Introduction

Serotonin syndrome is a rare and potentially fatal adverse drug reaction (ADR) caused by drugs with serotonergic properties and is due to increases in serotonin concentration or activation of the 5-HT, principally central postsynaptic 5HT-1 A and, most notably, 5HT-2 A receptors in the central nervous system. The syndrome is also called serotonin toxicity and severity depends on the amount of increased serotonin. Serotonin syndrome presents a classic triad of clinical symptoms: neuromuscular hyperactivity (such as clonus, myoclonus, tremor, hyperreflexia, rigidity), autonomic nervous system excitation (such as hyperthermia, tachycardia, diaphoresis) and altered mental state (agitation, confusion) [1,2,3].

Severe events can occur after stimulation of 5-HT receptors by drugs that increase serotonergic effects [1] (such as tryptophan, a serotonin precursor), drugs causing serotonin release (such as fenfluramine, sibutramine and amphetamines) or drugs that inhibit serotonin reuptake (such as selective serotonin reuptake inhibitors, SSRIs; serotonin and norepinephrine reuptake inhibitors, SNRIs, linezolid and monoamine oxidase (MAO) inhibitors). Many opioids may also block serotonin and noradrenaline reuptake, causing the serotonin syndrome [4]. Recent reports have also suggested that antiemetics and antinauseants acting as 5-HT3 antagonists can contribute to serotonin syndrome, through the excessive stimulation of other serotonin receptors, such as 5-HT1A and 5-HT2A, as a result of increased levels of serotonin due to 5-HT3 receptor antagonism [5].

Serotonin syndrome is only described in some case reports, but no epidemiological studies have investigated the drug classes most closely related to the syndrome and the frequency of this ADR in relation to age and sex. Again, the causal relationship with some medications (e.g. triptans or linezolid) is still not clear [6, 7].

We analysed adverse events in the US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) data set to investigate: (i) the main drug classes and active substances related to reports of serotonin syndrome; (ii) the reporting risk in relation to age and sex.

Methods

We ran a disproportionality analysis based on data from the FAERS database to assess whether there was any disproportion in the numbers of reports of serotonin syndrome in patients receiving drugs causing serotonin release or affecting serotonin reuptake. Reports were extracted from inception (1 January 2004) up to 30 September 2022. The FAERS database is a global repository for post-marketing safety reports, and pharmacovigilance using this database provides a warning of potential issues with marketed drugs [8, 9].

We used the online tool OpenVigil 2.1 (https://openvigil.sourceforge.net/) to query the FAERS database. OpenVigil 2.1 is a pharmacovigilance tool for data extraction, cleaning, mining and analysis on the FAERS database [10, 11]. OpenVigil 2.1 has already mapped arbitrary drug names (such as brand names, generic names, abbreviations, misspelt names, ecc.) to single names and allow one to query the database also through the Anatomical Therapeutic Chemical (ATC) Classification System of drugs.

Thus, we first identified each drug according to its ATC code, as reported in Supplementary Table S1. In the FAERS database, adverse events are coded according to the Medical Dictionary for Regulatory Activities (MedDRA) terminology. We detected all adverse events according to the Preferred Terms (PTs) from the MedDRA dictionary using the definition “Serotonin Syndrome” suggested by the tool, and for each report we selected the most recent version of reported cases.

Results are presented as Reporting Odds Ratios (ROR) [12, 13], a ratio similar to the odds ratio in case-control studies with their 95% confidence interval, calculated to establish the association between drugs investigated and the occurrences of serotonin syndrome reported events. The ROR is the ratio of the odds of reporting of one specific event (serotonin syndrome for a given drug) versus all other events compared to this reporting odds for all other drugs present in the database and registered during the same period. So the ROR compared cases and non-cases and a signal is considered when the lower limit of the 95% confidence interval (CI) of the ROR is greater than one [14]. A higher ROR indicates a stronger relation and any consideration was accorded to the criteria of Evans [15]. Institutional review board approval was not required because FAERS is an anonymized database open to pharmacovigilance centers.

Results

We found 8,997 cases of reported serotonin syndrome in 18 years of post-marketing safety reports of FAERS. SSRIs was the class of drugs with most reports of serotonin syndrome (4,548), followed by opioids (3,159) and other antidepressants (2,145). Table 1 shows the numbers of cases and the ROR regarding each class of drugs related to serotonin syndrome and individual substances. At class level the highest ROR were found for MAO inhibitors (45.99, 95% CI: 41.21–51.33) and SSRIs (32.66, 95% CI: 31.33–34.04).

Table 1 Main drug classes and substances related to serotonin syndrome
Full size table

The ten active substances with the highest ROR were moclobemide, isocarboxazid, oxitriptane, tranylcypromine, melitracen, phenelzine, linezolid, amoxapine, reboxetine and tryptophan, with values of ROR ranging from 44.19 (95% CI: 25.38–76.94) of tryptophan to 388.36 (95% CI: 314.58-479.46) of moclobemide.

