Triptans, Serotonin Agonists, and Serotonin Syndrome (Serotonin Toxicity): A Review
Conflicts of Interest: None
Dr P. Ken Gillman, Postal PO Box 86, Bucasia 4750, Queensland, Australia.
The US Food and Drug Administration (FDA) have suggested that fatal serotonin syndrome (SS) is possible with selective serotonin reuptake inhibitors (SSRIs) and triptans: this warning affects millions of patients as these drugs are frequently given simultaneously. SS is a complex topic about which there is much misinformation. The misconception that 5-HT1A receptors can cause serious SS is still widely perpetuated, despite quality evidence that it is activation of the 5-HT2A receptor that is required for serious SS. This review considers SS involving serotonin agonists: ergotamine, lysergic acid diethylamide, bromocriptine, and buspirone, as well as triptans, and reviews the experimental foundation underpinning the latest understanding of SS. It is concluded that there is neither significant clinical evidence, nor theoretical reason, to entertain speculation about serious SS from triptans and SSRIs. The misunderstandings about SS exhibited by the FDA, and shared by the UK Medicines and Healthcare products Regulatory Agency (in relation to methylene blue), are an important issue with wide ramifications.
US Food and Drug Administration
human cloned receptor
lysergic acid diethylamide
monoamine oxidase inhibitor
(selective) serotonin reuptake inhibitor
The purpose of this review is to elucidate the possible role of triptans in the causation of serious “serotonin mediated CNS morbidity” when co-administered with serotonergic drugs (selective serotonin reuptake inhibitor [SSRI] or serotonin and noradrenaline reuptake inhibitor antidepressants). “Serotonin mediated morbidity” is now often referred to as serotonin toxicity (ST), and also as serotonin syndrome (SS). SS with triptans is a subject about which there has been debate and comment,1-4 as the US Food and Drug Administration (FDA) alert suggesting that fatal SS is possible when triptans and certain antidepressants are coadministered,5,6 and Evans's subsequent publication of the case histories of the 29 cases, not originally provided by the FDA, but obtained via a Freedom of Information Act request.7,8
This review focuses on the relationships between the 3 classes of drugs known to be involved in severe SS, and the relative degrees of toxicity they characteristically precipitate. A key point to be emphasized is the spectrum concept of SS. SS is a synaptic serotonin (5-HT) concentration-related phenomenon.9 Readers are referred to recent selected reviews of SS for a broader perspective. There have been advances in the quantification of the frequency and severity of SS with different drugs,9-14 in the definition of SS in animals and humans,15-17 in the pathophysiology,15 in the clinical presentation,18 and in its management and treatment.19-21
Serotonin syndrome can be diagnosed with accuracy and confidence, but few reports classify it using recognized diagnostic criteria, hence diminishing their value. It is not, as far as current human and animal evidence indicates, an idiosyncratic response, but a predictable and inevitable result of toxicity (mediated via the final common pathway of elevated intra-synaptic serotonin). If a case history were to appear reporting SS following an overdose of vitamin C, it would be parsimonious to assume that there had been a failure to ascertain or recognize the simultaneous ingestion of a potently serotonergic drug, cf. the Stanford case report.22 It would not be logical to make an initial assumption that vitamin C had previously unknown serotonergic properties, especially as we have good reasons to predict that it does not affect serotonin. That is Bayesian reasoning, ie, considering the prior probability when estimating the likelihood of an outcome. Without such frameworks of knowledge and understanding of SS, case reports are often difficult to interpret, and the type of information they can yield reliably requires cautious consideration.23 The uncertainty and debate surrounding triptans demonstrates this problem clearly. The FDA alert was based on case reports, most of them informal, or “second-hand” and not peer-reviewed, and interpreted with an imperfect notion of the symptoms and pathophysiology of SS, and without using validated criteria to establish diagnoses (eg, the Hunter Serotonin Toxicity Criteria [HSTC]17,20).
The HSTC demonstrate unequivocally that clonus is the single most important sign required to diagnose SS, a fact that has now been established for many years, yet case reports of SS rarely, if ever, document the presence, or absence, of this sign. When such key information is lacking, little credence can be given to many reports. Case reports constitute a low grade of evidence, but they command undue attention and are repeatedly cited, even when they have been firmly rebutted (for just such an error that occurs in the FDA case reports, see24,25).
