For regulatory purposes prenatal developmental toxicity (OECD No. 414) studies are routinely performed in our laboratories. The suitability of metabolomics as technology to identify maternal toxicity in such studies was investigated. Plasma was sampled from pregnant, non-fasted rats on gestation day 20 before cesarean section. Metabolite profiling was performed by gas- and liquid-chromatography-tandem mass spectrometry techniques. The sensitivity of routinely examined maternal toxicity parameters (OECD No. 414) was compared to those of metabolome analysis. Evaluating 44 studies, the metabolome-derived NOEL was more sensitive in 45% of the cases in detecting maternal toxicity than the maternal NOAEL. Metabolome patterns indicative for liver effects and 4-hydroxyphenylpyruvate dioxygenase (HPPD) enzyme-inhibition were established in pregnant rats based on regulated metabolites using reference compounds. The HPPD inhibition and liver toxicity patterns in pregnant rats were reasonably comparable to the ones established in non-pregnant, fasted rats. Metabolomics is a useful tool for an improved and mechanism-based identification of maternal toxicity in maternal and prenatal toxicity studies. The data suggest that the current classical maternal toxicity parameters may underestimate the extent of effects of compounds on the dams.

The intestinal microbiota contributes to the metabolism of its host. Adequate identification of the microbiota's impact on the host plasma metabolites is lacking. As antibiotics have a profound effect on the microbial composition and hence on the mammalian-microbiota co-metabolism, we studied the effects of antibiotics on the "functionality of the microbiome"-defined as the production of metabolites absorbed by the host. This metabolomics study presents insights into the mammalian-microbiome co-metabolism of endogenous metabolites. To identify plasma metabolites related to microbiome changes due to antibiotic treatment, we have applied broad-spectrum antibiotics belonging to the class of aminoglycosides (neomycin, gentamicin), fluoroquinolones (moxifloxacin, levofloxacin) and tetracyclines (doxycycline, tetracycline). These were administered orally for 28 days to male rats including blood sampling for metabolic profiling after 7, 14 and 28 days. Fluoroquinolones and tetracyclines can be absorbed from the gut; whereas, aminoglycosides are poorly absorbed. Hippuric acid, indole-3-acetic acid and glycerol were identified as key metabolites affected by antibiotic treatment, beside changes mainly concerning amino acids and carbohydrates. Inter alia, effects on indole-3-propionic acid were found to be unique for aminoglycosides, and on 3-indoxylsulfate for tetracyclines. For each class of antibiotics, specific metabolome patterns could be established in the MetaMap®Tox data base, which contains metabolome data for more than 550 reference compounds. The results suggest that plasma-based metabolic profiling (metabolomics) could be a suitable tool to investigate the effect of antibiotics on the functionality of the microbiome and to obtain insight into the mammalian-microbiome co-metabolism.

New technologies, such as metabolomics, can address chemical grouping and read across from a biological perspective. In a virtual case study, we selected MCPP as target substance and MCPA and 2,4-DP as source substances with the goal to waive a 90-day study with MCPP. In order to develop a convincing case to show how biological data can substantiate read across, we used metabolomics on blood samples from the 28-day studies to show the qualitative and quantitative similarity of the substances. The 28-day metabolome evaluation of source substances and the target substance indicate liver and kidneys as target organs. 2,4-DP was identified as the best source substance. Using the information of the 90-day 2,4-DP study, we predicted MCPP's toxicity profile at 2500 ppm: reduced food consumption and body weight gain, liver and kidney weight increases with clinical-pathology changes and a moderate red blood cell parameter reduction. NOEL prediction for MCPP was below that of 2,4-DP (<500 ppm), and similar to that of MCPA (≥150 ppm). Qualitatively, these predictions are comparable to the results of the real MCPP 90-day study in rats (reduced food consumption and body weight gain, weight increases and clinical-pathology changes in liver and kidneys, reduced red blood cells values). Quantitatively, the predicted NOAEL (150 ppm) is similar to the actual study (NOEL = 75 ppm, NOAEL ≤ 500 ppm). Thus, the 90-day rat toxicity study of MCPP could have been waived and substituted by the 90-day results of 2,4-DP by using metabolome data of 28 day studies.