Among the classes with highest number of reported cases of serotonin syndrome, the ROR was similar for the individual substances of the SSRIs, ranging from 18.06 (95% CI: 16.82–19.39) for citalopram to 33.33 (95% CI: 27.54–40.34) for fluvoxamine. Differently, among the opioids the ROR ranged from 0.45 (95% CI: 0.19–1.09) for oxymorphone to 17.75 (95% CI: 15.24–20.67) for tapentadol.

Among the 6,938 cases with complete information on age, 526 (7.6%) were patients under 17 years of age; 5,010 (72.2%) were adults between 18 and 64 years; and 1402 (20.2%) were over 65 years. The ROR for the most commonly involved drugs was higher in the group of older adults (Table 2). The risk was higher for opioids; tricyclic and tetracyclic antidepressants; triptans; linezolid and methylphenidate.

Table 2 Drugs related to serotonin syndrome stratified by age groups
Full size table

Among the 7,891 cases with complete information on sex, 4,633 (58.7%) of cases were female, but ROR for the most commonly involved drugs was higher in males (Table 3). For example, ROR for SSRIs were 41.15 (male) vs. 28.09 (female); for SNRIs were 26.68 vs. 21.70; for tricyclic and tetracyclic antidepressants were 20.04 vs. 12.67. In contrast, the ROR was higher in females for linezolid (61.32 vs. 43.27) and MAO inhibitors (48.08 vs. 35.12).

Table 3 Drugs related to serotonin syndrome stratified by sex
Full size table

Discussion

This study investigated the classes of drug and active substances most at risk for serotonin syndrome reporting in clinical practice. The number of cases between 2004 and 2022 was highest for SSRIs, followed by opioids and SNRIs, while MAOi was the class with the highest risk of serotonin syndrome reporting risk. Again, the serotonin syndrome reporting risk was generally higher in older adults and the number of cases were similar for males and females, but the reporting risk was higher in males, who could be at higher risk for this severe central nervous system ADR. Even though our study confirms that ADR are more frequent in older adults, for this suspected adverse reaction we did not find a definite higher prevalence in women than men, as suggested by some pharmacovigilance studies [16, 17].

Our study confirms that among the opioids those classified as high-medium risk of serotonin toxicity, such as tramadol, fentanyl, dextromethorphan, tapentadol and pethidine, [4] had the highest relation with the risk of serotonin syndrome. In fact, many opioids, in particular synthetic opioids, have actions on other targets, for example blocking serotonin reuptake or having affinity with postsynaptic receptors 5-HT1A and 5-HT2A [18].

A recent review found a lack of data supporting severe consequences of serotonin toxicity for linezolid [7], but we found that linezolid was strictly related with serotonin syndrome despite the small number of cases reported. The study also found several reports of serotonin syndrome for triptans, though the relation is not clear because of the lack of biological plausibility since triptans do not act on 5-HT2A receptors and very little on 5-HT1A [6], and the number of published cases reported is small. Similarly, a significantly higher reporting risk was found for the 5-HT3 antagonists, although these medications do not have the pharmacological properties necessary to contribute to serotonergic effects. However, as the WHO report suggested, there may be greater vulnerability with the concomitant use of 5-HT3 antagonists and some serotonergic drugs, for the increased availability of serotonin (which can stimulate 5-HT1a and 5-HT2a receptors) due to 5-HT3 receptor antagonism [5].

The main strength of this study is that, as far as we know, it is the first systematic investigation of the association between different medications acting on serotonin receptors and serotonin syndrome reports, using real world data. With a validated method to detect differences in ADR reports, our analysis shows the medications at greater risk of serotonin syndrome reporting rates, as well as the reporting risk in relation to age and sex. Again, a selection bias can be excluded since we used all records available, even if under reporting or selective reporting cannot be excluded.

This work has some limitations inherent to the method, like every pharmacovigilance study [19]: the lack of systematic recording of drug exposure duration or drug doses meant we could not discuss the role of higher doses of the different medications in the occurrence of serotonin syndrome. Because FAERS is an adverse events database, our analysis can only show an increase in the frequency of reports of serotonin syndrome among all reported adverse events, with no direct comparison of classes of medication.

Our study suggests that some signals, especially those found for drugs with a small number of cases reported, should be more carefully analysed in larger pharmaco-epidemiological studies and that future studies should compare the risk of serotonin syndrome between the different active substances among the same drug class.

Conclusions

Prescribers need to be vigilant about drugs that can raise serotonin concentrations or influence serotonergic neurotransmission, such as SSRIs, opioids, SNRIs and MAOi, particularly in older adults and male patients. Clinicians should be careful when using drugs with less well-known risk for serotonin syndrome, like linezolid and triptans, and must take account of the risk of serotonin syndrome in their risk-benefit analysis.