide added value in a screening strategy for DILI; however, the possible occurrence of other organ toxicities and the immature larval stage of the zebrafish might complicate accurate and fast analysis. We investigated whether expression analysis of liver-specific fatty acid binding protein 10a (lfabp10a) was an appropriate endpoint for assessing hepatotoxic effects in zebrafish larvae. It was found that expression analysis of lfabp10a was a valid marker, as after treatment with hepatotoxicants, dose-response curves could be obtained and statistically significant abnormal lfabp10 expression levels correlated with hepatocellular histopathological changes in the liver. However, toxicity in other vital organs such as the heart could impact liver outgrowth and thus had to be assessed concurrently. Whether zebrafish larvae were suitable for assessing human relevant drug-induced hepatotoxicity was assessed with hepatotoxicants and non-hepatotoxicants that have been marketed for human use and classified according to their mechanism of toxicity. The zebrafish larva showed promising predictivity towards a number of mechanisms and was capable of distinguishing between hepatotoxic and non-hepatotoxic chemical analogues, thus implying its applicability as a potential screening model for DILI.

Flow cytometric 96-well microplate-based in vitro micronucleus assay with human TK6 cells: protocol optimization and transferability assessment.

An automated approach for scoring in vitro micronuclei (MN) has been described in which flow cytometric analysis is combined with compound exposure, processing, and sampling in a single 96-well plate (Bryce SM et al. [2010]: Mutat Res 703:191-199). The current report describes protocol optimization and an interlaboratory assessment of the assay's transferability and reproducibility. In a training phase, the methodology was refined and collaborating laboratories were qualified by repeatedly testing three compounds. Second, a set of 32 chemicals comprised of reference genotoxicants and presumed non-genotoxicants was tested at each of four sites. TK6 cells were exposed to 10 closely spaced compound concentrations for 1.5- to 2-cell population doublings, and were then stained and lysed for flow cytometric analysis. MN frequencies were determined by evaluating ≥ 5,000 cells per replicate well, and several indices of cytotoxicity were acquired. The prevalence of positive results varied according to the MN-fold increase used to signify a genotoxic result, as well as the endpoint used to define a cytotoxicity limit. By varying these parameters, assay sensitivity and specificity values ranged from 82 to 98%, and 86 to 97%, respectively. In a third phase, one laboratory tested a further six genotoxicants and five non-genotoxic apoptosis inducers. In these experiments assay specificity was markedly improved when top concentration selection was based on two cytotoxicity endpoints-relative survival and quantification of ethidium monoazide-positive events. Collectively, the results indicate that the miniaturized assay is transferable across laboratories. The 96-well format consumes considerably less compound than conventional in vitro MN test methods, and the high information content provided by flow cytometry helps guard against irrelevant positive results arising from overt toxicity.

Suitable top concentration for tests with mammalian cells: mouse lymphoma assay workgroup.

The Mouse Lymphoma Expert Workgroup of the International Workshop for Genotoxicity Tests (IWGT) met in Basel, Switzerland in August of 2009. The Workgroup (WG) was tasked with discussing the appropriate top concentration for non-pharmaceuticals that would be required for the conduct of the mouse lymphoma assay (MLA) when sufficient cytotoxicity [to between 10 and 20% relative total growth (RTG)] has not been attained. The WG approached this task by (1) enumerating the various regulatory decisions/use for MLA data, (2) discussing the appropriate assays to which MLA data and assay performance should be compared and (3) discussing all the proposals put forth concerning the top concentration for non-pharmaceuticals. In addition, one of the members presented a summary of a re-evaluation of the National Toxicology Program MLA data using the IWGT harmonized guidance that was underway as a separate (non IWGT) activity, being conducted by two members of the Expert WG. The WG was asked to vote on each of the various proposals for top concentration for when cytotoxicity is not concentration limiting. While there was general agreement that the top concentration for non-pharmaceuticals should be re-evaluated and likely lowered from the current recommended levels, there was no agreement on a specific new recommendation.

Follow-up actions from positive results of in vitro genetic toxicity testing.

