The 5-HT3 receptor

In brief

The exception in the serotonin family

Serotonin, the chemical whose proper name is 5-hydroxytryptamine and whose abbreviation is 5-HT, acts on a family of receptors. All but one of them are G-protein-coupled receptors: proteins that thread through the cell membrane seven times and, when they bind their chemical, activate an attached protein inside the cell which then sets off a slow cascade of further chemistry ^1,3. The 5-HT3 receptor is the exception. It is a hole in the membrane with a gate on it. When serotonin binds, the gate opens, positively charged ions flow through, the inside of the cell becomes less negative, and the cell fires ^2,12. That difference in kind, not degree, is why one drug class acting at this one receptor can stop a reflex within minutes.

The receptor was first recognized by its pharmacology, thirty years before anyone knew what it was made of. In 1957 John Gaddum and Zuleika Picarelli, working on strips of guinea-pig ileum, found that serotonin acted at two separable sites: a D receptor blocked by dibenzyline and probably on the muscle, and an M receptor blocked by morphine and probably in nerve tissue ¹¹. For decades the M receptor sat awkwardly outside serotonin pharmacology. In 1986 a committee convened to end the confusion proposed the modern names, and the M receptor of Gaddum and Picarelli became the 5-HT3 receptor ¹.

The proof that it was a channel came in 1989. Recording from excised patches of membrane taken from neurons of the guinea-pig submucous plexus, a nerve network in the gut wall, Derkach, Surprenant and North saw serotonin open individual channels of 15 and 9 picosiemens conductance, reversibly blocked by 5-HT3 antagonists, while acetylcholine opened a different, 40 picosiemens channel in the same patches ². A picosiemens measures how readily a channel passes current. These were the first single-channel recordings of a serotonin-gated current, and the authors drew the conclusion plainly: serotonin can be a fast synaptic transmitter, not only a slow modulator ².

Two years later the gene was cloned. Maricq and colleagues screened a complementary DNA library made from a mouse neuroblastoma cell line by injecting pools of messenger RNA into frog eggs and testing whether the eggs responded to serotonin with a current; the clone they recovered encoded a protein with the structural hallmarks of the ligand-gated ion channel family, and its messenger RNA was found in brain, spinal cord and heart ¹². The receptor was formally a new class of excitatory ligand-gated channel ¹².

Structurally it belongs to the Cys-loop receptors, named for a loop in the extracellular part of each subunit closed by a disulfide bridge between two cysteine residues. Its relatives are the nicotinic acetylcholine receptor, the GABA-A receptor on which benzodiazepines act, and the glycine receptor ^3,4. Five subunits assemble around a central pore, each contributing one membrane-spanning helix to the pore's lining ⁴. The mouse receptor's X-ray structure, published in 2014 at 3.5 angstrom resolution, showed the extracellular neurotransmitter site, a pore whose narrowest constriction measured 4.6 angstroms, and a bundle of five helices below the membrane forming a closed intracellular vestibule with its side portals obstructed by loops ⁴. Cryo-electron microscopy then filmed the sequence: a resting state ³³, two distinct serotonin-bound conformations with a continuous 165 angstrom permeation pathway and newly opened lateral portals where the membrane domain meets the intracellular domain ⁸, and a set of four structures including the receptor bound to the antiemetic tropisetron and to serotonin plus a positive modulator ³⁴. A 2026 review summarizes how far that structural work has come and where it stops ³⁵.

Five subunits, five genes, and the ones nobody understands

There are five 5-HT3 subunit genes in humans, named HTR3A, HTR3B, HTR3C, HTR3D and HTR3E, encoding subunits 5-HT3A to 5-HT3E. Every functional receptor must contain at least one A subunit; the others cannot form a working channel on their own ³. HTR3A sits on the long arm of chromosome 11, at 11q23.1 to 11q23.2 ⁵; that 1995 mapping study predates the discovery of the B subunit in 1999 ⁷, and because none of the sources used here maps HTR3B itself, its position is left unstated. HTR3D and HTR3E were identified and their full-length complementary DNAs cloned in 2003, and mapping placed them, together with the already known HTR3C, within a stretch of less than 100 kilobases on chromosome 3q27, a different chromosome entirely ⁶.

The A subunit alone makes a receptor that works but behaves wrongly. Homomeric receptors, built of five A subunits, conduct less than one picosiemens, so little that individual openings cannot be resolved, while the 5-HT3 receptors recorded from actual neurons conduct 9 to 17 picosiemens ⁷. In 1999 Davies and colleagues identified the B subunit and showed that A-plus-B heteromers conduct 16 picosiemens, are relatively impermeable to calcium, and reproduce the current-voltage behavior of native neuronal receptors, with distinct pharmacology ⁷. Oddly, the B subunit's pore-lining segment contains none of the structural features known to confer high conductance in related receptors ⁷. That puzzle was solved in 2003. The low conductance of the A-only receptor is set not in the lining of the pore but by three arginine residues in a stretch of the large intracellular loop, the HA-stretch, unique to the A subunit; replacing them with their B-subunit counterparts raised single-channel conductance 28-fold, which the authors said solved the conundrum ⁴².

The remaining three subunits are the least understood part of the receptor. Their tissue distribution is narrow: HTR3D messenger RNA was found only in kidney, colon and liver, HTR3E only in colon and intestine, while HTR3A and HTR3B are expressed widely including in the brain ⁶. What they contribute to a working receptor is largely inference. A 2025 computational study built homology models of heteromeric receptors containing the D subunit and predicted that it changes the conformational transitions of the channel, and that substitutions at glycine 110, the variant rs6443930, alter the extracellular domain ⁴³; these are predictions from models, not measurements from tissue, although the authors describe them as molecular evidence for the association of those variants with clinical disorders. No structure of a heteromeric 5-HT3 receptor in a native membrane has been solved, which a 2026 review names as the field's main unfinished business ³⁵.

Genetic variation in these genes has been collected into a dedicated database covering all five genes together with functional and pharmacogenetic annotations ⁴⁴. The clinical signal is real but modest. In 242 patients studied on the first day of chemotherapy, with nausea or vomiting in about 30 percent, sequencing the whole HTR3B gene turned up thirteen polymorphisms, and patients homozygous for a three-base deletion in the promoter region vomited more often than everyone else in both observation periods; the authors concluded that only a small fraction of antiemetic failure is explained by it ⁴⁵. In irritable bowel syndrome, patients whose stool consistency responded to ondansetron were more likely to carry the CC genotype of a variant in HTR3C, 33 percent against 14 percent ³¹. In cyclic vomiting syndrome, variants in HTR3B and HTR3C were associated with risk in a study of 70 patients ⁴⁶.

From a cell in the gut lining to the floor of the fourth ventricle

The circuit that the setrons interrupt begins in the lining of the intestine. Enterochromaffin cells are rare, scattered cells in the gut epithelium that store serotonin. For decades they were presumed to be sensors, but they are too few and too inaccessible for ordinary measurement. In 2017 a group at the University of California, San Francisco grew intestinal organoids, miniature cultured guts, and made single-cell recordings from the enterochromaffin cells within them. The cells expressed specific chemical sensors, were electrically excitable, and modulated serotonin-sensitive sensory nerve fibers through synapse-like connections, which is as direct a demonstration as exists that they transduce events in the gut lumen into neural signals ³⁶.

Cytotoxic chemotherapy sets this system off. In the 1990 trial that introduced ondansetron, urinary excretion of 5-hydroxyindoleacetic acid, abbreviated 5-HIAA, the main breakdown product of serotonin, rose in all patients two to six hours after cisplatin, and the rises paralleled the episodes of vomiting; the authors proposed that cisplatin releases serotonin from enterochromaffin cells and that ondansetron works by blocking the receptors it acts on ²⁸. A later study using intravenous microdialysis probes sharpened the picture considerably. In sixteen patients given about 89 milligrams of cisplatin per square meter of body surface, the metabolite in blood dialysate roughly doubled over four hours, but free serotonin in the blood did not rise at all ⁴⁷. That argues against the old idea that circulating serotonin reaches the area postrema, where the blood-brain barrier is leaky, and in favor of serotonin acting locally within the intestinal wall ⁴⁷.

The nerves carrying the signal were identified by cutting them. In ferrets, which vomit and are therefore used for this work, a combination of bilateral abdominal vagotomy and section of the greater splanchnic nerves completely abolished retching and vomiting in response to intraperitoneal cisplatin or intravenous cyclophosphamide ²⁹. The same study found that BRL 24924, a gastrokinetic drug with more specific M-receptor antagonist properties than metoclopramide, almost completely abolished the response, far outperforming metoclopramide; the authors concluded that serotonin acting on M receptors, as they were still called, on visceral afferent nerves was the probable mechanism, while noting that a central site could not be excluded ²⁹. The fibers that were cut end in the brainstem. Vagal afferents from the gut terminate in the nucleus tractus solitarius, the sensory nucleus that sits immediately beside the area postrema and relays to the dorsal motor nucleus of the vagus; those three together make up the dorsal vagal complex, and that is where the second site of action lies ^9,10.

It could not be excluded because the receptor is also in the brainstem, in exactly the right place. Autoradiography in ferrets with a high-affinity labeled 5-HT3 ligand found a high concentration of binding in the area postrema, and injecting tiny amounts of three different 5-HT3 antagonists directly into that region, ondansetron at 0.01 to 1 microgram among them, inhibited cisplatin-induced vomiting, while antagonists of other serotonin receptors injected into the same spot did not ⁹. In human tissue the result is just as clean: in slices of human medulla oblongata, saturable binding of tritium-labeled granisetron, then known by its laboratory code BRL 43694, appeared in the dorsal vagal complex, with a maximum binding capacity of 1.88 picomoles per milligram of protein and a dissociation constant of 1.21 nanomolar, and in no other medullary region ¹⁰.

So there are two candidate blockade sites, the vagal endings in the gut wall and the dorsal vagal complex in the brainstem, and the clinical evidence cannot separate them; route is no help, because an oral setron is absorbed into the blood and reaches both. What route does show is practical. A randomized trial of 1054 patients receiving cisplatin found a single 2 milligram oral granisetron tablet equivalent to a 32 milligram intravenous ondansetron infusion, with total control of nausea and vomiting in 54.7 percent against 58.3 percent ³⁷.

Before 1990: high-dose metoclopramide, and what replaced it

Cisplatin, introduced in the 1970s, is among the most reliably emetic drugs ever given to human beings, and for a decade there was almost nothing to give against it. The first real advance came in 1981, when Richard Gralla and colleagues at Memorial Sloan-Kettering ran two small double-blind trials in 41 patients receiving cisplatin. High intravenous doses of metoclopramide, a drug ordinarily used at much lower doses for its effects on gastric emptying, reduced the median number of vomiting episodes from 10.5 on placebo to 1.0, and from 12.0 on prochlorperazine to 1.5 ²⁰. The price was the dose. Mild sedation was frequent and one patient had an extrapyramidal reaction, the involuntary muscle spasm that dopamine blockade can cause ²⁰.

Why very high doses of a dopamine antagonist should work was answered backward, after the fact. The 1988 ferret study noted that other work had recently attributed metoclopramide's antiemetic action to antagonism at the serotonin M receptor, and showed directly that metoclopramide did not fully control retching or vomiting, while BRL 24924, a gastrokinetic drug with more specific M-receptor blocking properties, almost totally abolished both ²⁹. In other words, the best antiemetic of the 1980s was a poor 5-HT3 blocker used at doses high enough to hit a receptor it had not been designed for.

Two papers published back to back in the New England Journal of Medicine on 22 March 1990 changed that. The first, in 28 patients, compared three intravenous doses of ondansetron at 0.15 milligrams per kilogram against saline placebo. The median time to the first episode of vomiting went from 2.8 hours on placebo to 11.6 hours on ondansetron; median episodes in 24 hours fell from 5.5 to 1.5; and whereas 12 of 14 placebo patients needed rescue medication, none of the 14 given ondansetron did ²⁸. The second, a double-blind crossover study in which 76 of 97 chemotherapy-naive patients completed both arms, compared ondansetron with high-dose metoclopramide: complete or near-complete control of vomiting, which the trial defined as no episodes of vomiting or no more than two, in 57 of 76 treatments with ondansetron, 75 percent, against 32 of 76, 42 percent, with metoclopramide (p below 0.001), and dystonic reactions in three metoclopramide treatments and none with ondansetron ¹⁴.

The class filled out quickly. By 1998 a trial of 1054 patients could show that a single oral granisetron tablet matched a 32 milligram intravenous ondansetron infusion ³⁷. In 2003 palonosetron, a second-generation agent with a long half-life, beat a 32 milligram ondansetron infusion in a 570-patient phase III trial ⁴⁸. And in the same year the first neurokinin-1 antagonist, aprepitant, showed that the setrons had only solved half the problem ⁴⁹.