The MetaMap(®)-Tox database contains plasma-metabolome and toxicity data of rats obtained from oral administration of 550 reference compounds following a standardized adapted OECD 407 protocol. Here, metabolic profiles for aniline (A), chloroform (CL), ethylbenzene (EB), 2-methoxyethanol (ME), N,N-dimethylformamide (DMF) and tetrahydrofurane (THF), dosed inhalatively for six hours/day, five days a week for 4 weeks were compared to oral dosing performed daily for 4 weeks. To investigate if the oral and inhalative metabolome would be comparable statistical analyses were performed. Best correlations for metabolome changes via both routes of exposure were observed for toxicants that induced profound metabolome changes. e.g. CL and ME. Liver and testes were correctly identified as target organs. In contrast, route of exposure dependent differences in metabolic profiles were noted for low profile strength e.g. female rats dosed inhalatively with A or THF. Taken together, the current investigations demonstrate that plasma metabolome changes are generally comparable for systemic effects after oral and inhalation exposure. Differences may result from kinetics and first pass effects. For compounds inducing only weak changes, the differences between both routes of exposure are visible in the metabolome.

The identification of the no observed adverse effect level (NOAEL) is the key regulatory outcome of toxicity studies. With the introduction of "omics" technologies into toxicological research, the question arises as to how sensitive these technologies are relative to classical regulatory toxicity parameters. BASF SE and metanomics developed the in vivo metabolome database MetaMap®Tox containing metabolome data for more than 500 reference compounds. For several years metabolome analysis has been routinely performed in regulatory toxicity studies (REACH mandated testing or new compound development), mostly in the context of 28 day studies in rats (OECD 407 guideline). For those chemicals for which a toxicological NOAEL level was obtained at either high or mid-dose level, we evaluated the associated metabolome to investigate the sensitivity of metabolomics versus classical toxicology with respect to the NOAEL. For the definition of a metabolomics NOAEL the ECETOC criteria (ECETOC, 2007) were used. In this context we evaluated 104 cases. Comparable sensitivity was noted in 75% of the cases, increased sensitivity of metabolomics in 8%, and decreased sensitivity in 18% of the cases. In conclusion, these data suggest that metabolomics profiling has a similar sensitivity to the classical toxicological study (e.g. OECD 407) design.

While conventional parameters used to detect hepatotoxicity in drug safety assessment studies are generally informative, the need remains for parameters that can detect the potential for hepatotoxicity at lower doses and/or at earlier time points. Previous work has shown that metabolite profiling (metabonomics/metabolomics) can detect signals of potential hepatotoxicity in rats treated with doxorubicin at doses that do not elicit hepatotoxicity as monitored with conventional parameters. The current study extended this observation to the question of whether such signals could be detected in rats treated with compounds that can elicit hepatotoxicity in humans (i.e., drug-induced liver injury, DILI) but have not been reported to do so in rats. Nine compounds were selected on the basis of their known DILI potential, with six other compounds chosen as negative for DILI potential. A database of rat plasma metabolite profiles, MetaMap(®)Tox (developed by metanomics GmbH and BASF SE) was used for both metabolite profiles and mode of action (MoA) metabolite signatures for a number of known toxicities. Eight of the nine compounds with DILI potential elicited metabolite profiles that matched with MoA patterns of various rat liver toxicities, including cholestasis, oxidative stress, acetaminophen-type toxicity and peroxisome proliferation. By contrast, only one of the six non-DILI compounds showed a weak match with rat liver toxicity. These results suggest that metabolite profiling may indeed have promise to detect signals of hepatotoxicity in rats treated with compounds having DILI potential.

For identification of toxicological modes of action (MoAs) a database (MetaMap®Tox) was established containing plasma metabolome consisting of approximately 300 endogenous metabolites. Each five male and female Wistar rats per groups were treated with >500 reference compounds over a period of 28 days. More than 120 specific toxicity patterns of common metabolite changes associated with unique MoAs were established.To establish patterns predictive effects on the thyroid, animals have been treated with reference compounds directly acting on the thyroid hormone formation (such as methimazole, ethylenethiourea) as well as liver enzyme inducers leading to an increased excretion of thyroid hormones and therewith to a secondary response of the thyroid (such as aroclor 1254 and boscalid). Here we present the plasma metabolite changes which form the patterns for direct and indirect effects on the thyroid. It is possible to identify metabolites which are commonly regulated irrespective of an indirect or direct effect on the thyroid as well as groups of metabolites separating both MoAs. By putting the metabolite regulations in the context of affected pathways helps to identify thyroid hormone inhibiting MoAs even when the hormone levels are not consistently changed. E.g., direct thyroid hormone synthesis inhibitors affect some enzymes in the urea cycle, increase the ω-oxidation of fatty acids and decrease glutamate and oxoproline levels, whereas indirect thyroid hormone inhibiting compounds interact with the lipid mediated and liver metabolism.

