According to its aim of being a leading source of summarized knowledge of structural alerts for reactive metabolite formation, the expert database SpotRM provides several attractive features:
- A comprehensive set of structural alerts, currently numbering 75, for reactive metabolite (RM) formation that you can match your planned test compounds against.
- Facile and unprecedented access to the most relevant background information on RM mechanisms without you having to sift through piles of scientific articles; currently, there are 140+ short summaries/monographs and seven micro-reviews. This readily available material should vastly strengthen your knowledge base for prioritizing test substances.
- Continuous updates regarding new alerts, drugs, experimental compounds, and new references keep you at the knowledge front-line.
Designed as a didactic tool, SpotRM with a solid knowledge base will hopefully lead to a deeper understanding of potential toxicity risks with planned test compounds and thus result in a more efficient lead selection and optimization. Background is further discussed in this perspective paper: A Claesson & A Minidis. Chem. Res. Toxicol. 2018.
We caution that users of this tool should appreciate the large limitations of our current understanding of structure-toxicity in the area of RMs. One should be careful not to under-estimate the difficulties involved. This regards several aspects: which reactive species are actually formed, the ways of their formation, and in particular what linkage there is of a certain transformation to the observed adverse effects. To add to this, it is generally agreed that many drugs have side-effects caused by multiple types of metabolic conversions. Most importantly, many of the adverse effects mediated via RMs are idiosyncratic and hence unpredictable. The majority of this type of adverse reactions appears to be immune mediated.
A few general considerations regarding drugs with toxicity issues emanating from RMs are found in the end of this document.
Classification of drugs is based on perceived severity of adverse drug effects
In the SpotRM database all alerts are linked to selected examples of drugs that have the alert in their structures. The drug examples have been chosen to reflect in a didactic way on a certain structural alert while avoiding too much speculation on alternatives. Structures with several potential structural alerts are therefore largely avoided although we do not hesitate to propose hypotheses. For a rapid visual impression of the hazard associated with a given alert, the drug examples are classified as Red, Yellow, Green or Neutral. Definitions of these classes are:
The drug has shown clinical adverse effects that have a proven or highly probable association with bioactivation to RMs or, regarding preclinical compounds, experiments have shown extensive formation of RMs.
A drug used clinically has been withdrawn or bears regulatory warnings.
Examples: sudoxicam, felbamate, amodiaquine, sitaxentan, lamotrigine and halothane.
The drug contains an alert included in SpotRM, and there have been some reports of adverse effects that have been discussed in terms of RM formation.
Preclinical compounds have displayed RM formation.
Examples: mirtazapine, paroxetine, atorvastatin, tolmetin and phenazone.
Despite having triggered an RM alert, the drug has been used clinically without reported findings of adverse effects that can be associated with RM formation. The explanations for vindication typically include low dosage or a very low degree of metabolism that involves the relevant substructure. The last-mentioned case usually depends on metabolism in other parts of the molecule.
Examples: rivaroxaban, cefuroxime, aripiprazole and tolterodine.
In spite of having a RM alert in the structure the clinical information is insufficient to classify the drug into any of the other categories. This might be because the drug has only recently been introduced or is not widely used, and therefore sufficient safety data have not accumulated. For experimental compounds the relevant in vitro experiments might not have been run.
Examples: formoterol, fluorofelbamate and ranolazine.
This is a new feature since April 2021 that is in a testing phase and will be evaluated in due course. Eventually, it might not be deemed to meet our users’ or our own value criteria.