The drugs, and the half of the problem they do not solve

The drugs that block this receptor are called setrons, after the suffix they share. Ondansetron, granisetron, tropisetron and dolasetron are the first generation; palonosetron is the second; ramosetron is used mainly in Asia; alosetron was developed for the bowel rather than for vomiting ^3,13. They are competitive antagonists, sitting in the serotonin-binding pocket at the interface between two subunits, and cryo-electron microscopy structures of the receptor in complex with palonosetron, ondansetron and alosetron show that each leaves a distinct fingerprint of interactions with the pocket, and that the degree to which each distorts the channel's activation pathway differs ⁴⁰.

Within the first generation, the differences are small enough that the choice is usually made on cost and route. The 1054-patient trial comparing a single 2 milligram oral granisetron tablet with a 32 milligram intravenous ondansetron infusion before cisplatin-based chemotherapy found total control in 54.7 percent against 58.3 percent, with a 95 percent confidence interval on the difference of minus 9.6 to 2.4, that is, equivalence; complete control of emesis was 61.2 against 67.1 percent, a difference whose interval just excluded zero ³⁷. Every patient was receiving cisplatin at 60 milligrams per square meter or more, and both arms were allowed, though not required, to receive dexamethasone or methylprednisolone alongside the setron, so the trial establishes the equivalence of the two drugs within a regimen rather than the sufficiency of either alone ³⁷. A confidence interval is the range within which the true value probably lies given the size of the study.

Palonosetron is the one genuine advance within the class. In a double-blind phase III trial, 570 patients receiving moderately emetogenic chemotherapy were randomized to a single intravenous dose of palonosetron 0.25 milligrams, palonosetron 0.75 milligrams, or ondansetron 32 milligrams; 563 were evaluable. Complete response rates for palonosetron 0.25 milligrams against ondansetron were 81.0 against 68.6 percent in the first 24 hours, 74.1 against 55.1 percent from 24 to 120 hours, and 69.3 against 50.3 percent overall, each significant at p below 0.01 ⁴⁸. Curiously, the higher 0.75 milligram dose was not statistically better than ondansetron ⁴⁸. A 2026 systematic review and meta-analysis of eleven randomized trials in adults and children confirmed the pattern with moderate to high certainty: relative risk for complete response 1.06 acute, 1.30 delayed, 1.29 overall, with comparable adverse events ⁴¹.

Two laboratory papers proposed a mechanism for that difference: palonosetron binds allosterically and with positive cooperativity, unlike the simple bimolecular binding of ondansetron and granisetron, and accelerates their dissociation from the receptor ³⁸; and it triggers internalization of the receptor into the cell, with inhibition of calcium influx persisting after the drug has been washed off, which ondansetron and granisetron do not do ³⁹. The 2008 paper came from scientists at MGI Pharma, which marketed palonosetron in the United States, and the 2009 paper from a Johns Hopkins laboratory with co-authors from the same commercial collaboration; the affiliations declare it. Later cryo-electron microscopy placed palonosetron in the same orthosteric binding pocket as ondansetron and alosetron and described all three as competitive antagonists ⁴⁰, which sits uneasily with the allosteric reading. The clinical difference itself was confirmed in the 2026 meta-analysis ⁴¹; the mechanism offered for it should be treated as a plausible, commercially originated hypothesis. Allosteric action at this receptor is, however, no longer a claim unique to one company's drug. In 2026 an independent group recording macroscopic currents from human 5-HT3A receptors, and then mutating candidate residues, showed that piperazine, a drug used against intestinal worms, and its derivative hydroxyzine, a sedating antihistamine, inhibit the receptor as negative allosteric modulators acting at two sites at the interfaces between subunits, one in the lower extracellular domain and one in the upper membrane domain; mutations at E75, E208, S248 and D293 abolished or enhanced the effect, which is the test of whether a proposed site is real ⁵⁶. That is cell-based work on drugs given for other reasons, not a clinical finding.

What no setron fully controls is the delayed phase. In 2003 a multinational trial analyzed 520 patients, by modified intention to treat, who were receiving at least 70 milligrams of cisplatin per square meter and were given either standard therapy, ondansetron plus dexamethasone, or the same plus the oral neurokinin-1 antagonist aprepitant. Complete response over days one to five was 72.7 percent with aprepitant against 52.3 percent without, and the gap was widest on days two to five ⁴⁹. That trial defined the modern regimen. Current guidelines, including the American Society of Clinical Oncology update and the 2026 Chinese Anti-Cancer Association guideline, build antiemetic prophylaxis from a 5-HT3 antagonist combined with dexamethasone, a neurokinin-1 antagonist and, for highly emetogenic regimens, olanzapine, an antipsychotic drug that blocks several receptors at once and is used here purely as an antiemetic, selected by the emetic risk of the chemotherapy and the individual patient ^21,57,58. The 5-HT3 receptor is one pillar of a four-pillar regimen, not the regimen.

Where else blocking the channel works

After surgery, it works but is not special. A landmark factorial trial in postoperative nausea and vomiting enrolled 5199 high-risk patients and randomized 4123 of them to one of 64 possible combinations of six prophylactic interventions, among them 4 milligrams of ondansetron, 4 milligrams of dexamethasone and 1.25 milligrams of droperidol. Each of the three antiemetics reduced the risk by about 26 percent; propofol anesthesia reduced it by 19 percent and avoiding nitrous oxide by 12 percent ²². The interventions acted independently and their effects multiplied, which means the sensible rule is to use the safest or cheapest first and to reserve combinations for high-risk patients ²². A 2020 Cochrane network meta-analysis of 585 randomized trials and 97,516 participants confirmed the picture: ondansetron reduced vomiting in the first 24 hours with high certainty, relative risk 0.55 (95 percent confidence interval 0.51 to 0.60), but it ranked thirteenth of 28 single drugs, behind aprepitant at 0.26, ramosetron at 0.44, granisetron at 0.45 and dexamethasone at 0.51 ⁵⁹. On this evidence ondansetron is one interchangeable option among several, not the best one. The current standard of care follows the same logic: the Fourth Consensus Guidelines, issued in 2020 by an international panel convened under the American Society for Enhanced Recovery and the Society for Ambulatory Anesthesia, endorsed by 23 societies and based on literature to September 2019, recommend identifying high-risk patients, reducing baseline risk, and multimodal prophylaxis combining agents from different classes, with 5-HT3 blockade as one component rather than the regimen ⁶⁰.

In children with gastroenteritis, a single oral dose is genuinely useful. In a double-blind trial of 215 children aged six months to ten years treated in a pediatric emergency department, orally disintegrating ondansetron reduced the proportion who vomited during oral rehydration from 35 percent to 14 percent, relative risk 0.40 (95 percent confidence interval 0.26 to 0.61), and the proportion needing intravenous rehydration from 31 percent to 14 percent, relative risk 0.46 ²³. Hospital admissions and return visits did not differ ²³.

In diarrhea-predominant irritable bowel syndrome, the drug does something distinct from its antiemetic action. A randomized, double-blind crossover trial of 120 patients compared five weeks of ondansetron 4 milligrams, with dose titration, against placebo. Stool consistency improved by 0.9 points on the seven-point Bristol scale (95 percent confidence interval 0.6 to 1.1, p below 0.001); days with urgency, urgency scores, stool frequency and bloating all fell; the symptom severity score fell by 83 points against 37 on placebo; and 65 percent reported adequate relief on ondansetron but not placebo, against 14 percent the other way round, relative risk 4.7 (2.6 to 8.5) ³⁰. Abdominal pain, notably, did not improve significantly ³⁰. A companion analysis found that these patients had roughly twice the control concentration of serotonin's metabolite in rectal biopsies, and that the patients who needed the least drug had the lowest mucosal serotonin, which is a plausible, if still correlational, account of who responds ³¹. A meta-analysis of twenty-one randomized trials across the class found global symptoms improved with relative risk 1.56 (1.43 to 1.71), with constipation 3.71 times as likely as on placebo ¹³.

In schizophrenia, adding a setron to an antipsychotic has been tested repeatedly and the effect is small and uncertain. A 2025 meta-analysis of eight randomized trials in 533 patients found that adjunctive ondansetron improved the negative-symptom subscale of the Positive and Negative Syndrome Scale by 2.96 points at twelve weeks (95 percent confidence interval 1.24 to 4.69 points of improvement) and general psychopathology by 2.71 points, with no effect on positive symptoms and more constipation ⁶¹. An earlier meta-analysis of six trials in 311 patients reported larger standardized effects but with heterogeneity above 80 percent, meaning the trials disagreed with each other substantially ⁶². These are small trials of a cheap old drug in a hard illness, and the right reading is that the question is open, not that the treatment is established.

Two other findings mark the edges of what the receptor does. In mice lacking the receptor or given an antagonist, the pain that follows tissue injury is reduced while brief acute pain is unaffected, and the receptor transcripts sit in a previously uncharacterized subset of sensory fibers; this is a mouse result, and it has not translated into setrons being used as analgesics ²⁵. And in rats, the increase in new neuron formation in the hippocampus produced by fluoxetine was blocked by co-treatment with ondansetron, implying the receptor is required for that effect; again, rats, and no human counterpart ⁶³.

The heart: hERG, the QT interval, and the dose that was withdrawn

The setrons have an off-target action. Beyond the 5-HT3 receptor, they block hERG, the potassium channel that carries the rapid delayed-rectifier current responsible for resetting the electrical charge of heart muscle cells after each beat. Blocking it lengthens the QT interval on the electrocardiogram, the time from the start of ventricular depolarization to the end of repolarization, and a long QT interval is the substrate for torsades de pointes, a ventricular rhythm that can degenerate into fibrillation and sudden death. Measured directly in cells expressing the human channels, the concentrations producing half-maximal block of hERG were 0.81 micromolar for ondansetron, 3.73 for granisetron, 5.95 for dolasetron and 12.1 for the active metabolite of dolasetron; all four also blocked the cardiac sodium channel in a frequency-dependent way, ondansetron least potently at 88.5 micromolar ¹⁵. Ondansetron's half-maximal blocking concentration for hERG, 0.81 micromolar, is the lowest of the four, which makes it the most potent hERG blocker among them, and the authors concluded that this submicromolar affinity likely underlies the prolongation of cardiac repolarization reported with the drug ¹⁵.

In patients, the effect is measurable at ordinary doses. In 85 patients treated for postoperative nausea and vomiting, droperidol 0.75 milligrams prolonged the corrected QT interval by a mean of 17 plus or minus 9 milliseconds and ondansetron 4 milligrams by 20 plus or minus 13 milliseconds, both peaking within three minutes and both returning below baseline by 90 minutes ⁶⁴. A controlled crossover study in sixteen healthy volunteers measured maximum placebo-corrected, baseline-subtracted prolongation of 25 plus or minus 8 milliseconds for droperidol 1 milligram, 17 plus or minus 10 milliseconds for ondansetron 4 milligrams, and 28 plus or minus 10 milliseconds for the two together, with no pharmacokinetic interaction ⁶⁵. The practical conclusion from the first of those studies was uncomfortable and worth stating plainly: when used for postoperative nausea, the safety of 5-HT3 antagonists may not be superior to that of low-dose droperidol ⁶⁴, which had itself been the subject of a boxed warning.

The dose question was settled by a formal thorough QT study conducted to the international regulatory standard, with placebo and a positive control. Single intravenous infusions over 15 minutes produced a maximum mean placebo-corrected QTcF prolongation of less than 10 milliseconds after 8 milligrams and approximately 20 milliseconds after 32 milligrams; modeling predicted 9.2 milliseconds, upper 90 percent confidence bound 11.2, for 16 milligrams. The conclusion was that single intravenous doses above 16 milligrams should no longer be used ¹⁶. That study was designed and authored by scientists at GlaxoSmithKline, which marketed ondansetron, and it is the rare case of a manufacturer's own study tightening its own label; the direction of the commercial interest is worth noting even when the result runs against the sponsor ¹⁶. The 32 milligram intravenous dose, which had been the comparator in the pivotal trials of both granisetron and palonosetron ^37,48, was restricted by the US Food and Drug Administration in 2012 and removed ^17,51.

What that leaves is a drug with a real but modest electrophysiological effect, a handful of alarming case reports, and a pharmacovigilance signal that is easy to over-read. Two children died of refractory cardiac arrest after receiving ondansetron for gastroenteritis, one a previously healthy ten-year-old who developed wide-complex tachycardia, the other an 86-day-old infant with previously unrecognized congenital cardiomyopathy; the authors of the report were careful to frame these as cases in which other proarrhythmic factors were present ⁶⁶. A 2026 analysis of the Food and Drug Administration's adverse event reporting system covered 2608 ondansetron and 15,257 olanzapine reports from adults that named the drug as primary suspect between January 2015 and October 2025, with the narrow outcome defined by two preferred terms, torsades de pointes and electrocardiogram QT prolonged: 4.91 percent of the ondansetron reports concerned those events against 1.61 percent of the olanzapine reports, with reporting odds ratios of 27.24 (95 percent confidence interval 22.78 to 32.58) against 8.70 (7.65 to 9.88) ⁶⁷. A reporting odds ratio measures how disproportionately an event appears in a voluntary reporting database; it cannot establish causation, as the authors state, and it cannot estimate incidence and is distorted by publicity ⁶⁷.