Appropriate follow-up actions and decisions are needed when evaluating and interpreting clear positive results obtained in the in vitro assays used in the initial genotoxicity screening battery (i.e., the battery of tests generally required by regulatory authorities) to assist in overall risk-based decision making concerning the potential effects of human exposure to the agent under test. Over the past few years, the International Life Sciences Institute (ILSI) Health and Environmental Sciences Institute (HESI) Project Committee on the Relevance and Follow-up of Positive Results in In Vitro Genetic Toxicity (IVGT) Testing developed a decision process flow chart to be applied in case of clear positive results in vitro. It provides for a variety of different possibilities and allows flexibility in choosing follow-up action(s), depending on the results obtained in the initial battery of assays and available information. The intent of the Review Subgroup was not to provide a prescriptive testing strategy, but rather to reinforce the concept of weighing the totality of the evidence. The Review Subgroup of the IVGT committee highlighted the importance of properly analyzing the existing data, and considering potential confounding factors (e.g., possible interactions with the test systems, presence of impurities, irrelevant metabolism), and chemical modes of action when analyzing and interpreting positive results in the in vitro genotoxicity assays and determining appropriate follow-up testing. The Review Subgroup also examined the characteristics, strengths, and limitations of each of the existing in vitro and in vivo genotoxicity assays to determine their usefulness in any follow-up testing.

Flow cytometric analysis of micronuclei in mammalian cell cultures: past, present and future.

The relative simplicity of the in vitro micronucleus (MNvit) endpoint has made it amenable to several automated scoring approaches. Flow cytometry is one such scoring platform that has been successfully employed. This review describes the origins of the MNvit assay, as well as the evolution and properties of flow cytometry-based scoring systems. While the current state-of-the-art methods acquire micronucleus (MN) frequency data very efficiently, it is becoming clear that they also endow the assay with high information content. For instance, simultaneous with MN frequency determinations, several additional endpoints are acquired that provide insights into cytotoxicity, cell cycle perturbations and, in the event of MN induction, information about genotoxic mode of action. This review concludes with a discussion regarding data gaps and also recommendations for additional work that is needed to more fully realise the potential of flow cytometric MNvit scoring.

Development of a new opacitometer for the bovine corneal opacity and permeability (BCOP) assay.

The bovine corneal opacity and permeability (BCOP) assay is an in vitro eye irritation test that relies on the measurement of induced opacity and permeability evaluated in isolated bovine corneas. Routinely, opacity is measured by an OP-KIT opacitometer which provides a centre-weighted reading of light transmission through the corneal tissue. However, this may underestimate opacity that develops as spots or heterogeneous opaque areas on the periphery of isolated cornea. Consequently, these findings question the reliability of the obtained data when opacity is manifested in a non-homogeneous manner. In addition, the BCOP test is only recommended to identify corrosive and severe eye irritants, and has not proven as sensitive in distinguishing among mild and moderate eye irritants. Within this context, a new opacitometer was developed which uses an adjustable laser beam in combination with a calibrated photometer. This work focused on a technical optimisation and describes the optical characteristics of the new device. The laser-based opacitometer allowed the analysis of the complete corneal surface, showed improved sensitivity to detect subtle changes in corneal transparency and was able to detect more efficiently opaque spots located along the sides of the excised corneas.

Screening for phospholipidosis induced by central nervous drugs: comparing the predictivity of an in vitro assay to high throughput in silico assays.

Drug-induced phospholipidosis is a side effect for which drug candidates can be screened in the drug discovery phase. The numerous in silico models that have been developed as a first line of screening are based on the characteristic physicochemical properties of phospholipidosis-inducing drugs, e.g. high logP and pK(b) values. However, applying these models on a predominantly high lipophilic, basic CNS chemistry results in a high false positive rate and consequently in a wrong classification of a large number of valuable drug candidates. Here, we tested 33 CNS-compounds (24 in vivo negative and 9 in vivo positive phospholipidosis-inducers) in our in house developed in vitro phospholipidosis screening assay (Mesens et al., 2009) and compared its predictivity with the outcome of three different, well established in silico prediction models. Our in vitro assay demonstrates an increased specificity of 79% over the in silico models (29%). Moreover, by considering the proposed plasma concentration at the efficacious dose we can show a clear correlation between the in vitro and in vivo occurrence of phospholipidosis, improving the specificity of prediction to 96%. Through its high predictive value, the in vitro low throughput assay is thus preferred above high throughput in silico assays, characterized by a high false positive rate.