The 3 classes of therapeutic drugs that, in certain combinations at usual doses, have been reliably documented to be capable of precipitating severe SS are: monoamine oxidase inhibitors (MAOIs), SRIs, and releasers. There are other drugs and combinations that might possibly, or rarely, produce severe SS; but in usual treatment settings involving therapeutic drugs and doses, only an SSRI+an MAOI is likely (or has been reliably documented) to produce severe SS. As SSRIs by themselves, even if taken in overdose, do not precipitate severe SS, the occurrence of severe SS in patients on an SSRI means that an MAOI is likely to have been coingested.9,26
Concern about copharmacy resulting in SS makes it logical and helpful to consider combinations including not only the triptans, but also other serotonin receptor agonists such as the related indolealkylamines, including ergotamine, bromocriptine, lysergic acid diethylamide (LSD), lisuride, pergolide, and buspirone.27,28 It is useful to revisit similar false positive reports and speculations previously published about SS with other drugs, such as medications we now know are not serotonergic and cannot precipitate SS, for example, mirtazapine.10
The risk of serious morbidity and death in SS is from hyperthermia which is mediated in a dose-related manner by the action of 5-HT or 5-HT agonists, on 5-HT2A receptors, and is ameliorated, or prevented, by 2A antagonists such as cyproheptadine, but not by 1A, antagonists.14-16,18,21,29,30 The degree of 5-HT elevation required for toxicity is of the order of 10-50 times above the baseline level9 and drugs such as mirtazapine or nefazodone, which are incapable of engendering elevations of even twice the baseline level, simply cannot cause SS, any more than can vitamin C.
SEARCH STRATEGY AND SELECTION CRITERIA
The author (P.K.G.) has maintained a database of all cases identified relating to hyperthermia and SS for 15 years. References for this review were identified by searches of his own database and PubMed from 1966 until February 2009 with multiple terms including: “hyperthermia,”“pyrexia,”“temperature,”“serotonin syndrome,”“serotonin toxicity,”“toxicity,”“triptans,”“individual drug names,” and by hand searches of the bibliographies of these papers. All papers published in all languages were accessed for review, whether reported as SS, ST, or central nervous system (CNS) toxicity, or where considered by this author to be relevant.
CRITERIA FOR SS
Serotonin syndrome is a well-delineated and discrete toxic syndrome (often referred to as a toxidrome). The HSTC are validated criteria used to define SS and have been validated in over 2000 overdoses of SSRIs, and other serotonergic drugs, during extensive clinical use by experienced toxicologists, and are highly sensitive (84%) and specific (97%) for SS.17 As stated in the introduction, the term ST is preferable to SS and should be confined to more severe cases, as inferred by the term toxicity (ie, poisoning, in contrast to side effects). It is not helpful or logical to use the terms SS (or ST) to describe what are in fact typical and usual side effects of therapeutic doses of drugs. This practice has led to the absurd situation of innumerable reports of “SS” with drugs (or combinations) that have never been documented to cause severe serotonergic symptoms or fatalities.
These reports have been too frequently accepted because there is a generally insufficient understanding of the spectrum concept of SS, which explains the progression of dose-related effects that result from synaptic serotonin concentration-related phenomena.9 The spectrum concept provides both a framework for understanding the subject, and the disparate literature, and makes testable (and proven) predictions about the frequency and severity of toxicity. There will be only rare exceptions to the statement that case reports will not contribute much more to the literature, because large prospective case series (eg, Professor Whyte's data) provide much more reliable evidence concerning the typical picture of toxicity with individual drugs, with the additional advantage that the signs and symptoms have been consistently collected and evaluated by an experienced team of toxicologists.
Ergots are indolealkylamines, mostly derived directly from ergotamine and its many natural analogs.27,28 There seem to be no cases of typical or hyperthermic SS resulting from ergotamine in the literature. Mathew's cases31 related to IV use of dihydroergotamine and were atypical (and rightly reported as only “suggestive” of SS) and do not meet the HSTC for SS.