The current synthesis, from a 2026 clinical review and risk-stratification framework, is that QT liability is highly heterogeneous within the antiemetic classes: among the setrons, ondansetron's effect is dose-dependent, which is what prompted the 2012 restriction on the 32 milligram intravenous dose, while palonosetron has minimal effect; neurokinin-1 antagonists, low-risk antihistamines and anticholinergics carry little liability, whereas the butyrophenones, the phenothiazines and promethazine carry more; and serious events cluster in patients with low potassium, slow heart rates, several QT-active drugs at once, or an inherited long QT syndrome, particularly when the corrected QT exceeds 500 milliseconds or rises more than 60 milliseconds from baseline ¹⁷. The authors' conclusion is that antiemetic-associated arrhythmia is largely preventable by electrolyte correction, electrocardiographic monitoring and choosing a low-liability agent, not by avoiding treatment ¹⁷.

Pregnancy and hyperemesis gravidarum: the signal, the absolute risks, and the split

Nausea and vomiting affect most pregnancies. Hyperemesis gravidarum is the severe end of that spectrum: vomiting severe enough to cause weight loss, dehydration and hospital admission, graded in current practice with validated scores such as the Pregnancy-Unique Quantification of Emesis, and explicitly not graded by ketones in the urine, which the 2024 United Kingdom guideline states is not an indicator of dehydration and should not be used ²⁴. Ondansetron is not licensed for this. It is used off-label, as second-line treatment after antihistamines and the doxylamine-pyridoxine combination, and its use grew rapidly: among control mothers in two large American birth-defect studies, first-trimester use rose from under one percent before 2000 to 13 percent in 2013 and 2014 ⁶⁸.

The first question is whether it works, and the evidence is thinner than the volume of prescribing suggests. A double-blind randomized trial of 160 women admitted with hyperemesis gravidarum compared intravenous ondansetron 4 milligrams with metoclopramide 10 milligrams eight-hourly for 24 hours and found them similar on nausea, vomiting and well-being, with less drowsiness, 12.5 against 30 percent, less dry mouth, 10.0 against 23.8 percent, and less persistent ketonuria, 12.5 against 30 percent, on ondansetron ⁶⁹. A smaller trial of 36 women with nausea and vomiting of pregnancy found ondansetron better than doxylamine with pyridoxine, with median nausea scores falling 51 millimeters against 20 on a 100 millimeter scale ⁷⁰. Both compared the drug with another drug. Until 2026 nobody had compared it with placebo in hyperemesis gravidarum.

When somebody finally did, it did not separate. The Danish VOMIT trial randomized 59 pregnant people with hyperemesis gravidarum, across seven hospitals in Denmark between 2019 and 2022, to 14 days of mirtazapine, an antidepressant long used off-label in this setting because it blunts nausea and restores appetite, or ondansetron, or placebo. The reduction in symptom score at day 2 was 1.86 points greater with mirtazapine than placebo (95 percent confidence interval 0.12 to 3.61, p equals 0.04) and 0.51 points greater with ondansetron than placebo (interval spanning no effect, from 1.30 points worse to 2.32 points better) ¹⁸. Mild to moderate adverse events were commoner with mirtazapine, while serious adverse events were few and evenly spread between the groups ¹⁸. Fifty-nine participants is small, and the trial was not designed to prove ondansetron ineffective; the honest statement is that the one placebo-controlled trial that exists did not show a significant benefit, and that this is a reversal of expectation in a field where the drug is standard.

The safety question is older and much larger. It began reassuringly. A Danish cohort drawn from 608,385 pregnancies, with 1:4 propensity-score matching, found no significant increase in spontaneous abortion, stillbirth, preterm delivery, low birth weight or major birth defects, the last at 2.9 percent in both exposed and unexposed, prevalence odds ratio 1.12 (95 percent confidence interval 0.69 to 1.82) ⁷¹. Then a Swedish registry study of 1349 infants exposed in early pregnancy between 1998 and 2012 reported no increase in major malformation overall but an increased risk of cardiovascular defects, odds ratio 1.62 (95 percent confidence interval 1.04 to 2.14), and specifically of cardiac septal defects, risk ratio 2.05 (1.19 to 3.28); the authors wrote that an increased risk of cardiac septum defect was likely ⁵².

The study that defined the modern answer was published in JAMA in 2018. Huybrechts and colleagues analyzed 1,816,414 pregnancies in the United States Medicaid program, of which 88,467, or 4.9 percent, were exposed to ondansetron in the first trimester, with propensity-score stratification to control for the indication and other confounders. The absolute risk of cardiac malformation was 84.4 per 10,000 births unexposed and 94.4 per 10,000 exposed, and after adjustment the association vanished: adjusted relative risk 0.99 (0.93 to 1.06), risk difference minus 0.8 per 10,000 ⁵³. The absolute risk of oral clefts was 11.1 per 10,000 unexposed and 14.0 per 10,000 exposed, and that association survived adjustment: relative risk 1.24 (1.03 to 1.48), risk difference 2.7 per 10,000 (0.2 to 5.2) ⁵³. Congenital malformations overall showed nothing: relative risk 1.01 ⁵³. The same group returned to the question of route in a 2020 JAMA research letter on first-trimester intravenous exposure in the same Medicaid data; its estimates are given only in the letter itself and are therefore not quoted here ⁷². The 2018 paper is the pivot of the whole literature: it overturned the cardiac signal and left a small cleft signal standing.

Other cohorts cut both ways, and the disagreements are informative. Two American case-control studies, the National Birth Defects Prevention Study and the Slone Birth Defects Study, found no increased risk for most of 51 defect groups examined, with modest increases for cleft palate, adjusted odds ratio 1.6 (1.1 to 2.3) in the first, and for renal agenesis-dysgenesis, 1.8 (1.1 to 3.0) in the second, which the authors themselves said might be chance ⁶⁸. An analysis of 864,083 commercially insured mother-infant pairs reported a cardiac odds ratio of 1.52 (1.35 to 1.70) and an orofacial cleft odds ratio of 1.32 whose interval, 0.76 to 2.28, included no effect; its comparison group was women with no antiemetic exposure at all, which leaves confounding by indication wide open, since women sick enough to need an antiemetic differ from women who are not ⁷³. A single-hospital study of 33,677 deliveries, with 3733 first-trimester exposures at doses ranging from 2.4 to 1008 milligrams, found risk ratios for ventricular septal defect of 1.7 to 2.1 depending on method, a dose-dependence with the highest cumulative doses carrying an adjusted risk ratio of 3.2 (1.0 to 9.9), and translated the finding usefully: approximately one additional ventricular septal defect for every 330 pregnancies exposed in the first trimester ⁷⁴.

The meta-analyses land close to each other and close to the null. A 2019 systematic review found no significant increase in major malformations, heart defects, orofacial clefts or genitourinary malformations in its primary analysis, with elevated point estimates in secondary analyses warranting continued surveillance ⁷⁵. A 2020 meta-analysis of twelve studies found ventricular septal defects at odds ratio 1.11 (1.00 to 1.23), oral clefts at 1.22 (1.00 to 1.49), and, after excluding the study responsible for the heterogeneity, cleft palate at 1.48 (1.19 to 1.84) ⁷⁶. A multinational cohort study with meta-analysis across five Canadian provinces, United States claims data and United Kingdom practice records, 456,963 pregnancies in all, compared ondansetron against other antiemetics rather than against nothing, and found no association with fetal death, adjusted hazard ratio 0.91 (0.67 to 1.23), or major congenital malformations, odds ratio 1.06 (0.91 to 1.22) ⁷⁷. A 2022 systematic review of twenty observational studies found small associations with cardiac and neural tube defects in its primary analysis that did not survive sensitivity analysis, no association with orofacial clefts, 1.09 (0.95 to 1.25), and a reduced incidence of miscarriage ⁷⁸. Across all of this, the framing that best serves a patient is the one a 2020 review set out: even if ondansetron is causally implicated, the absolute increase is small, on the order of 0.03 percent for orofacial clefts and 0.3 percent for ventricular septal defect, and must be balanced against the risks of inadequately treated vomiting ⁷⁹. The most recent synthesis is a 2026 scoping review, which searched four databases from inception to March 2024 and updated the search in May 2025, screened 1942 records and included 39 original studies, 34 of them on ondansetron and 5 on granisetron; it found no elevated risk of congenital anomalies overall, described the cardiac and orofacial findings as conflicting with minimal absolute changes in risk either way, and concluded that ondansetron likely poses a low risk of harm when set against the excess risks of untreated vomiting, while granisetron has been studied too little to be used widely in pregnancy ⁸⁰. A scoping review maps a literature rather than pooling it, so it produces no summary estimate of its own.

A mechanism has been proposed, and it is the same mechanism as the cardiac one. Ondansetron is a potent hERG blocker, and hERG carries the current that controls the rhythm of the embryonic heart. Applied to gestational day 13 rat embryos in culture, ondansetron caused concentration-dependent slowing and arrhythmia of the embryonic heartbeat, and cardiovascular malformations occurred in rats at exposures only slightly higher than those of early human pregnancy; the authors integrated this with published animal teratology and pharmacokinetics and argued that risk might be increased in human pregnancy when other factors such as low potassium are present ⁸¹. This is a rat study with in vitro embryo culture. It establishes plausibility, not human causation, and the human cardiac association that it was proposed to explain is the one that did not hold up ⁵³.

The regulators split, and the split has not closed. In July 2019 the European Medicines Agency's Pharmacovigilance Risk Assessment Committee reassessed first-trimester use, and in November 2019 the European product information was changed to state that ondansetron should not be used in the first trimester of pregnancy ⁵⁴. The European Network of Teratology Information Services, the body that fields clinicians' questions about drugs in pregnancy, published a direct rebuttal arguing that the step was insufficiently substantiated and did not serve the interests of pregnant women with severe nausea and vomiting ⁵⁴. The United Kingdom's Royal College of Obstetricians and Gynaecologists took the other position in its 2024 Green-top Guideline No. 69: ondansetron is safe and effective, its use as a second-line antiemetic should not be discouraged where first-line agents fail, and women can be reassured regarding a very small increase in the absolute risk of orofacial clefting, which should be balanced against the risks of poorly managed hyperemesis ²⁴.

The work since 2024 has mostly been about whether the remaining cleft signal is real or an artifact of who agrees to be in a study. A 2024 quantitative bias analysis asked what degree of differential participation would be needed to produce the cleft palate association seen in the National Birth Defects Prevention Study if the true odds ratio were in fact null, and found that ondansetron-exposed cleft palate cases would need to have been 1.6 times more likely to take part than exposed controls, a magnitude the authors regarded as possible ⁸². A prospective cohort from an Israeli teratology information service followed 260 women with nausea and vomiting of pregnancy, 137 exposed to ondansetron and 123 not; major congenital malformations occurred in 5.1 against 4.1 percent, p equals 0.695, with no consistent pattern, though the authors note the sample cannot detect rare outcomes ⁸³. A different safety question was answered in 2025: a South Korean national cohort of 630,904 children, designed to emulate a pragmatic trial, examined seven neurodevelopmental disorders and formal developmental screening after exposure to six antiemetics in pregnancy, and among the 5673 children, 0.9 percent of the cohort, exposed to ondansetron found no substantial association with neurodevelopmental disorders or delays; the only signals involved metoclopramide, with a hazard ratio of 1.12 (95 percent confidence interval 1.06 to 1.18) for attention deficit hyperactivity disorder after exposure in the second half of pregnancy and delays after seven or more days of use ⁸⁴. A 2026 retrospective cohort restricted to women with hyperemesis gravidarum, using records from more than 500 American healthcare facilities between 2001 and 2016 and propensity-matching 10,238 pairs for spontaneous abortion and 13,585 for stillbirth against women given no antiemetic at all, found ondansetron associated with a lower risk of spontaneous abortion, adjusted hazard ratio 0.82 (0.73 to 0.91), with no dose-response, and no association with stillbirth; a secondary comparison against users of other antiemetics pointed the same way, 0.85 (0.73 to 0.98). A protective association of that kind, in a population defined by the severity of its vomiting, invites the same confounding that inflates risk estimates elsewhere, and is better read as reassurance about harm than as evidence of benefit ⁸⁵. Earlier work in American hyperemesis patients had pointed the same way ⁸⁶. A disproportionality analysis of pregnancy reports in the Food and Drug Administration database flagged congenital and cardiac categories, with the authors stating that causal relationship and risk level cannot be inferred from such data ⁸⁷. And a 2025 study found an association between ondansetron use in pregnancy and urinary tract infection in 7023 women, with more severe bladder infection in ondansetron-treated pregnant mice, a finding that joins human association to an animal mechanism and needs replication before it changes anything ⁸⁸.