In short, every structural alert is assigned a severity value of 1 to10 based on our reading of the literature and authorization documentation for the drugs most relevant to the specific alert. Obviously, there is a rather wide scope for discussions here due to the many unknowns involved. However, the intention is simple in its purpose: to label a structural alert with a high score when it, in the metabolic process, often has proven to result in extensive conversion(s) to its reactive intermediate(s) leading to repeated cases of adverse events in patients. A typical example is a primary benzeneamine (aniline), which gives rise to nitroso intermediates that react with proteins thus causing allergic reactions (an antibacterial sulphonamide being a good example). In the other end of the spectrum one might put a phenol that when further oxidized gives rise to reactive benzoquinones. However, phenols are often metabolically inactivated by efficient sulfation and glucuronidation processes thus mitigating the conversion to RMs and justifying a low score. (The phenol alert in SpotRM is required to have a free ortho position). A middle point in the severity spectrum might be represented by a thiazolidinedione ring, which is oxidized to several non-characterized reactive species. This alert has not been much investigated.
Bioactivation is required, chemical reactivity per se is excluded
SpotRM, with very few exceptions, focuses on drugs that require metabolic activation to form reactive intermediates. It is well-known that there are quite a number of drugs on the market that give rise to toxic effects, including hepatotoxicity and IADR, by being chemically reactive per se. This type of intrinsically reactive compounds are normally filtered away early in the drug design process but might occasionally have beneficial effects. Examples of important drugs of this class range from beta-lactams, like amoxicillin, to cytotoxic anticancer agents, for example busulfan. Intriguingly, such intrinsic reactivity has since several years been a focus of new kinase inhibitors used for cancer.
Cytotoxic anticancer agents are largely neglected
A large number of anticancer agents require metabolic activation to achieve cytotoxic therapeutic activity, a principle that entails many well-known and feared side-effects. For the purpose of SpotRM, which is to help spotting hidden alerts in “normal” drugs, this kind of overt toxicity is of minor interest and therefore the majority of these drugs, for example dacarbazine and cyclo-phosphamide, have been excluded. Nevertheless, some anticancer agents may well offer guidance regarding what safety-focused medicinal chemists ought to keep out of their drug design tool-box. We have tried to incorporate some of this type of learning in SpotRM.
A comment on irreversible CYP enzyme inhibition
Many drugs interfere with the metabolism of other drugs, causing drug-drug interactions. This can take place in several ways, but the most common mechanism, which is also relevant in the current context, is irreversible inhibition of a CYP enzyme that is responsible for metabolizing another drug. The most common mechanism of such inhibition is via a reactive species that is formed in situ at the enzyme’s active site, also called mechanism-based inhibition (MBI). The broader concept of time-dependent inhibition (TDI) can also occur via other mechanisms. Many drugs that have a reported significant evidence of MBI are included in SpotRM, since effects collateral with RMs formed in situ most often cause other types of toxicity. However, there is a need to systematically include even more drugs that inhibit CYPs; future updates of SpotRM will gradually fill the gap in order to ensure comprehensiveness.
Selection of alerts – focus on the less well-known
It is a delicate task to select the really useful substructures (RM alerts), which will be represented by a well-written SMARTS string for matching in target structures. The dilemma can be illustrated by including a string that corresponds to a benzene ring with nitrogen attached to it. This would encompass very many drugs. We have instead chosen to split this substructure class into many smaller classes; a few examples are indoles, o-methyl-benzeneamines, N-acyl-benzeneamines, benzimidazoles, and more. In this way we can fine-tune the examples that are most appropriate. The user can still search using as input benzene with attached nitrogen but as result may have to handle up to a hundred or more hits.
SpotRM is not complete regarding reported examples of RM formation: the more widespread the knowledge of the hazard of a certain substructure is the less is the marginal gain to the user of having it included. Many alerts in SpotRM are of this type, for example a nitroarene, which is more or less taboo in drug candidates of today though we choose to include several examples of this for the sake of reasonable comprehensiveness. Regarding the number of examples of drugs (in the red/yellow/green/neutral scale) we find less value in listing in SpotRM all the drugs that are anilines (including “hidden” anilines), which have displayed hepatotoxicity. Besides, many of these are hugely obsolete. On the other hand, safe anilines and an explanation as to why they may be considered safe are certainly within the scope of SpotRM.