Eadie reviewed “Convulsive ergotism: epidemics of the serotonin syndrome,”32 in which he suggests that SS “. . . may therefore have been a public-health problem long before it was recognized as a complication of modern psychopharmacology.” The features, described from old sources (particularly the studies by Barger and Tissot33,34), were: “. . . distortion of the trunk and limbs, painful involuntary flexion of the fingers and wrists, . . . drowsy, sometimes delirious, lethargic, and melancholic or manic and [patients] could have hallucinations and double vision. Some affected individuals developed profuse sweating, fever, muscle stiffness, and twitching. . . . It seems likely that some of the involuntary postures would now be classified as torsion dystonia or other dyskinesias, . . . .” It is known that ergot alkaloids also have dopaminergic properties, which are more likely to be the cause of these signs and symptoms.35
Lysergic acid diethylamide is a potent compound; the typical active dose is between only 0.05 and 0.20 mg in humans. It does not seem to be associated with serotonergic symptoms or features of SS, either in “usual” doses or in “over-doses.” There appear to be no cases of SS in more than 50 years of use, and Fantegrossi states, “Indeed, there are no documented cases of death due to LSD overdose, . . . .”36 The absence of mortality or documented SS with LSD is particularly interesting because it is probably the indolealkylamine with the greatest agonist potency at the 5-HT2A receptor.37 The only case reports of toxicity in the literature are not typical SS38,39 although one case exhibited severe hyperthermia. It has become clear recently that LSD acts as a 5-HT2A agonist exhibiting “functional selectivity,”40,41 and is a full 1A agonist. Of the various human cloned receptor (HCR) serotonin sub-types, LSD appears to have the highest inhibition constant (Ki) at the 1A receptor, closely followed by 2A, 2B, 2C, and 5B and 6 (all Kis from PDSP database (National Institute of Mental Health, Psychoactive Drug Screening Program) are between 1 and 10 nM, ie, very potent. The observation that LSD does not usually produce hyperthermia is evidence that this “functional selectivity” has real consequences in humans.
Bromocriptine is most potent as a dopamine D2 agonist (HCR Ki ∼3 nM), but less potent than pergolide or cabergoline28 and may slightly increase serotonin levels, perhaps via 5HT-1A receptors. It is approximately equipotent as an agonist at 1A and 1D receptors (HCR Ki ∼10 nM). The most current HCR affinity data may be viewed at the PDSP website. The evidence that it raises brain 5-HT is from 30 years ago, and has not been subsequently replicated.42 Cases where bromocriptine was thought to have worsened SS were misdiagnosed as neuroleptic malignant syndrome and have been reviewed.21 Any contribution of bromocriptine to SS symptoms or severity seems minor or doubtful.
Buspirone is a 5-HT1A partial agonist thought to act mainly via post-synaptic 1A receptors,43 and is thus a weakly serotonergic drug. There is no good evidence it precipitates SS despite years of coadministration with SSRIs and MAOIs. Both pre- and post-synaptic 1A receptors mediate hypothermia,15,44 and the animal “5-HT syndrome” (which does not involve hyperthermia) that these receptors mediate is quite different from human SS.15 Case reports involving buspirone are unconvincing case reports (discussed in the study by Gillman14 adequately accounted for by the actions of other coingested serotonergic drugs45-52).
Triptans are, by partly serendipitous design, agonists at 5HT1B/1D, but not at 2A (or 1A) receptors. Triptans were not initially synthesized for CNS penetration, which was thought undesirable. Receptor data on most of them are sparse.53 Measured HCR affinities at 5-HT2A receptors are thousands of times less than at 1D (Kis of ∼10 thousand nM vs ∼10 nM), and for 1A receptors hundreds of times less.54-56 There are few data on CNS penetrance, which for sumatriptan is clearly low.57 In rodents given 100 times the usual dose of naratriptan (30 mg/kg), during testing for analgesic properties, no behavioral effects relevant to SS were observed.58 Zolmitriptan shows some CNS penetrance.59
There is no identifiable basis for the FDA statements5 that triptans “increase serotonin levels,” or that there is “biological plausibility” for interactions causing SS. The contrary is the more reasoned initial presumption, namely that there is no biological plausibility for triptans to cause SS, especially because they have no agonist action at 5-HT2A receptors. Even if 5-HT1A receptors were thought to be relevant, triptans have less activity there than other widely used drugs that are not associated with SS, eg, buspirone. Also, drugs showing significant activity at CNS 5-HT1A receptors, such as buspirone, cause hypothermia, not hyperthermia.
There have been uncertain case reports published, none of which meet the HSTC for SS, and are unlikely to be SS.31,60-65 The reservations about case reports highlighted above apply. The cases described by Soldin et al suffer an even more serious disadvantage because they consisted of patient self-reports,64 which are even less reliable than doctor reports; they contain no symptom descriptions that suggest diagnosable SS.
The logical fallacy post hoc, ergo propter hoc becomes especially relevant when there are many hundreds of thousands of instances where patients have developed symptoms only rarely with commonly used drugs. The chance distribution of (close) temporal association between a random event or other drug, and a symptom, will inevitably occur, erroneously strengthening the assumption of causality.