The deepest change of the last three years is not about the drug at all. In 2023 and 2024 a large collaborative study made the strongest case yet that the nausea and vomiting of pregnancy is caused in large part by GDF15, a hormone that in pregnancy comes mainly from the feto-placental unit and acts on the brainstem. Mass spectrometry of a naturally labeled variant of the hormone showed that the great majority of the GDF15 in maternal plasma comes from the feto-placental unit; higher maternal levels accompanied vomiting and hyperemesis; carriers of rare and common genetic variants that give low levels before pregnancy were at higher risk, so maternal sensitivity matters as much as fetal production; and women with beta-thalassemia, an inherited anemia in which GDF15 is chronically high, reported very little nausea and vomiting in pregnancy. The authors describe their findings as support for a putative causal role rather than as proof ⁵⁵. That is a different receptor, a different signal, and a different therapeutic target from the one ondansetron blocks. Read alongside the VOMIT trial's failure to separate ondansetron from placebo ¹⁸, it suggests that the pregnancy question of the next decade may not be how safe the setron is but whether it was ever the right drug.

One further practical point. Ondansetron is metabolized in the liver by several cytochrome P450 enzymes, and CYP2D6 is only one of them, although it is the one whose genetic variation is known to matter: the extent of CYP2D6 involvement differs across the setrons, and its influence on efficacy is clearest for ondansetron and tropisetron ⁸⁹. Ultrarapid metabolizers have therefore long been suspected of failing to respond. A 2026 case-control study genotyped 264 pregnant patients treated for nausea and vomiting and found no significant association between metabolizer status and response; ultrarapid and normal metabolizers had a non-significantly higher odds of non-response than poor and intermediate metabolizers, odds ratio 1.53 (0.88 to 2.66) ⁹⁰. The hypothesis is not refuted by that study, and guideline writers have not abandoned it either: the Clinical Pharmacogenetics Implementation Consortium updated its CYP2D6 guidance for this drug class in 2026 and gives genotype-directed therapeutic recommendations where variation is associated with reduced efficacy ⁸⁹. The fair summary is narrower than it first appears. A guideline says what to do with a CYP2D6 result that already exists, including choosing a setron less dependent on the enzyme; it is not a case for genotyping pregnant patients routinely, and the one pregnancy-specific study found no association between metabolizer status and response ^89,90.

Alosetron: the withdrawal, the public campaign, and the restricted return

Alosetron is a 5-HT3 antagonist developed not for vomiting but for the bowel. The logic was sound: the receptor sits on enteric and vagal neurons, blocking it slows colonic transit, and the commonest complaint in women with diarrhea-predominant irritable bowel syndrome is urgency. One pivotal trial randomized 647 women with diarrhea-predominant or alternating bowel habit to alosetron 1 milligram twice daily or placebo for twelve weeks. Adequate relief of abdominal pain and discomfort for all three months was reported by 41 percent on alosetron against 29 percent on placebo, a difference of 12 percentage points (95 percent confidence interval 4.7 to 19.2); urgency and stool frequency fell and stools firmed ⁵⁰. Constipation occurred in 30 percent on alosetron against 3 percent on placebo, and drop-outs were 24 against 16 percent, mostly for that reason; the authors described the drug as well tolerated ⁵⁰.

It was approved in the United States early in 2000 and withdrawn voluntarily in November of the same year, after post-marketing reports of ischemic colitis, a loss of blood supply to the colon, and of serious complications of constipation ^26,27. That sequence, from approval to withdrawal in under a year, became a case study in drug regulation. The British Medical Journal's account framed the question that the episode posed and did not answer: whether the drug's reintroduction represented regulatory capture by the manufacturer or a victory for patients who had organized to demand it back ²⁶. It returned in 2002 under a Risk Management Plan, with a narrower indication, restricted to women with severe diarrhea-predominant disease who have not responded to conventional therapy, and a formal prescribing program ^26,27.

Whether the restrictions worked is a genuinely difficult question, and the best data come from the manufacturer's own safety surveillance, which is a conflict a reader should hold in mind. Between November 2002 and June 2008, 29,072 patients received 203,939 prescriptions under the Risk Management Plan. Incidence rates were 0.95 cases of ischemic colitis and 0.36 cases of serious constipation complication per 1000 patient-years, which the authors noted were similar to the rates before withdrawal; in clinical trials conducted since reintroduction, ischemic colitis occurred in 4 per 1000 patients on alosetron, 998 patients, against 2 per 1000 on placebo, 411 patients, and serious constipation complications in 2 against 0 per 1000 ⁹¹. No mesenteric ischemia, that is loss of blood supply to the intestine itself, and no surgery, transfusion or death occurred among the ischemic colitis cases, and all improved on prompt withdrawal of the drug ⁹¹. In other words, the rate of the event did not fall; what the Risk Management Plan appears to have changed is how quickly it was recognized and stopped. An independent review of the same period concluded that the drug was being used far less than its efficacy warranted, partly from fear of the plan itself ²⁷.

A 2025 analysis of the United States adverse event reporting database, covering reports from each drug's approval through 30 June 2024 and excluding reports that named other suspect drugs, counted 3832 adverse events attributed to alosetron. Constipation was the commonest, in 2007 events; colitis appeared in 235, ischemic colitis in 140, obstruction in 110 and perforation in 26 ⁹². The percentages the paper prints beside those counts, 23.1 percent for constipation, 2.7 for colitis, 1.6 for ischemic colitis, 1.3 for obstruction and 0.3 for perforation, are calculated on a larger denominator than the 3832 events and cannot be read as shares of them ⁹². These are counts of reports, not incidences, and a drug prescribed under a formal risk program is reported on more assiduously than one that is not; the comparison the authors draw is with the distinctive hazards of the other two approved drugs for the same condition, pancreatitis with eluxadoline, a drug acting on opioid receptors in the gut, and infection with rifaximin, an antibiotic that stays inside the bowel ⁹².

The class lesson is that the risk belongs to the drug and the dose, not automatically to the receptor. Ondansetron, used off-label for the same condition, produces the same therapeutic effect through the same receptor, and its characteristic adverse effect is also constipation: across twenty-one randomized trials of the class, constipation was 3.71 times as likely as on placebo, and the subgroup analysis found that the excess of adverse events was driven by alosetron while ramosetron's rate did not differ significantly from placebo, with serious adverse events rare overall ¹³. Nobody has run a trial large enough to exclude a small ischemic risk with ondansetron in the bowel, and the honest position is that none has been demonstrated, not that none exists ^13,30.

Cannabinoid hyperemesis: where blocking the channel does not work

Cannabinoid hyperemesis syndrome is cyclical, violent vomiting in people who use cannabis heavily and daily, classically relieved by hot showers or baths, and cured only by stopping cannabis. It has had a diagnostic category of its own since the Rome IV criteria of 2016, which define cannabis-induced cyclic vomiting as cannabinoid hyperemesis syndrome and treat it as a special variant of cyclic vomiting syndrome ⁹³; the Rome V classification of disorders of gut-brain interaction keeps the same three nausea and vomiting subcategories, chronic nausea and vomiting syndrome, cyclic vomiting syndrome and cannabinoid hyperemesis syndrome, the last of which shares nearly everything with the second except the cannabis ^94,95. It is not rare. In a nationally representative survey of 7034 United States adults conducted between May and September 2025, 2.7 percent of all respondents and 17.8 percent of those who used cannabis daily reported symptoms consistent with the syndrome ⁹⁶. Canadian hospital discharge records, 7533 records in all, show rates rising about two and a half fold across nine years, from 1.5 per 100,000 population in 2016-17 to 3.8 per 100,000 in 2024-25, with a mean patient age of 32.4 years; the significant rise ran from 2016-17 to 2020-21 and then stabilized for most subgroups, although rates went on rising across the whole period among youth and among older adults, and administrative coding cannot separate rising recognition from rising incidence ⁹⁷.

The patients present vomiting, and so they are given antiemetics, and the antiemetics do not work. The question was tested properly in 2020 by the HaVOC trial, a triple-blind randomized controlled trial in a Canadian emergency department. Thirty-three cannabis users with active vomiting were randomized to intravenous haloperidol, with a nested randomization to 0.05 or 0.1 milligrams per kilogram, or to ondansetron 8 milligrams; 30 received at least one treatment, 13 haloperidol and 17 ondansetron. The participants were young, a mean of 29 years, using a mean of 1.5 grams a day since the age of 19. At two hours, haloperidol was superior by 2.3 centimeters on the combined 10 centimeter visual analog scales for abdominal pain and nausea (95 percent confidence interval 0.6 to 4.0, p equals 0.01), with less use of rescue antiemetics, 31 against 59 percent, a difference of 28 percentage points whose 95 percent confidence interval, from minus 61 to plus 13 percentage points, includes no difference at all, and a shorter time to emergency department departure, 3.1 against 5.6 hours (p equals 0.03) ¹⁹. Two patients returned with acute dystonia, both in the higher-dose haloperidol group ¹⁹. The authors drew the mechanistic inference explicitly: the superiority of a dopamine antagonist over a 5-HT3 antagonist is information about what this illness is ¹⁹.

The finding had been anticipated by case reports. A single case of haloperidol resolving the syndrome after conventional antiemetics failed was published in 2013 ⁹⁸, followed by a four-patient series in which haloperidol improved patients who had failed standard emergency department therapy ⁹⁹. Droperidol, haloperidol's shorter-acting relative, has accumulated a consistent observational record: in an eleven-year Australian single-center review, covering 2006 to 2016, of 76 presentations meeting diagnostic criteria, median length of stay was 6.7 hours in the 37 presentations treated with droperidol against 13.9 hours in the 39 that were not (p equals 0.014), and ondansetron and metoclopramide were used twice as often in the group that did not receive it ¹⁰⁰. A 2025 retrospective study of 211 encounters at two academic emergency departments in Atlanta found droperidol, at a median dose of 1.25 milligrams, associated with a reduced length of stay, adjusted mean ratio 0.76 (0.62 to 0.94), less total medication and markedly less opioid use, with two mild adverse reactions ¹⁰¹. A prospective multicenter cohort of 47 patients given droperidol with diphenhydramine, a sedating antihistamine, recorded symptom scores falling from 8.3 at baseline to 3.1 at 30 minutes and 1.4 at 120 minutes, with 12.9 percent returning within seven days ¹⁰². These are observational and uncontrolled; droperidol has not had its HaVOC. There is also a trade-off that the enthusiasm obscures. The same 2026 review that makes ondansetron the most QT-prolonging of the setrons ranks the butyrophenones, haloperidol and droperidol among them, and the phenothiazines above the setrons for repolarization liability ¹⁷, which is why the doses used here are small, a median of 1.25 milligrams of droperidol in the Atlanta series ¹⁰¹, and why electrolytes and an electrocardiogram matter in a patient who has been vomiting for days ^17,64.

The other thing that works is odd and instructive: capsaicin cream, the pepper extract, rubbed on the abdomen. It is an agonist at TRPV1, the heat and capsaicin receptor on sensory nerves, and the hypothesis is that it reproduces pharmacologically what the hot shower does. A systematic review and meta-analysis pooled seven studies with 106 patients and found a mean time to symptom resolution after application of 325 minutes (95 percent confidence interval 234 to 787) and an emergency department length of stay of 379 minutes (10 to 747); the intervals are wide, the studies small, and the authors call for randomized trials ¹⁰³. A German narrative review summarizes current practice as hot showers, intramuscular haloperidol 5 milligrams for acute severe vomiting, and capsaicin cream at 0.075 to 0.1 percent as an antiemetic that works less quickly, with the standard antiemetics often simply not helping ⁹³.

Why 5-HT3 blockade fails here is not known, and the leading simple explanations have been tested and have not held. If the syndrome were a dose effect of accumulated tetrahydrocannabinol, affected patients should carry more of it; hair analysis in 46 subjects with positive urine found cannabinoid concentrations in patients with the syndrome indistinguishable from recreational users without vomiting and from emergency department controls, with only the ratio of tetrahydrocannabinol to cannabinol differing, 2.6-fold lower in cases, suggesting that heavy use is necessary but not sufficient ¹⁰⁴. A 2026 study measured serum concentrations of tetrahydrocannabinol, its metabolites and the non-psychotropic phytocannabinoids in 35 emergency patients with the syndrome and 15 with simple acute intoxication and found no significant differences in any of them or their ratios ¹⁰⁵. What is left is a syndrome that looks like a vomiting disorder, responds better to a dopamine blocker than to the 5-HT3 blockade that dominates chemotherapy vomiting, and may be relieved by stimulating TRPV1 ^19,93,103. The only treatment that resolves it permanently is sustained abstinence from cannabis ⁹³.

The serotonin syndrome warning, and how thin the evidence for it is

Serotonin syndrome, better called serotonin toxicity, is the clinical state produced by too much serotonin acting in the central nervous system: agitation and confusion, autonomic disturbance with fever, sweating and a racing pulse, and the characteristic neuromuscular signs of hyperreflexia, meaning reflexes that are exaggerated when tested, with tremor and clonus. It is produced by drugs that increase serotonin in the synapse, most dangerously by combining a monoamine oxidase inhibitor with a serotonin reuptake inhibitor, and in overdose of serotonergic antidepressants it is common, with moderate toxicity in about 14 percent of poisonings with a selective serotonin reuptake inhibitor ¹⁰⁶.

Regulators have added a serotonin syndrome warning to the labeling of ondansetron ¹⁰⁶. The pharmacological reasoning runs the wrong way. Ondansetron does not raise serotonin anywhere; it blocks one postsynaptic receptor, and that receptor is not the one that mediates the syndrome. For a drug to cause serotonin toxicity it must act, by one route or another, to increase serotonergic neurotransmission, and ondansetron does the opposite at its therapeutic target ¹⁰⁶.