Optimisation of the cell cultivation methods in the embryonic stem cell test results in an increased differentiation potential of the cells into strong beating myocard cells.

In order to support drug research in the selection process for non-embryotoxic pharmaceutical compounds, a screening method for embryotoxicity is needed. The murine embryonic stem cell test (EST) is a validated in vitro test based on two permanent mouse cell lines and delivering results in 10-days. Implementation of this test within our laboratory, revealed variability in the differentiation potential of the embryonic stem cells and, as a consequence, a lot of assays needed to be rejected due the fact the acceptance criteria were not reached. In order to gain a better yield of contracting myocardial cells, we used (1) a stringent control of the cell growth during subcultivation and a standardised hanging drop culture method and (2) a non-enzymatic cell harvest instead of a trypsin/EDTA cell harvest. Implementing of these cell culture modifications resulted in a decreased variability in the size of embryonic bodies, an increase of the number of acceptable tests and a significant increase of the differentiation potential of embryonic cells into strong beating myocardium, which made scoring less time consuming. Testing of 6 reference compounds in the optimized EST showed that the cell culture modifications did not changed the in vitro classification.

Interlaboratory evaluation of a flow cytometric, high content in vitro micronucleus assay.

An international, multi-lab trial was conducted to evaluate a flow cytometry-based method for scoring micronuclei in mouse lymphoma L5178Y cells [S.L. Avlasevich, S.M. Bryce, S.E. Cairns, S.D. Dertinger, In vitro micronucleus scoring by flow cytometry: differential staining of micronuclei versus apoptotic and necrotic chromatin enhances assay reliability, Environ. Mol. Mutagen. 47 (2006) 56-66]. A reference laboratory investigated the potential of six chemicals to induce micronuclei -- the genotoxicants mitomycin C (MMC), etoposide (ETOPO), and vinblastine (VB), and the non-genotoxicants sucrose (SUC), staurosporine (STS), and dexamethasone (DEX). The latter two non-genotoxicants were selected as extreme challenges to the assay because of their potent apoptogenic activity. Three collaborating laboratories were supplied with prototype In Vitro MicroFlow kits, and each was assigned one genotoxicant and one non-genotoxicant. Cells were treated continuously for 24h over a range of concentrations up to 5 mg/ml, or overtly cytotoxic concentrations. Micronuclei were scored via standard microscopy and flow cytometry. In addition to enumerating micronucleus frequencies, a cytotoxicity measurement that is simultaneously acquired with the flow cytometric micronucleus scoring procedure was evaluated (Flow-NBR). With this method, latex particles served as counting beads, and facilitated relative survival measurements that exclude the presence of dead/dying cells. For comparison purposes, additional cytotoxicity endpoints were measured, including several that are based on cell number, and others that reflect compromised membrane integrity, including dye permeability and/or phospholipid distribution. Key findings for this set of compounds include the following: (1) significant discrepancies in top concentration selection were found when cytotoxicity measurements were based on different methods, with the Flow-NBR approach tending to be the most sensitive, (2) both microscopy- and flow cytometry-based scoring methods detected concentration-dependent micronucleus formation for the three genotoxic agents studied, with good agreement between the reference laboratory and the collaborating laboratories, and (3) whereas flow cytometric analyses showed no significant increases for the non-genotoxicants when top concentration selection was based on Flow-NBR, significantly elevated micronucleus frequencies were observed for concentrations that were chosen based on less-sensitive cytotoxicity assays. Collectively, these results indicate that rapid assessment of genotoxicity can be accomplished with a relatively simple flow cytometric technique, and that the scoring system is transferable across laboratories. Furthermore, a concurrent assessment of cytotoxicity, Flow-NBR, may help reduce the occurrence of irrelevant positive results, as it may represent a more appropriate means for choosing top concentration levels. Finally, the data presented herein reinforce concerns about the manner in which cytotoxicity limits are described in guidance documents, since these recommendations tend to cite fixed cut-off values without reference to methodology.