In line with the notion of SpotRM not being comprehensive: the exploding drug class of kinase inhibitors contains many dubious members (from a RM perspective). Even new ones tend to exhibit clear structural alerts. We have not given full coverage of these compounds since kinase inhibitors are also fraught with complicating pharmacology.
To increase the usefulness of SpotRM we have made considerable efforts to include as alerts substructures that have not been mentioned explicitly as such in the literature but probably have contributed to adverse effects. In particular, our selection of alerts focuses on the less well-studied hazards of certain combinations of ring systems and substituents. Here you may find, for example a range of substructures that ultimately result in various types of activated benzylic alcohols such as derivatives of 2- or 4-methyl-benzeneamine, or five-membered heterocycles having alkyl groups that can be oxidised to (benzylic type) alcohols.
Since there is a huge lack of exact knowledge of how adverse effects might be associated with RM formation, we have been generous with hints of potential compound liabilities and offer many hypotheses regarding less obvious reaction pathways based on sound chemical reasoning and guessing of routes of metabolism. Our sincere hope is that our users will be involved in the continuous improvement of SpotRM by communicating their points of view to us.
Delimiting an alert/toxicophore is a delicate task as well. We hope to work together with the users of SpotRM in a process to refine the SMARTS definitions and the linking of alerts to the most illustrative example drugs.
General considerations regarding drugs with toxicity issues from RMs
The number of drugs that have been on the market and then withdrawn due to toxicity associated with their metabolism is frustratingly large, well over one hundred. The dominant problem here is metabolic activation to form reactive metabolites leading to modifications of proteins and DNA, or to inexpedient influence on oxidation processes. Since this kind of toxicity exhibited by candidate drugs usually leads to termination of their development, early identification of compounds that likely exhibit a risky metabolic conversion cannot be stressed enough (cf. reviews collated here). It is our hope and belief that SpotRM can contribute in facilitating the process of paying proper attention to these drug structures.
The precise biochemical mechanisms of adverse events of many drugs, in particular no longer-marketed ones, have rarely been investigated. However, in many cases involvement of RMs might be considered when the clinical symptoms involve:
- Hepatotoxicity (drug-induced liver injury, DILI)
- Effects on blood cells (blood dyscrasias), e.g. agranulocytosis
- Anaphylaxis and hypersensitivity reactions, which are immune-mediated; they most often manifest as cutaneous reactions
- Mitochondrial toxicity
- Drug-drug interaction caused by time-dependent inhibition of cytochrome P450 enzymes (via mechanism-based inhibition, MBI, a subgroup of time-dependent inhibition, TDI)
Furthermore, many unknowns are involved in estimating the clinical consequences of RM formation since the reasons for discontinuation of clinical trials are often unknown to outsiders. Such terminated trials also naturally produce limited amounts of safety data. Additionally, it is important to keep in mind that experts generally consider side-effects of licensed drugs to be underreported.
In SpotRM we try to draw examples of drugs from the first four of the above-mentioned types of clinical adverse effects. Naturally, our conclusions on drugs where the literature is scarce (as hinted, a very common situation) should be regarded as provisional hypotheses.
Another growing set of compounds consists of those described in the medicinal chemistry/ pharmacology literature with in vitro data on adduct formation with trapping agents, usually glutathione. Illustrative examples of such compounds are included where pertinent. Updates of SpotRM will include even more examples of this type.
Abbreviations used within SpotRM+
ADR, adverse drug reaction; AE, adverse event; CYP, cytochrome P450; DILI, drug induced liver injury; DRESS, Drug reaction with eosinophilia and systemic symptoms; EMA, European Medicines Agency; FDA, Food and Drug Administration (USA); GSH, glutathione; GST, glutathione transferase; HLM, human liver microsomes; IADR, idiosyncratic adverse drug reaction; MAO, monoamine oxidase; MBI, mechanism-based inhibition; SULT, sulfotransferase; TDI, time-dependent inhibition; ULN, upper level of normal.