The evidence that prompted the warning consists of case reports in patients taking several serotonergic drugs at once. A typical example: an elderly woman who developed the syndrome after general anesthesia while taking paroxetine, duloxetine and bupropion, and receiving fentanyl and ondansetron ¹⁰⁷. A second: a woman on peritoneal dialysis taking duloxetine and tramadol who developed hyperreflexia and inducible clonus and whose ondansetron, given for the nausea the syndrome itself was causing, was listed among the serotonergic agents and withdrawn along with the rest; the authors of that report concluded that the ondansetron they had given had likely made the syndrome worse ¹⁰⁸. In both, several drugs that genuinely do increase synaptic serotonin were present, and in the second the nausea that prompted the ondansetron was a symptom of the syndrome already under way.

A 2022 review of serotonin toxicity diagnostic criteria set out the problem in the general case and named this drug specifically. Of 55 published fatal cases reviewed, 12, or 22 percent, were judged unlikely to have been serotonin toxicity at all; the Sternbach and Radomski criteria were met by 25, or 45 percent, and 20 had insufficient data reported; only 13 could be confirmed as meeting the more specific Hunter criteria, because 42 lacked the necessary detail ¹⁰⁶. The authors' conclusion was that diagnosing the syndrome from symptom criteria alone, without asking whether the drugs taken can plausibly raise central serotonin, has produced case reports and government warnings for drugs that cannot plausibly cause significant serotonin toxicity, and they give ondansetron and the antipsychotics as their examples ¹⁰⁶.

The practical position, then, is this. The warning exists. The mechanism is implausible. The cases are confounded by co-prescription of drugs that do cause the syndrome and by the fact that nausea is itself one of its features, which guarantees that an antiemetic will often have been given shortly before the diagnosis is made. A clinician treating a patient with serotonin toxicity should stop the serotonergic drugs; whether ondansetron belongs on that list is, on the published evidence, doubtful ¹⁰⁶. This is not a reason to be cavalier: the same review's point is that better reporting is needed, not that the question is closed.

Cyclic vomiting syndrome, where the setrons are an option and not an answer

Cyclic vomiting syndrome is a disorder of gut-brain interaction in which a person is well between attacks and then, with little warning, vomits relentlessly for hours or days. It affects children and adults, is closely associated with migraine, and in the United States has an estimated prevalence of about 2 percent ¹⁰⁹. Its mechanism is not known; genetic, autonomic, neurohormonal and environmental contributions have all been proposed ¹⁰⁹. Rome V classifies it among the nausea and vomiting disorders, alongside chronic nausea and vomiting syndrome and cannabinoid hyperemesis syndrome, which has been listed separately since Rome IV ^93,94,95.

The adult guideline, developed jointly by the American Neurogastroenterology and Motility Society and the Cyclic Vomiting Syndrome Association using the GRADE framework, strongly recommends a tricyclic antidepressant, usually amitriptyline, as first-line prevention for moderate to severe disease, with topiramate, a drug developed for epilepsy and used to prevent migraine, or aprepitant as alternatives ³². For aborting an attack, it recommends serotonin antagonists such as ondansetron, and triptans or aprepitant, conditionally ³². The word conditional carries weight: the accompanying evidence review graded the evidence for most of these agents as very low quality, with moderate certainty only for triptans as abortive therapy, and described the evidence supporting ondansetron specifically as limited ¹¹⁰.

The 2025 pediatric guidelines from the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition reached the same shape of conclusion by the same method. Sixteen recommendations were formulated; the only strong recommendation for aborting an attack was for anti-migraine agents in children with a personal or family history of migraine, while oral and intravenous 5-HT3 and neurokinin-1 receptor antagonists were recommended conditionally ¹¹¹. A companion diagnostic guideline recommends screening blood and urine tests and an upper gastrointestinal radiographic series for all children with suggestive symptoms, finds other screening tests low-yield, and sets out a separate consensus approach to pediatric cannabinoid hyperemesis as a related but distinct entity ¹¹².

Why should a drug that abolishes cisplatin vomiting only partly help here? Probably because the trigger is not a flood of serotonin from the gut lining. The clinical picture argues the same way: the drugs that prevent attacks are antimigraine and antidepressant agents acting centrally, and the drugs that abort them are a mixture ^32,109. There is a thread of receptor genetics. A case-control study comparing 70 patients with cyclic vomiting syndrome against 2504 controls from a public reference genome collection found that specific genotypes in HTR3B and HTR3C were associated with risk, with one HTR3B genotype increasing it and particular alleles in HTR3B and HTR3C decreasing it ⁴⁶. The design, a small clinical series against a public panel rather than a matched control group, makes this hypothesis-generating; it is the first report to implicate these genes in the disease and needs replication.

Two practical observations round this out. Comorbidity is the rule, and the adult guideline recommends screening for and treating anxiety, depression, migraine, autonomic dysfunction, sleep disorders and substance use, with referral where needed ³². And distinguishing cyclic vomiting syndrome from cannabinoid hyperemesis is often impossible at the first visit, because about a third of patients with the former also use cannabis, sometimes therapeutically; the distinction can only be made reliably by observing full remission during six to twelve months of abstinence ^93,109.

What was believed and is not, and what is still unknown

Believed and abandoned: that serotonin released into the bloodstream reaches the area postrema, where the blood-brain barrier is incomplete, and triggers vomiting from there. That was the natural reading of the 1990 observation that urinary 5-HIAA rose with cisplatin ²⁸. Continuous intravenous microdialysis in sixteen patients then showed that after cisplatin the metabolite in blood roughly doubled while free serotonin in the blood did not rise at all, which argues against an action of circulating serotonin on the area postrema and for serotonin released inside the intestinal wall acting locally on vagal endings ⁴⁷. Both sites matter for the drug, since antagonists microinjected into the area postrema also work ⁹; what fell was the idea that the blood carries the signal.

Believed and abandoned: that 5-HT3 blockade is general-purpose antiemesis. The HaVOC trial showed that in cannabinoid hyperemesis a dopamine antagonist beats ondansetron outright ¹⁹; in cyclic vomiting syndrome the guideline recommendation for ondansetron is conditional and rests on limited evidence ^32,110; in hyperemesis gravidarum the one placebo-controlled trial did not show separation ¹⁸; and in postoperative nausea, ondansetron is one of three interchangeable options each worth about a 26 percent relative risk reduction ²². The receptor is the dominant pathway for chemotherapy and radiotherapy emesis and for a good deal of postoperative and gastroenteritis vomiting, and it is one pathway among several everywhere else.

Believed and abandoned: that ondansetron in early pregnancy causes heart defects. The Swedish registry signal of cardiac septal defects ⁵² did not survive the 1.8 million pregnancy Medicaid cohort, where the adjusted relative risk for cardiac malformation was 0.99 ⁵³, nor the multinational active-comparator cohort ⁷⁷, nor sensitivity analysis in the largest meta-analysis ⁷⁸. What remains is a small oral cleft association, 11.1 against 14.0 per 10,000 births, seen in the Medicaid cohort and in one meta-analysis ^53,76 but not in two others ^75,78, and possibly explained by who agrees to take part in case-control studies ⁸².

Weakly believed and still in the labeling: that ondansetron causes serotonin syndrome. The drug blocks a postsynaptic receptor and does not raise synaptic serotonin; the published cases involve several genuinely serotonergic drugs at once; and a narrative review of diagnostic criteria names it as an example of a drug for which warnings exist despite physiological implausibility ^106,107,108.

Still unknown, first: what the receptor actually looks like in the body. Every solved structure is of a homomeric receptor made of A subunits, often stabilized by antibody fragments and studied outside a native membrane, while the receptors that matter in people are heteromers; obtaining heteropentameric structures in a native membrane is named as the central unfinished task in the most recent review ³⁵. Second: what the C, D and E subunits do. Their expression is restricted to a few organs ⁶, their genetic variants track with drug response and disease ^31,44,46, and what they contribute to a channel is so far a matter of computational modeling ⁴³.

Still unknown, third: why palonosetron is better in the delayed phase. The internalization and allosteric-binding account came from a commercially linked laboratory ^38,39; independent structures place palonosetron in the same pocket as the other setrons and call all of them competitive antagonists ⁴⁰; and the clinical superiority the mechanism was offered to explain was confirmed in a 2026 meta-analysis of eleven trials ⁴¹. Fourth, and most consequential for patients now: why 5-HT3 blockade fails in cannabinoid hyperemesis. The two simplest hypotheses, accumulation of tetrahydrocannabinol and differential exposure to other cannabinoids, have both been tested against measured concentrations in blood and hair and have not held up ^104,105; meanwhile the condition is common, with 2.7 percent of surveyed United States adults and 17.8 percent of daily cannabis users reporting compatible symptoms in 2025 ⁹⁶, while Canadian hospitalization rates rose about two and a half fold to 2024-25, mostly before 2020-21 and stable since for most subgroups ⁹⁷.

Finally, a question that has only just been opened. If the nausea and vomiting of pregnancy is driven by fetal GDF15 acting on brainstem receptors, with maternal prior sensitivity determining severity ⁵⁵, then the most-prescribed drug for it blocks a receptor that may lie well downstream of the cause, which would explain why it failed to separate from placebo in the only trial that tried ¹⁸. Two findings since have begun to test that idea. In a questionnaire study of 5414 people recruited through a hyperemesis patient foundation, daily metformin in the month before a first pregnancy, which raises circulating GDF15 and should therefore desensitize the receiving system, was associated with a reduction of more than 70 percent in the risk of hyperemesis gravidarum, adjusted relative risk 0.29 (95 percent confidence interval 0.12 to 0.71); selective serotonin reuptake inhibitors before the first pregnancy, and cannabis use before a second, were associated with higher risk. The study is self-reported, recall-based and self-selected, which the authors state, and it is a reason to run a trial rather than to prescribe ¹¹³. And a 2026 longitudinal study, using serial samples from two Danish early-pregnancy cohorts, one of them the VOMIT trial itself, showed that GDF15 rises steeply to about week 7 and plateaus by week 9, that levels are roughly 19 percent higher with a female fetus, and that although hyperemesis pregnancies run higher early their confidence intervals overlap those of normal pregnancies by week 8, which limits the hormone's use as a diagnostic test for now ¹¹⁴. Whether a GDF15-directed treatment works in people is still unproven, but it is now a question with trials in prospect rather than a hypothesis alone. It is the most interesting open question about this receptor, and it is a question about something else.

The tests, one by one

• Radioligand bindingA radioactively labeled antagonist is applied to tissue slices and the radioactivity mapped, showing where the receptor sits. In human medulla oblongata, tritium-labeled granisetron bound saturably in the dorsal vagal complex with a maximum binding capacity of 1.88 picomoles per milligram of protein and a dissociation constant of 1.21 nanomolar, and nowhere else in the medulla ¹⁰.Locating the receptor in tissue and measuring how tightly a candidate drug binds.Excellent for affinity and anatomy; the 1989 human study resolved a single medullary region cleanly ¹⁰.Cannot tell which subunits make up the receptor it is labeling, and says nothing about whether the channel opens.
• Patch-clamp electrophysiologyA glass pipette seals onto a patch of membrane and records current through single channels. This is how the receptor was proven to be a channel at all: in guinea-pig submucous plexus neurons, serotonin opened channels of 15 and 9 picosiemens that were blocked by 5-HT3 antagonists, while acetylcholine opened a 40 picosiemens channel ².Distinguishing homomeric from heteromeric receptors and measuring what a drug does to function.The only method that measures the thing the receptor actually does; it separated the sub-picosiemens A-only receptor from the 16 picosiemens A-plus-B receptor ⁷.Done in cells or expression systems, not in patients; a recording says nothing about what the receptor is doing in a person being sick.
• X-ray crystallography and cryo-electron microscopyPurified receptor is frozen or crystallized and its atomic structure reconstructed. The mouse receptor was solved by X-ray at 3.5 angstroms in 2014 ⁴, then by cryo-electron microscopy in resting ³³, serotonin-bound ⁸ and drug-bound forms at 3.2 to 4.5 angstroms ³⁴, and in complex with palonosetron, ondansetron and alosetron ⁴⁰.Showing how serotonin opens the channel and how each setron blocks it.Resolutions of 3.2 to 4.5 angstroms resolve side chains in the binding pocket but not every water molecule ³⁴.Every published structure is of a homomeric A-subunit receptor, often stabilized by antibody fragments; no structure of the heteromeric receptor that predominates in the body has been solved ³⁵.
• HTR3 genotypingSequencing the five genes and relating variants to drug response. A curated database collects the known variants of HTR3A to HTR3E with their functional and clinical associations ⁴⁴. In 242 patients starting chemotherapy, those homozygous for a three-base deletion in the HTR3B promoter vomited more often than the rest ⁴⁵.Predicting who will fail a setron, and testing whether receptor variation contributes to irritable bowel syndrome, cyclic vomiting and psychiatric illness.Associations are real but small; the authors of the chemotherapy study wrote that only a small fraction of therapeutic failure is explained by the variant they found ⁴⁵.Most of the variation in who vomits; and the 2026 study of CYP2D6, one of the liver enzymes that metabolize ondansetron, found no significant effect of metabolizer status on response in 264 pregnant patients ⁹⁰, although pharmacogenetic guidance still gives genotype-directed advice for this drug class ⁸⁹.
• The bedside test: does a setron workThere is no clinical assay for this receptor. In practice the receptor is tested by giving a drug that blocks it and watching, which is how the field learned that in cannabis hyperemesis ondansetron is beaten by haloperidol ¹⁹, while cisplatin vomiting responds well ¹⁴.Everyday clinical decisions, and, retrospectively, as evidence about mechanism.A blunt instrument, but the HaVOC trial shows it can be done properly, with randomization and blinding ¹⁹.Everything about why a given patient responds; a failure to respond can mean the wrong receptor, the wrong dose, or the wrong diagnosis.