Two prospective investigations of interactions between triptans and antidepressants were negative,66,67 as have been subsequent retrospective analyses.68,69 Shapiro and Tepper estimate that “1 million relevant patient-month exposures to the combination of triptans and SSRIs occurred during the period of the reporting of the FDA cases.”3 This degree of exposure has probably been so for a decade or more. Thus, millions of people in the world have likely taken such combinations. If triptans were capable of inducing serious SS it is likely there would be large numbers of serious cases, including deaths. It can be confidently inferred that is not the case, because it is almost inconceivable that large numbers severe cases, or deaths, could have been missed.
The 29 cases that the FDA apparently evaluated include some confusion as to which ones were published.24,31,60,61,65 It is not noted in the FDA data that case 27 is a published case,24 which was rebutted (as not being SS) by Professor Whyte, a prominent commentator on SS, and the author of the HSTC.25 Evans analyzed the FDA cases and concluded that none met the Hunter criteria for SS.7 This present analysis agrees with Evans' opinion. None of the FDA cases convincingly support their proposition.
The case with the most obviously incorrect assertion that triptans played a role in SS is case 28, “an overdose of sumatriptan, sertraline, citalopram, moclobemide,” which was indeed severe, and potentially fatal, SS. Moclobemide (an MAOI)+2 SSRIs (sertraline and citalopram) is a classic combination documented to precipitate fatal SS70: thus, to suggest that a triptan contributed to the toxicity is pure speculation. A representative example of the FDA material presented is Case 3. “Sumatriptan Unspecified SSRI. A nurse reported to a sales representative that a patient received sumatriptan and an unspecified SSRI and developed SS. Age and sex not provided.” Imprecise second-hand information, that is not even suggestive of SS, cannot be construed as scientific evidence and one is obliged to conclude, as most of the FDA data are of similar quality, that they are almost bereft of value.
There are only 2 cases in the whole series (22 and 23) that deserve attention, on the basis that an experienced clinician would want more information before giving an opinion, eg, case 22, “Sertraline lithium methysergide 6 mg of sumatriptan sc. On examination, she was dysarthric, excited, hypomanic, shivering, with dilated pupils, weakness of all limbs more pronounced on the right . . . frequent myoclonic jerking in all limbs with hemiballistic movements in the right upper extremity. She was diffusely hyperreflexic . . . ataxia of limb and gait.” Applying the Hunter ST Criteria decision rules17: rule 1, the cardinal sign of clonus was not present. Rule 2 requires “inducible clonus with agitation or diaphoresis” (negative), and rule 3 requires “other clonus with agitation” (also negative). Rule 4 requires tremor and hyperreflexia (negative) and the criteria state if these are not present, the case is “not SS”; so this case fails to meet the criteria. Also, it does not appear likely to be SS on the basis of “clinical judgment.”
It is also relevant to note that the HSTC symptoms have been defined by experienced toxicologists. The threshold for assigning pathological significance to particular signs such as hyperreflexia and agitation is likely to be different (lower) in less experienced hands, so the bias is likely to be toward including false positives (cf. nefazodone below).
Different classes of drugs exhibit distinct degrees to which they can elevate serotonin. The 3 relevant categories of drugs are: (1) releasers (eg, MDMA, 3,4-methylenedioxymethamphetamine); (2) MAOIs; and (3) SSRIs. Altering each of these mechanisms (that is, catecholamine release, breakdown, and uptake) each produces a characteristic maximum effect. Thus overdoses of SSRIs do not precipitate either severe or fatal SS, or temperature elevation beyond 38.5°C. No other classes of serotonin enhancing drugs have ever been demonstrated to produce severe SS, neither l-tryptophan, lithium, 5-HT1A antagonists or agonists, nor catechol-O-methyltransferase inhibitors.
Three types of data support such deductions: (1) the presence of serotonergic side effects at therapeutic doses; (2) serotonergic side effects or toxicity after overdose; and (3) serotonergic effects on coadministration with MAOIs. All other drugs that produce SS exhibit congruent findings in these 3 categories which, for instance, predict those tricyclic antidepressants that do, or do not, precipitate SS, for discussion and elaboration of this point see Gillman.10,71,72 There is no good evidence that triptans produce serotonergic side effects at therapeutic doses; serotonergic side effects, or toxicity after overdose; or, likewise on coadministration with MAOIs.