Addressing safety concerns such as drug-induced kidney injury (DIKI) early in the drug pharmaceutical development process ensures both patient safety and efficient clinical development. We describe a unique adjunct to standard safety assessment wherein the metabolite profile of treated animals is compared with the MetaMap Tox metabolomics database in order to predict the potential for a wide variety of adverse events, including DIKI. To examine this approach, a study of five compounds (phenytoin, cyclosporin A, doxorubicin, captopril, and lisinopril) was initiated by the Technology Evaluation Consortium under the auspices of the Drug Safety Executive Council (DSEC). The metabolite profiles for rats treated with these compounds matched established reference patterns in the MetaMap Tox metabolomics database indicative of each compound’s well-described clinical toxicities. For example, the DIKI associated with cyclosporine A and doxorubicin was correctly predicted by metabolite profiling, while no evidence for DIKI was found for phenytoin, consistent with its clinical picture. In some cases the clinical toxicity (hepatotoxicity), not generally seen in animal studies, was detected with MetaMap Tox. Thus metabolite profiling coupled with the MetaMap Tox metabolomics database offers a unique and powerful approach for augmenting safety assessment and avoiding clinical adverse events such as DIKI.

For identification of toxicological modes of action (MoAs) a database (MetaMap(®)Tox) was established containing plasma metabolome consisting of approximately 300 endogenous metabolites. Each five male and female Wistar rats per groups were treated with >500 reference compounds over a period of 28 days. More than 120 specific toxicity patterns of common metabolite changes associated with unique MoAs were established. To establish patterns predictive effects on the thyroid, animals have been treated with reference compounds directly acting on the thyroid hormone formation (such as methimazole, ethylenethiourea) as well as liver enzyme inducers leading to an increased excretion of thyroid hormones and therewith to a secondary response of the thyroid (such as aroclor 1254 and boscalid). Here we present the plasma metabolite changes which form the patterns for direct and indirect effects on the thyroid. It is possible to identify metabolites which are commonly regulated irrespective of an indirect or direct effect on the thyroid as well as groups of metabolites separating both MoAs. By putting the metabolite regulations in the context of affected pathways helps to identify thyroid hormone inhibiting MoAs even when the hormone levels are not consistently changed. E.g., direct thyroid hormone synthesis inhibitors affect some enzymes in the urea cycle, increase the ω-oxidation of fatty acids and decrease glutamate and oxoproline levels, whereas indirect thyroid hormone inhibiting compounds interact with the lipid mediated and liver metabolism.

Read-across and grouping is one of the most commonly used alternative approaches for data gap filling in registrations submitted under the REACH Regulation as defined by the European Chemicals Agency (ECHA) in their ‘Read-Across Assessment Framework’ (RAAF, 2017). At the same time, the application of read-across is rejected by ECHA frequently due to various reasons. As a major reason hereof, applicants fail to reduce the level of ‘remaining uncertainty’ intrinsical to every read-across approach compared to testing a substance experimentally. Recently, the use of metabolomics to support read-across cases with biological information has been reported in a case study with phenoxy herbicides (Ravenzwaay et al., 2016). In the present case-study a ‘weight-of-evidence’ read-across approach from 2-aminoethanol (MEA = ‘source’) to 3-aminopropanol (3AP = ‘target’) with metabolomics as ‘supporting evidence’ reducing the remaining uncertainties is reported. We demonstrate the high structural similarity of the two analogous substances based on the available data and we report how metabolome data add confidence concerning mechanistic similarity in this read-across approach. Finally, the herein described read-across case supported by metabolomics is used to cover the data gaps in repeated dose and reproductive toxicity endpoint of 3AP via weight of evidence for the REACH-registration.