Where it connects

Sources

114 sources, numbered as they are cited. Every one was checked against PubMed or its publisher before it was cited here; the note under each says what it shows and what it does not.

1. 1
   Bradley PB, Engel G, Feniuk W, Fozard JR, Humphrey PP, Middlemiss DN, Mylecharane EJ, Richardson BP, Saxena PR. Proposals for the classification and nomenclature of functional receptors for 5-hydroxytryptamine.Neuropharmacology · 1986 · 25(6):563-76doi:10.1016/0028-3908(86)90207-8 · PMID 2875415

   The committee paper that named the 5-HT3 receptor, explicitly identifying it with Gaddum's M receptor. It notes that no brain binding studies with 5-HT3-selective radioligands had then been published. A classification, not new data.

2. 2
   Derkach V, Surprenant A, North RA. 5-HT3 receptors are membrane ion channels.Nature · 1989 · 339(6227):706-9doi:10.1038/339706a0 · PMID 2472553

   The first single-channel recordings of serotonin-gated currents, in excised patches from guinea-pig submucous plexus neurons: channels of 15 and 9 picosiemens blocked by 5-HT3 antagonists, against a 40 picosiemens acetylcholine channel. Establishes that this receptor is a channel, in guinea pig.

3. 3
   Lummis SCR. 5-HT(3) receptors.J Biol Chem · 2012 · 287(48):40239-45doi:10.1074/jbc.R112.406496 · PMID 23038271

   Concise authoritative review: cation-selective Cys-loop receptors in central and peripheral nervous systems; five subunits A to E; all functional receptors require at least one A subunit; the setrons are the standard of care for chemotherapy-induced and postoperative nausea and vomiting. A review.

4. 4
   Hassaine G, Deluz C, Grasso L, Wyss R, Tol MB, Hovius R, Graff A, Stahlberg H, Tomizaki T, Desmyter A, Moreau C, Li XD, Poitevin F, Vogel H, Nury H. X-ray structure of the mouse serotonin 5-HT3 receptor.Nature · 2014 · 512(7514):276-81doi:10.1038/nature13552 · PMID 25119048

   The first structure of a mammalian Cys-loop receptor, at 3.5 angstroms: neurotransmitter site, an aqueous pore with a 4.6 angstrom constriction, and five intracellular helices forming a closed vestibule. A proteolyzed receptor stabilized by nanobodies, not a native membrane preparation.

5. 5
   Weiss B, Mertz A, Schröck E, Koenen M, Rappold G. Assignment of a human homolog of the mouse Htr3 receptor gene to chromosome 11q23.1-q23.2.Genomics · 1995 · 29(1):304-5doi:10.1006/geno.1995.1254 · PMID 8530095

   Maps the human HTR3A gene by fluorescence in situ hybridization and somatic cell hybrids. A short mapping report; the PubMed record carries no abstract.

6. 6
   Niesler B, Frank B, Kapeller J, Rappold GA. Cloning, physical mapping and expression analysis of the human 5-HT3 serotonin receptor-like genes HTR3C, HTR3D and HTR3E.Gene · 2003 · 310:101-11doi:10.1016/s0378-1119(03)00503-1 · PMID 12801637

   Clones HTR3D and HTR3E, maps all three of HTR3C, HTR3D and HTR3E to a region under 100 kilobases on chromosome 3q27, and reports expression restricted to kidney, colon and liver for HTR3D and to colon and intestine for HTR3E. Expression by tissue messenger RNA, not protein.

7. 7
   Davies PA, Pistis M, Hanna MC, Peters JA, Lambert JJ, Hales TG, Kirkness EF. The 5-HT3B subunit is a major determinant of serotonin-receptor function.Nature · 1999 · 397(6717):359-63doi:10.1038/16941 · PMID 9950429

   Identifies the second subunit and shows why it matters: homomeric A receptors conduct under 1 picosiemens against 9-17 picosiemens for neuronal receptors, while A-plus-B heteromers conduct 16 picosiemens with low calcium permeability and native-like current-voltage behavior. Expression systems, not native tissue.

8. 8
   Basak S, Gicheru Y, Rao S, Sansom MSP, Chakrapani S. Cryo-EM reveals two distinct serotonin-bound conformations of full-length 5-HT3A receptor.Nature · 2018 · 563(7730):270-274doi:10.1038/s41586-018-0660-7 · PMID 30401837

   Two serotonin-bound states at 3.32 and 3.89 angstroms, showing a large twisting motion that opens a 165 angstrom permeation pathway and creates lateral portals between the membrane and intracellular domains. Homomeric mouse receptor in detergent, with molecular dynamics.

9. 9
   Higgins GA, Kilpatrick GJ, Bunce KT, Jones BJ, Tyers MB. 5-HT3 receptor antagonists injected into the area postrema inhibit cisplatin-induced emesis in the ferret.Br J Pharmacol · 1989 · 97(1):247-55doi:10.1111/j.1476-5381.1989.tb11948.x · PMID 2720310

   High concentrations of 5-HT3 binding in the ferret area postrema, and discrete microinjection of ondansetron (0.01-1 microgram) or two other antagonists into that region inhibited cisplatin emesis, while ketanserin and methiothepin did not. Ferret; identifies a central site of action. All five authors worked at Glaxo Group Research, which developed ondansetron.

10. 10
    Reynolds DJ, Leslie RA, Grahame-Smith DG, Harvey JM. Localization of 5-HT3 receptor binding sites in human dorsal vagal complex.Eur J Pharmacol · 1989 · 174(1):127-30doi:10.1016/0014-2999(89)90884-4 · PMID 2612576

    In vitro autoradiography of human medulla oblongata: saturable, displaceable binding in the dorsal vagal complex with maximum binding 1.88 picomoles per milligram protein and dissociation constant 1.21 nanomolar, and in no other medullary region. Post-mortem human tissue.

11. 11
    Gaddum JH, Picarelli ZP. Two kinds of tryptamine receptor.Br J Pharmacol Chemother · 1957 · 12(3):323-8doi:10.1111/j.1476-5381.1957.tb00142.x · PMID 13460238

    The origin of the receptor. In guinea-pig ileum, serotonin acts at a morphine-blockable M receptor, probably in nervous tissue, and a dibenzyline-blockable D receptor, probably in muscle. Pharmacology only; no molecular identification.

12. 12
    Maricq AV, Peterson AS, Brake AJ, Myers RM, Julius D. Primary structure and functional expression of the 5HT3 receptor, a serotonin-gated ion channel.Science · 1991 · 254(5030):432-7doi:10.1126/science.1718042 · PMID 1718042

    The cloning paper. A complementary DNA from a mouse neuroblastoma expression library, identified by serotonin-gated currents in Xenopus oocytes, encodes a protein with ligand-gated ion channel features; messenger RNA in brain, spinal cord and heart. Mouse sequence; human genes mapped later.

13. 13
    Zheng Y, Yu T, Tang Y, Xiong W, Shen X, Jiang L, Lin L. Efficacy and safety of 5-hydroxytryptamine 3 receptor antagonists in irritable bowel syndrome: A systematic review and meta-analysis of randomized controlled trials.PLoS One · 2017 · 12(3):e0172846doi:10.1371/journal.pone.0172846 · PMID 28291778

    21 randomized trials, 16 rated high quality. Global symptom improvement relative risk 1.56 (1.43-1.71); abdominal pain 1.33 (1.26-1.39); constipation 3.71 (2.98-4.61) against placebo. Subgroup analysis found the excess of adverse events driven by alosetron, with ramosetron not significantly different from placebo, and serious adverse events rare. Pools different drugs and doses.

14. 14
    Marty M, Pouillart P, Scholl S, Droz JP, Azab M, Brion N, Pujade-Lauraine E, Paule B, Paes D, Bons J. Comparison of the 5-hydroxytryptamine3 (serotonin) antagonist ondansetron (GR 38032F) with high-dose metoclopramide in the control of cisplatin-induced emesis.N Engl J Med · 1990 · 322(12):816-21doi:10.1056/NEJM199003223221205 · PMID 2137902

    Double-blind crossover in 97 chemotherapy-naive patients, 76 completing both arms: complete or near-complete control in 57 of 76 treatments (75 percent) with ondansetron against 32 of 76 (42 percent) with metoclopramide, p below 0.001, with dystonic reactions in three metoclopramide treatments. Acute phase only.

15. 15
    Kuryshev YA, Brown AM, Wang L, Benedict CR, Rampe D. Interactions of the 5-hydroxytryptamine 3 antagonist class of antiemetic drugs with human cardiac ion channels.J Pharmacol Exp Ther · 2000 · 295(2):614-20PMID 11046096

    Patch-clamp measurements in cells expressing human channels. Half-maximal block of the hERG potassium channel at 0.81 micromolar for ondansetron, 3.73 granisetron, 5.95 dolasetron, 12.1 for dolasetron's active metabolite; all blocked the cardiac sodium channel frequency-dependently. In vitro; the authors conclude ondansetron's submicromolar hERG affinity likely underlies its repolarization effect. No DOI in the PubMed record.

16. 16
    Zuo P, Haberer LJ, Fang L, Hunt TL, Ridgway D, Russo MW. Integration of modeling and simulation to support changes to ondansetron dosing following a randomized, double-blind, placebo-, and active-controlled thorough QT study.J Clin Pharmacol · 2014 · 54(11):1221-9doi:10.1002/jcph.322 · PMID 24782199

    The formal thorough QT study, with placebo and moxifloxacin control. Maximum mean placebo-corrected QTcF under 10 milliseconds after 8 mg intravenously and about 20 milliseconds after 32 mg infused over 15 minutes; modeling predicted 9.2 milliseconds (upper 90 percent bound 11.2) for 16 mg. Concludes single intravenous doses above 16 mg should no longer be used. Authors employed by GlaxoSmithKline, which marketed ondansetron.

17. 17
    Moustafa AT, Dabash O, Panigrahi PK, Khan HR. Antiemetic-Associated QT Prolongation: A Clinical Review and Risk-Stratified Prescribing Framework for Long QT Syndrome.Eur Heart J Cardiovasc Pharmacother · 2026 · online ahead of printdoi:10.1093/ehjcvp/pvag077 · PMID 42748473

    Narrative review through early 2026 with a three-tier prescribing framework. Attributes antiemetic QT liability principally to hERG blockade, notes the 2012 Food and Drug Administration restriction on the 32 mg intravenous ondansetron dose, reports palonosetron as having minimal effect, and locates serious events in patients with electrolyte disturbance, bradycardia, polypharmacy or inherited long QT, especially above 500 milliseconds or a rise over 60 milliseconds. A narrative review, not a systematic one.

18. 18
    Ostenfeld A, Carlsen SE, Jensen AK, Futtrup TB, Westergaard HB, Pedersen LH, Andersen JT, Petersen TS, Løkkegaard ECL. Mirtazapine or ondansetron for hyperemesis gravidarum: a randomized placebo-controlled trial.Am J Obstet Gynecol · 2026 · 234(6):1700-1729doi:10.1016/j.ajog.2025.12.061 · PMID 41478546

    The VOMIT trial: 59 pregnant people in seven Danish hospitals randomized 1:1:1 to mirtazapine, ondansetron or placebo for 14 days, 2019-2022. Change in Pregnancy-Unique Quantification of Emesis score at day 2 against placebo: mirtazapine minus 1.86 (minus 3.61 to minus 0.12, p=0.04); ondansetron minus 0.51 (minus 2.32 to 1.30), not significant. The first placebo-controlled test of ondansetron in hyperemesis gravidarum; with 59 participants it cannot exclude a modest benefit, but it did not demonstrate one.

19. 19
    Ruberto AJ, Sivilotti MLA, Forrester S, Hall AK, Crawford FM, Day AG. Intravenous Haloperidol Versus Ondansetron for Cannabis Hyperemesis Syndrome (HaVOC): A Randomized, Controlled Trial.Ann Emerg Med · 2021 · 77(6):613-619doi:10.1016/j.annemergmed.2020.08.021 · PMID 33160719

    The only randomized controlled trial. 33 cannabis users with active emesis randomized triple-blind to intravenous haloperidol, nested at 0.05 or 0.1 mg/kg, or ondansetron 8 mg; 30 received treatment (13 haloperidol, 17 ondansetron), mean age 29, using 1.5 g daily since age 19. At two hours haloperidol was superior by 2.3 cm on combined 10 cm scales for pain and nausea (0.6-4.0, p=0.01), with less rescue medication (31 against 59 percent) and shorter time to departure (3.1 against 5.6 hours, p=0.03). Two returned with acute dystonia, both on the higher haloperidol dose. Small, single-center, but randomized and blinded.