However, directly acting post-synaptic receptor agonists are a particular case, as all the other mechanisms are indirect and are mediated by changes in the endogenous ligand, 5-HT. The consequences of functional selectivity suggest that the inherent activity and toxicity profile of each individual agonist might be different and relevant to deducing what could result from idiosyncratic responses or overdose,73 but toxicity data are not usually published. Functional selectivity means that drugs that are equipotent ligands at the 5-HT2A receptor can have different “downstream” effects.40,74 Some clinically used dopamine agonists are also agonists at the 5-HT2A receptor, eg, lisuride and pergolide. LSD and pergolide are hallucinogenic, but lisuride is not.74 There appear to be no reports of complications involving hyperthermia with these drugs. Is it just luck that they happen to exhibit the “functional selectivity” that avoids this? One might assume that if experimental compounds did precipitate hyperthermic toxicity they would rapidly be screened out; however, without published toxicity data doubt remains.
Be that as it may, the triptans are inactive at the 2A receptor, and the evidence indicates that their very weak activity at the 1A receptor is almost certainly of no relevance or consequence. Indeed, if there were to be a convincing report of definite severe hyperthermic SS with a triptan, it would be a valuable report deserving critical evaluation. Such a case might provide insights about functional selectivity and genetic variants of receptors.75,76 It is therefore clear, in my opinion, that triptans do not pose a risk of causing severe SS for 2 reasons: (1) they do not show serotonergic side effects or toxicity by themselves or with other serotonergic drugs; (2) they do not posses the pharmacological properties that we are confident are required to mediate SS.
It is important to note that there have been clear precedents of many false-positive reports with other drugs that we know are not serotonergic and cannot precipitate SS, eg, amitriptyline, trazodone, nefazodone, and mirtazapine. These have all been analysed in detail elsewhere9,10,14,15: but there are numerous incorrect reports of supposed SS from them, so they serve to remind us of the importance of establishing the pharmacology of the drug and its ability to raise serotonin, which constitute the sine qua non for SS, and of using our knowledge of the spectrum concept of SS to predict the serotonergic potency of drugs in humans from data about their propensity to induce SS.9
There is a great contradiction between the estimate of risk, and the conclusions and recommendations, of the USA FDA and the UK Medicines and Healthcare products Regulatory Agency (MHRA) in their respective assessments of methylene blue,77 an MAOI, see Stanford,22 and triptans. On the one hand, the MHRA fail to warn of SS despite strong evidence of severe SS, whereas the FDA does warn of “fatal” SS where no substantive evidence exists. This is most likely attributable to a difficulty in comprehending the complex data in the explanatory context of the spectrum concept of SS, and of appreciating the details and the extent of SS drug-drug interactions. When such misunderstanding is perpetuated by official agencies, which are likely to be regarded by many as authorities, it becomes especially counter productive and may encourage misconceived, but expensive and distressing, legal actions against doctors.
Shapiro and Tepper repeat the frequently voiced but unnecessary concern that “. . . mild cases may evade detection.”3 It is crucial to appreciate the idea that this is not a major issue because the spectrum concept of SS clarifies that it is a dose-related phenomenon. Although it is a continuous spectrum, nevertheless, at some defined point of increasing severity (see the study by Gillman14), it becomes “toxicity” in the sense of dangerous/poisonous. Therefore, by definition, mild cases are of little, or no, consequence. In that sense mild toxicity is an oxymoron. It is likely that medical professionals are spending valuable but wasted time with patients in needless reassurance. It is also possible that the perceived inappropriateness of the FDA warnings may devalue the authority of future pronouncements.
Acknowledgments: I would like to acknowledge the assistance of Stewart J. Tepper, MD, Center for Headache and Pain, Cleveland Clinic in the preparation of this manuscript, and the expertise of my wife Isobel, who maintains the indispensable computers and programs. Ki determinations, receptor binding profiles, agonist and/or antagonist functional data, was generously provided by the National Institute of Mental Health's Psychoactive Drug Screening Program, Contract # NO1MH32004 (NIMH PDSP). The NIMH PDSP is directed by Bryan L. Roth, MD, PhD, at the University of North Carolina at Chapel Hill and Project Officer Jamie Driscol at NIMH, Bethesda, MD, USA. For experimental details refer to the PDSP website: http://pdsp.med.unc.edu/
STATEMENT OF AUTHORSHIP
- (a) Conception and DesignP. Ken Gillman
- (b) Acquisition of DataP. Ken Gillman
- (c) Analysis and Interpretation of DataP. Ken Gillman
- (a) Drafting the ManuscriptP. Ken Gillman
- (b) Revising It for Intellectual ContentP. Ken Gillman
- (a) Final Approval of the Completed ManuscriptP. Ken Gillman