BASF has developed a rat plasma metabolomics database (MetaMap®Tox) containing the metabolome of more than 500 chemicals, agrochemicals and drugs, for which the toxicity is well known, derived from 28-day repeated dose toxicity studies in rats. The quality/reproducibility of data was assessed by comparing the metabolome of 16 reference compounds tested at least twice under identical experimental conditions at three time points (day 7, day 14 and day 28). Statistical correlation analysis showed that the repeated treatment induced very similar changes to the metabolome. For all repetitions the modes of action of the compounds were always correctly identified. Moreover, when compared against the metabolome of all compounds available in the MetaMap®Tox database, the repetitions showed in most cases the highest degree of overall similarity with the metabolome of the original study. In addition, we also evaluated the robustness of our metabolomics technique, displayed by constancy of variability in control groups over time. Based on these results, it can be concluded, that metabolomics can reproducibly be applied during toxicological in vivo testing in rats under the conditions applied here.

BASF has developed a Metabolomics database (MetaMap(®) Tox) containing approximately 500 data rich chemicals, agrochemicals and drugs. This metabolome-database has been built based upon 28-day studies in rats (adapted to OECD 407 guideline) with blood sampling and metabolic profiling after 7, 14 and 28 days of test substance treatment. Numerous metabolome patterns have been established for different toxicological targets (liver, kidney, thyroid, testes, blood, nervous system and endocrine system) which are specific for different toxicological modes of action. With these patterns early detection of toxicological effects and the underlying mechanism can now be obtained from routine studies. Early recognition of toxicological mode of action will help to develop new compounds with a more favourable toxicological profile and will also help to reduce the number of animal studies necessary to do so. Thus this technology contributes to animal welfare by means of reduction through refinement (2R), but also has potential as a replacement method by analyzing samples from in vitro studies. With respect to the REACH legislation for which a large number of animal studies will need to be performed, one of the most promising methods to reduce the number of animal experiments is grouping of chemicals and read-across to those which are data rich. So far mostly chemical similarity or QSAR models are driving the selection process of chemical grouping. However, "omics" technologies such as metabolomics may help to optimize the chemical grouping process by providing biologically based criteria for toxicological equivalence. "From QSAR to QBAR" (quantitative biological activity relationship).

BASF and Metanomics have built-up the database MetaMap(®)-Tox containing rat plasma metabolome data for more than 500 reference compounds. Phenytoin was administered to five Wistar rats of both sexes at dietary dose levels of 600 and 2400 ppm over 28 days and metabolome analysis was performed on days 7, 14 and 28. Clinical pathology did not indicate clear evidence for liver toxicity, whereas liver histopathology revealed slight centrilobular hepatocellular hypertrophy. The metabolome analysis of phenytoin shows metabolome changes at both dose levels and the comparison with MetaMap-Tox indicated strong evidence for liver enzyme induction, as well as liver toxicity. Moreover, evidence for kidney and indirect thyroid effects were observed. This assessment was based on the metabolite changes induced, similarities to specific toxicity patterns and the whole metabolome correlation within MetaMap-Tox. As compared with the classical read-out, a more comprehensive picture of phenytoin's effects is obtained from the metabolome analysis, demonstrating the added value of metabolome data in preclinical/ toxicological studies.

Combination therapies with fibrates and statins are used to treat cardiovascular diseases, because of their synergistic effect on lowering plasma lipids. However, fatal side-effects like rhabdomyolysis followed by acute renal necrosis sometimes occur. To elucidate biochemical changes resulting from the interaction of fibrates and statins, doses of 100 mg/kg fenofibrate, 50mg/kg clofibrate, 70 mg/kg atorvastatin and 200 mg/kg pravastatin as well as combinations thereof were administered to Crl:Wi(Han) rats for 4 weeks. Plasma metabolome profile was measured on study days 7, 14 and 28. Upon study termination, clinical pathology parameters were measured. In a separate experiment plasmakinetic data were measured in male rats after 1 week of drug administration in monotherapy as well as in combinations. Lowering of blood lipid levels as well as toxicological effects, like liver cell degradation (statins) and anemia (fibrates) and distinct blood metabolite level alterations were observed in monotherapy. When fibrates and statins were co-administered metabolite profile interactions were generally underadditive or at the utmost additive according to the linear mixed effect model. However, more metabolite levels were significantly altered during combination therapy. New effects on the antioxidant status and the cardiovascular system were found which may be related to a development of rhabdomyolysis. Accumulation of drugs during the combination therapy was not observed.