20. 20
    Gralla RJ, Itri LM, Pisko SE, Squillante AE, Kelsen DP, Braun DW, Bordin LA, Braun TJ, Young CW. Antiemetic efficacy of high-dose metoclopramide: randomized trials with placebo and prochlorperazine in patients with chemotherapy-induced nausea and vomiting.N Engl J Med · 1981 · 305(16):905-9doi:10.1056/NEJM198110153051601 · PMID 7024807

    Two double-blind trials in 41 patients on cisplatin: median emetic episodes 1.0 with metoclopramide against 10.5 with placebo and 1.5 against 12.0 with prochlorperazine. Sedation was frequent and one patient had an extrapyramidal reaction. The state of the art before the setrons.

21. 21
    Hesketh PJ, Kris MG, Basch E, Bohlke K, Barbour SY, Clark-Snow RA, Danso MA, Dennis K, Dupuis LL, Dusetzina SB, Eng C, Feyer PC, Jordan K, Noonan K, Sparacio D, Lyman GH. Antiemetics: ASCO Guideline Update.J Clin Oncol · 2020 · 38(24):2782-2797doi:10.1200/JCO.20.01296 · PMID 32658626

    The American Society of Clinical Oncology guideline, based on randomized trials and meta-analyses published to January 2020, setting out combination prophylaxis by emetic risk and adding olanzapine options. A guideline; recommendations, not new data.

22. 22
    Apfel CC, Korttila K, Abdalla M, Kerger H, Turan A, Vedder I, Zernak C, Danner K, Jokela R, Pocock SJ, Trenkler S, Kredel M, Biedler A, Sessler DI, Roewer N. A factorial trial of six interventions for the prevention of postoperative nausea and vomiting.N Engl J Med · 2004 · 350(24):2441-51doi:10.1056/NEJMoa032196 · PMID 15190136

    5199 high-risk patients enrolled, 4123 randomized among 64 combinations. Ondansetron 4 mg, dexamethasone 4 mg and droperidol 1.25 mg each reduced risk by about 26 percent; propofol by 19 percent and nitrogen by 12 percent; effects were independent and multiplicative. Shows ondansetron is one interchangeable option, not a privileged one.

23. 23
    Freedman SB, Adler M, Seshadri R, Powell EC. Oral ondansetron for gastroenteritis in a pediatric emergency department.N Engl J Med · 2006 · 354(16):1698-705doi:10.1056/NEJMoa055119 · PMID 16625009

    215 children aged 6 months to 10 years. A single orally disintegrating dose cut vomiting during oral rehydration from 35 to 14 percent (relative risk 0.40, 0.26-0.61) and intravenous rehydration from 31 to 14 percent (0.46, 0.26-0.79). Hospitalization and return visits did not differ.

24. 24
    Nelson-Piercy C, Dean C, Shehmar M, Gadsby R, O'Hara M, Hodson K, Nana M. The Management of Nausea and Vomiting in Pregnancy and Hyperemesis Gravidarum (Green-top Guideline No. 69).BJOG · 2024 · 131(7):e1-e30doi:10.1111/1471-0528.17739 · PMID 38311315

    Royal College of Obstetricians and Gynaecologists guideline. Grades the evidence that ondansetron is safe and effective as B, states its use as a second-line antiemetic should not be discouraged, that women can be reassured about a very small absolute increase in orofacial clefting balanced against the risks of poorly managed hyperemesis, and that ketonuria is not an indicator of dehydration and should not be used to assess severity. A guideline.

25. 25
    Zeitz KP, Guy N, Malmberg AB, Dirajlal S, Martin WJ, Sun L, Bonhaus DW, Stucky CL, Julius D, Basbaum AI. The 5-HT3 subtype of serotonin receptor contributes to nociceptive processing via a novel subset of myelinated and unmyelinated nociceptors.J Neurosci · 2002 · 22(3):1010-9doi:10.1523/JNEUROSCI.22-03-01010.2002 · PMID 11826129

    Mice with the receptor genetically deleted or pharmacologically blocked show significantly reduced persistent pain after tissue injury but unchanged acute nociception, with receptor transcripts in a previously uncharacterized subset of afferents, few expressing substance P. Mouse work; it has not produced a clinical analgesic.

26. 26
    Moynihan R. Alosetron: a case study in regulatory capture, or a victory for patients' rights?BMJ · 2002 · 325(7364):592-5doi:10.1136/bmj.325.7364.592 · PMID 12228140

    Journalistic analysis of the withdrawal and reintroduction, setting out the question in its title and the role of organized patient pressure. The PubMed record carries no abstract. A feature article, not research; cited for the regulatory narrative.

27. 27
    Lewis JH. Alosetron for severe diarrhea-predominant irritable bowel syndrome: safety and efficacy in perspective.Expert Rev Gastroenterol Hepatol · 2010 · 4(1):13-29doi:10.1586/egh.09.72 · PMID 20136586

    Review of the regulatory history: approval in early 2000, voluntary withdrawal in November 2000 after ischemic colitis and severe constipation complications, reintroduction in 2002 under a Risk Management Plan with a restricted indication and a prescribing program, and very limited use since, which the author attributes partly to fear of the plan itself. A review by a single author.

28. 28
    Cubeddu LX, Hoffmann IS, Fuenmayor NT, Finn AL. Efficacy of ondansetron (GR 38032F) and the role of serotonin in cisplatin-induced nausea and vomiting.N Engl J Med · 1990 · 322(12):810-6doi:10.1056/NEJM199003223221204 · PMID 1689807

    28 patients, ondansetron 0.15 mg/kg three times against saline: median time to first emesis 11.6 against 2.8 hours, median 24-hour episodes 1.5 against 5.5, rescue needed in 0 of 14 against 12 of 14. Urinary 5-HIAA rose in all patients two to six hours after cisplatin, paralleling emesis. The serotonin mechanism is the authors' inference from the metabolite, not a direct measurement.

29. 29
    Hawthorn J, Ostler KJ, Andrews PL. The role of the abdominal visceral innervation and 5-hydroxytryptamine M-receptors in vomiting induced by the cytotoxic drugs cyclophosphamide and cis-platin in the ferret.Q J Exp Physiol · 1988 · 73(1):7-21doi:10.1113/expphysiol.1988.sp003124 · PMID 3347698

    Bilateral abdominal vagotomy plus greater splanchnic section totally abolished retching and vomiting to intraperitoneal cisplatin or intravenous cyclophosphamide; BRL 24924, a gastrokinetic agent with more specific M-receptor antagonist properties, nearly abolished it, outperforming metoclopramide, which did not fully control either retching or vomiting. Ferret; the authors state a central site cannot be excluded.

30. 30
    Garsed K, Chernova J, Hastings M, Lam C, Marciani L, Singh G, Henry A, Hall I, Whorwell P, Spiller R. A randomised trial of ondansetron for the treatment of irritable bowel syndrome with diarrhoea.Gut · 2014 · 63(10):1617-25doi:10.1136/gutjnl-2013-305989 · PMID 24334242

    120 patients meeting Rome III criteria, five weeks of ondansetron 4 mg with titration against placebo, crossover. Stool consistency improved by 0.9 Bristol points (0.6-1.1, p below 0.001); urgency, frequency and bloating improved; symptom severity fell 83 against 37 points; 65 percent had adequate relief on drug but not placebo against 14 percent the reverse, relative risk 4.7 (2.6-8.5). Pain scores did not change significantly.

31. 31
    Gunn D, Garsed K, Lam C, Singh G, Lingaya M, Wahl V, Niesler B, Henry A, Hall IP, Whorwell P, Spiller R. Abnormalities of mucosal serotonin metabolism and 5-HT3 receptor subunit 3C polymorphism in irritable bowel syndrome with diarrhoea predict responsiveness to ondansetron.Aliment Pharmacol Ther · 2019 · 50(5):538-546doi:10.1111/apt.15420 · PMID 31342534

    Companion analysis of the ondansetron trial. Rectal biopsy 5-HIAA was 2.1 against 1.1 picomoles per milligram protein in controls (p below 0.0001); the 39 patients needing under 4 mg daily had lower mucosal serotonin and greater transit slowing; stool consistency responders were more often CC homozygotes at HTR3C rs6766410, 33 against 14 percent (p=0.0066). Correlational, in a modest sample.

32. 32
    Venkatesan T, Levinthal DJ, Tarbell SE, Jaradeh SS, Hasler WL, Issenman RM, Adams KA, Sarosiek I, Stave CD, Sharaf RN, Sultan S, Li BUK. Guidelines on management of cyclic vomiting syndrome in adults by the American Neurogastroenterology and Motility Society and the Cyclic Vomiting Syndrome Association.Neurogastroenterol Motil · 2019 · 31 Suppl 2:e13604doi:10.1111/nmo.13604 · PMID 31241819

    GRADE-based guideline. Strongly recommends a tricyclic antidepressant such as amitriptyline as first-line prophylaxis for moderate to severe disease, with topiramate or aprepitant as alternatives; conditionally recommends serotonin antagonists such as ondansetron, and triptans or aprepitant, to abort attacks; recommends screening and treating comorbid anxiety, depression, migraine, autonomic dysfunction, sleep disorder and substance use.

33. 33
    Basak S, Gicheru Y, Samanta A, Molugu SK, Huang W, Fuente M, Hughes T, Taylor DJ, Nieman MT, Moiseenkova-Bell V, Chakrapani S. Cryo-EM structure of 5-HT3A receptor in its resting conformation.Nat Commun · 2018 · 9(1):514doi:10.1038/s41467-018-02997-4 · PMID 29410406

    The unliganded full-length receptor at 4.3 angstroms, with a closed pore and an expanded conformation relative to the crystal structure. Homomeric A-subunit receptor.

34. 34
    Polovinkin L, Hassaine G, Perot J, Neumann E, Jensen AA, Lefebvre SN, Corringer PJ, Neyton J, Chipot C, Dehez F, Schoehn G, Nury H. Conformational transitions of the serotonin 5-HT3 receptor.Nature · 2018 · 563(7730):275-279doi:10.1038/s41586-018-0672-3 · PMID 30401839

    Four cryo-electron microscopy structures of the full-length mouse receptor at 3.2 to 4.5 angstroms: with the antiemetic tropisetron, with serotonin, and with serotonin plus a positive allosteric modulator, including an open state. Tropisetron-bound receptor resembles the unliganded state.

35. 35
    Ye ZJ, Wu CY, Zhang H, Zhang HW, Vogel H. The serotonin-gated 5-HT3 receptor: a tale of functions and structures of a prototypical pentameric ligand-gated ion channel.Acta Pharmacol Sin · 2026 · 47(7):1718-1732doi:10.1038/s41401-026-01763-3 · PMID 41813973

    Current review of structure, subunit assembly and pharmacology, and the source for the statement that heteropentameric receptors predominate in vivo while the solved structures are homopentameric, with structures in native membranes named as the main unfinished work. A review, not new data.

36. 36
    Bellono NW, Bayrer JR, Leitch DB, Castro J, Zhang C, O'Donnell TA, Brierley SM, Ingraham HA, Julius D. Enterochromaffin Cells Are Gut Chemosensors that Couple to Sensory Neural Pathways.Cell · 2017 · 170(1):185-198.e16doi:10.1016/j.cell.2017.05.034 · PMID 28648659

    Single-cell measurements in intestinal organoids show enterochromaffin cells express specific chemosensory receptors, are electrically excitable and modulate serotonin-sensitive primary afferent fibers through synaptic connections. Mouse organoids and cultured afferents.

37. 37
    Gralla RJ, Navari RM, Hesketh PJ, Popovic W, Strupp J, Noy J, Einhorn L, Ettinger D, Bushnell W, Friedman C. Single-dose oral granisetron has equivalent antiemetic efficacy to intravenous ondansetron for highly emetogenic cisplatin-based chemotherapy.J Clin Oncol · 1998 · 16(4):1568-73doi:10.1200/JCO.1998.16.4.1568 · PMID 9552067

    1054 patients randomized: total control over 24 hours 54.7 percent with a single 2 mg oral granisetron tablet against 58.3 percent with 32 mg intravenous ondansetron, 95 percent confidence interval on the difference minus 9.6 to 2.4. Establishes equivalence within the first generation and across routes. The ondansetron comparator dose was later withdrawn for QT reasons.