Mouse lymphoma thymidine kinase gene mutation assay: meeting of the International Workshop on Genotoxicity Testing, San Francisco, 2005, recommendations for 24-h treatment.

The Mouse Lymphoma Assay (MLA) Workgroup of the International Workshop on Genotoxicity Testing (IWGT), comprised of experts from Japan, Europe and the United States, met on September 9, 2005, in San Francisco, CA, USA. This meeting of the MLA Workgroup was devoted to reaching a consensus on issues involved with 24-h treatment. Recommendations were made concerning the acceptable values for the negative/solvent control (mutant frequency, cloning efficiency and suspension growth) and the criteria to define an acceptable positive control response. Consensus was also reached concerning the use of the global evaluation factor (GEF) and appropriate statistical trend analysis to define positive and negative responses for the 24-h treatment. The Workgroup agreed to continue their support of the International Committee on Harmonization (ICH) recommendation that the MLA assay should include a 24-h treatment (without S-9) in those situations where the short treatment (3-4 h) gives negative results.

Mouse lymphoma thymidine kinase gene mutation assay: follow-up meeting of the International Workshop on Genotoxicity Testing--Aberdeen, Scotland, 2003--Assay acceptance criteria, positive controls, and data evaluation.

The Mouse Lymphoma Assay (MLA) Workgroup of the International Workshop on Genotoxicity Testing (IWGT), comprised of experts from Japan, Europe, and the United States, met on August 29, 2003, in Aberdeen, Scotland, United Kingdom. This meeting of the MLA Workgroup was devoted to reaching a consensus on the appropriate approach to data evaluation and on acceptance criteria for both the positive and negative/vehicle controls. The Workgroup reached consensus on the acceptance criteria for both the agar and microwell versions of the MLA. Recommendations include acceptable ranges for mutant frequency, cloning efficiency, and suspension growth of the negative/vehicle controls and on criteria to define an acceptable positive control response. The recommendation for the determination of a positive/negative test chemical response includes both the requirement that the response exceeds a defined value [the global evaluation factor (GEF)] and that there also be a positive dose-response (evaluated by an appropriate statistical method).

Flow cytometric analysis of micronucleated reticulocytes: Time- and dose-dependent response of known mutagens in mice, using multiple blood sampling.

According to the current Organization of Economic Cooperation and Development (OECD) and International Committee on Harmonization (ICH) guidelines for the mammalian erythrocyte micronucleus (MN) test, analysis of peripheral blood reticulocytes (RETs) for the presence of micronuclei can be performed using flow cytometry. The MicroFlow PLUS method (Litron Laboratories, Rochester, NY) for MN analysis by flow cytometry is based on the binding of FITC-labeled antibodies to the CD71 transferrin receptor of immature RETs, on parallel RNA degradation, and on propidium iodide staining of DNA present as micronuclei. The objective of this study was to assess the sensitivity of this flow cytometry method to detect time- and dose-dependent induction of micronuclei in mouse peripheral blood RETs after treatment with nine chemical agents. Five known clastogens, two known aneugens, and two compounds previously reported to be inactive in the mouse bone marrow MN test were evaluated at three dose levels. Multiple blood sampling of the same animal before and at two time points after treatment was conducted. All known mutagens produced a dose-dependent increase in micronucleated reticulocytes (MN-RETs); the compounds previously shown to be inactive in the in vivo MN test were also negative using the present methodology. The highest frequency of MN-RETs was observed at 48 hr after treatment, except for 5-fluorouracil, which had its peak response at 72 hr. The results indicate that micronuclei can be measured by multiple blood sampling of the same animal before and after treatment without altering the sensitivity of the assay. The results confirm that the flow cytometric assessment of MN-RETs in mouse peripheral blood using MicroFlow PLUS is a sensitive method with high analysis throughput, and robust quality control.