38. 38
    Rojas C, Stathis M, Thomas AG, Massuda EB, Alt J, Zhang J, Rubenstein E, Sebastiani S, Cantoreggi S, Snyder SH, Slusher B. Palonosetron exhibits unique molecular interactions with the 5-HT3 receptor.Anesth Analg · 2008 · 107(2):469-78doi:10.1213/ane.0b013e318172fa74 · PMID 18633025

    Binding studies finding positive cooperativity and allosteric behavior for palonosetron, which accelerated dissociation of granisetron and ondansetron, against simple bimolecular binding for those two. Commercial conflict of interest: the corresponding author and several co-authors were employed by MGI Pharma, which marketed palonosetron. In vitro, in transfected cells.

39. 39
    Rojas C, Thomas AG, Alt J, Stathis M, Zhang J, Rubenstein EB, Sebastiani S, Cantoreggi S, Slusher BS. Palonosetron triggers 5-HT(3) receptor internalization and causes prolonged inhibition of receptor function.Eur J Pharmacol · 2009 · 626(2-3):193-9doi:10.1016/j.ejphar.2009.10.002 · PMID 19836386

    Reports that palonosetron, unlike granisetron and ondansetron, causes receptor internalization visualized by confocal microscopy and inhibition of function persisting after washout. Same commercial conflict of interest as the 2008 paper; the authors state the work may explain published clinical trial differences.

40. 40
    Basak S, Kumar A, Ramsey S, Gibbs E, Kapoor A, Filizola M, Chakrapani S. High-resolution structures of multiple 5-HT3AR-setron complexes reveal a novel mechanism of competitive inhibition.Elife · 2020 · 9:e57870doi:10.7554/eLife.57870 · PMID 33063666

    Structures of the full-length receptor with palonosetron, ondansetron and alosetron, with molecular dynamics in a hydrated lipid environment, showing distinct interaction fingerprints and differing degrees of conformational change along the activation pathway. Homomeric receptor.

41. 41
    Bright B, Vivekanandan A, Ayyasamy L, Haridoss M. Comparative efficacy of palonosetron and ondansetron for chemotherapy-induced nausea and vomiting in pediatric and adult populations: A systematic review and meta-analysis of randomized controlled trials.J Oncol Pharm Pract · 2026 · online ahead of printdoi:10.1177/10781552261492619 · PMID 42808489

    Eleven randomized trials, searched to October 2025, registered on PROSPERO. Complete response relative risk 1.06 (1.02-1.11) acute, 1.30 (1.17-1.43) delayed, 1.29 (1.17-1.42) overall for palonosetron, with comparable adverse events; heterogeneity low, certainty moderate to high. Confirms the clinical difference; says nothing about mechanism.

42. 42
    Kelley SP, Dunlop JI, Kirkness EF, Lambert JJ, Peters JA. A cytoplasmic region determines single-channel conductance in 5-HT3 receptors.Nature · 2003 · 424(6946):321-4doi:10.1038/nature01788 · PMID 12867984

    Chimeric A and B subunits localized the determinant of single-channel conductance to the HA-stretch of the large cytoplasmic loop; replacing three arginine residues unique to the A subunit with their B-subunit counterparts raised conductance 28-fold. The authors state that this solves the conundrum of the anomalously low conductance of homomeric A receptors. Recombinant human receptors in cells.

43. 43
    Rao STRB, Irving HR. Predicted conformations of 5-HT3 receptor ion channels are modified by subunit D.Comput Struct Biotechnol J · 2025 · 27:2394-2402doi:10.1016/j.csbj.2025.05.048 · PMID 40529179

    Homology models of heteromeric receptors containing the D subunit, with and without a variant at residue 110, rs6443930, compared ligand-free and with serotonin or granisetron. Entirely computational; predictions, not measurements.

44. 44
    Celli J, Rappold G, Niesler B. The Human Serotonin Type 3 Receptor Gene (HTR3A-E) Allelic Variant Database.Hum Mutat · 2017 · 38(2):137-147doi:10.1002/humu.23136 · PMID 27763704

    Describes the curated database of variants in all five HTR3 genes with functional relevance, associated phenotypes and pharmacogenetic data. A resource paper collecting published associations; it does not itself test causality.

45. 45
    Tremblay PB, Kaiser R, Sezer O, Rosler N, Schelenz C, Possinger K, Roots I, Brockmoller J. Variations in the 5-hydroxytryptamine type 3B receptor gene as predictors of the efficacy of antiemetic treatment in cancer patients.J Clin Oncol · 2003 · 21(11):2147-55doi:10.1200/JCO.2003.05.164 · PMID 12775740

    242 patients on the first day of chemotherapy, with nausea or vomiting in about 30 percent, had the whole HTR3B gene sequenced: thirteen polymorphisms, two amino acid changes and two deletions. Homozygotes for the promoter deletion vomited more often in both observation periods. The authors state that only a small fraction of therapeutic failure is explained by it and that confirmation is needed.

46. 46
    Binienda A, Salaga M, Patel M, Włodarczyk J, Fichna J, Venkatesan T. Serotonin Receptors Polymorphisms Are Associated With Cyclic Vomiting Syndrome.Neurogastroenterol Motil · 2025 · 37(5):e15012doi:10.1111/nmo.15012 · PMID 39888244

    70 patients with cyclic vomiting syndrome against 2504 controls from the 1000 Genomes Project. Genotypes in HTR1B, HTR1D and the CC genotype of HTR3B rs176744 were associated with increased risk, while alleles in HTR3B rs3788987 and HTR3C rs6766410 were associated with decreased risk. First report implicating these genes; a small clinical series compared against a public reference panel rather than matched controls, so it is hypothesis-generating.

47. 47
    Castejon AM, Paez X, Hernandez L, Cubeddu LX. Use of intravenous microdialysis to monitor changes in serotonin release and metabolism induced by cisplatin in cancer patients: comparative effects of granisetron and ondansetron.J Pharmacol Exp Ther · 1999 · 291(3):960-6PMID 10565811

    In 16 patients given about 89 mg/m2 of cisplatin, blood dialysate 5-HIAA rose 104 percent at four hours while free serotonin did not rise, arguing against circulating serotonin acting on the area postrema and for local release in the intestinal wall. Also found no difference between granisetron and ondansetron in serotonin release or metabolism. No DOI in the PubMed record.

48. 48
    Gralla R, Lichinitser M, Van Der Vegt S, Sleeboom H, Mezger J, Peschel C, Tonini G, Labianca R, Macciocchi A, Aapro M. Palonosetron improves prevention of chemotherapy-induced nausea and vomiting following moderately emetogenic chemotherapy: results of a double-blind randomized phase III trial comparing single doses of palonosetron with ondansetron.Ann Oncol · 2003 · 14(10):1570-7doi:10.1093/annonc/mdg417 · PMID 14504060

    570 randomized, 563 evaluable. Palonosetron 0.25 mg against ondansetron 32 mg: complete response 81.0 against 68.6 percent acute, 74.1 against 55.1 percent delayed, 69.3 against 50.3 percent overall, all p below 0.01. The higher 0.75 mg palonosetron dose was not statistically superior to ondansetron, which the paper reports without explaining.

49. 49
    Hesketh PJ, Grunberg SM, Gralla RJ, Warr DG, Roila F, de Wit R, Chawla SP, Carides AD, Ianus J, Elmer ME, Evans JK, Beck K, Reines S, Horgan KJ. The oral neurokinin-1 antagonist aprepitant for the prevention of chemotherapy-induced nausea and vomiting: a multinational, randomized, double-blind, placebo-controlled trial in patients receiving high-dose cisplatin.J Clin Oncol · 2003 · 21(22):4112-9doi:10.1200/JCO.2003.01.095 · PMID 14559886

    520 patients on cisplatin at least 70 mg/m2. Complete response days 1 to 5 was 72.7 percent with aprepitant added to ondansetron and dexamethasone against 52.3 percent with standard therapy, p below 0.001, with the largest gap on days 2 to 5. Showed that adding a neurokinin-1 antagonist to a 5-HT3 antagonist and dexamethasone improves control, most of all on days 2 to 5. The comparator arm was not 5-HT3 blockade alone.

50. 50
    Camilleri M, Northcutt AR, Kong S, Dukes GE, McSorley D, Mangel AW. Efficacy and safety of alosetron in women with irritable bowel syndrome: a randomised, placebo-controlled trial.Lancet · 2000 · 355(9209):1035-40doi:10.1016/S0140-6736(00)02033-X · PMID 10744088

    647 women with diarrhea-predominant or alternating irritable bowel syndrome, alosetron 1 mg twice daily for 12 weeks. Adequate relief for all three months in 41 percent against 29 percent on placebo, difference 12 percentage points (4.7-19.2), with improved urgency, frequency and consistency. Constipation in 30 against 3 percent, drop-outs 24 against 16 percent. The paper concludes the drug was well tolerated; the post-marketing record disagreed.

51. 51
    Doggrell SA, Hancox JC. Cardiac safety concerns for ondansetron, an antiemetic commonly used for nausea linked to cancer treatment and following anaesthesia.Expert Opin Drug Saf · 2013 · 12(3):421-31doi:10.1517/14740338.2013.780026 · PMID 23578248

    Contemporaneous review of the regulatory action, identifying the 32 mg intravenous dose as the focus of the Food and Drug Administration's concern and noting that a lower dose had also prolonged the QT interval in healthy volunteers, which the authors said needed clarification. A review of published and internet sources.

52. 52
    Danielsson B, Wikner BN, Källén B. Use of ondansetron during pregnancy and congenital malformations in the infant.Reprod Toxicol · 2014 · 50:134-7doi:10.1016/j.reprotox.2014.10.017 · PMID 25450422

    Swedish registers, 1349 infants exposed in early pregnancy 1998-2012, adjusted for year, maternal age, parity, smoking and body mass index. No significant increase in major malformation; cardiovascular defect odds ratio 1.62 and cardiac septum defect risk ratio 2.05 (1.19-3.28). The authors judged an increased septal risk likely; the far larger 2018 cohort did not reproduce it.

53. 53
    Huybrechts KF, Hernández-Díaz S, Straub L, Gray KJ, Zhu Y, Patorno E, Desai RJ, Mogun H, Bateman BT. Association of Maternal First-Trimester Ondansetron Use With Cardiac Malformations and Oral Clefts in Offspring.JAMA · 2018 · 320(23):2429-2437doi:10.1001/jama.2018.18307 · PMID 30561479

    The pivotal study. 1,816,414 Medicaid pregnancies, 88,467 (4.9 percent) exposed in the first trimester, propensity-score stratified. Cardiac malformations 84.4 against 94.4 per 10,000, adjusted relative risk 0.99 (0.93-1.06). Oral clefts 11.1 against 14.0 per 10,000, adjusted relative risk 1.24 (1.03-1.48), risk difference 2.7 per 10,000 (0.2-5.2). Overall malformations 1.01 (0.98-1.05). Dispensing data, Medicaid population, residual confounding by severity possible.

54. 54
    Damkier P, Kaplan YC, Shechtman S, Diav-Citrin O, Cassina M, Weber-Schoendorfer C. Ondansetron in pregnancy revisited: Assessment and pregnancy labelling by the European Medicines Agency (EMA) & Pharmacovigilance Risk Assessment Committee (PRAC).Basic Clin Pharmacol Toxicol · 2021 · 128(4):579-582doi:10.1111/bcpt.13541 · PMID 33275828

    The record of the European regulatory decision and the dissent. The Pharmacovigilance Risk Assessment Committee released an updated assessment in July 2019 and the Summary of Product Characteristics was changed in November 2019 to state that ondansetron should not be used in the first trimester; the European Network of Teratology Information Services argues the step is insufficiently substantiated and not in the interest of pregnant women with severe symptoms. A position paper by one side of the dispute.

55. 55
    Fejzo M, Rocha N, Cimino I, Lockhart SM, Petry CJ, Kay RG, et al. GDF15 linked to maternal risk of nausea and vomiting during pregnancy.Nature · 2024 · 625(7996):760-767doi:10.1038/s41586-023-06921-9 · PMID 38092039

    Shows that most maternal GDF15 comes from the feto-placental unit, that low pre-pregnancy levels raise the risk of hyperemesis gravidarum, that women with beta-thalassemia and chronically high levels report very little nausea and vomiting, and that in mice the response to GDF15 desensitizes with prior exposure. Human genetics, mass spectrometry and mouse physiology; it reframes the cause of the illness ondansetron is used to treat.

56. 56
    Rodriguez Araujo N, Corradi J, Bergé I, Bouzat C. Novel allosteric modulation of the human 5-HT3A receptor by the clinically used drugs piperazine and hydroxyzine.Neuropharmacology · 2026 · 301:111196doi:10.1016/j.neuropharm.2026.111196 · PMID 42790730

    Macroscopic current recordings from human 5-HT3A receptors, with computational docking and mutagenesis at E75, E208, S248 and D293, identify piperazine and hydroxyzine as negative allosteric modulators acting at two sites at the interfaces between subunits. An in vitro drug-repositioning study with no human data; the first allosteric account of this receptor from a laboratory with no commercial interest in a setron.

57. 57
    Hesketh PJ. Chemotherapy-induced nausea and vomiting.N Engl J Med · 2008 · 358(23):2482-94doi:10.1056/NEJMra0706547 · PMID 18525044

    Standard clinical review of the acute and delayed phases, risk factors and combination prophylaxis. The PubMed record carries no abstract. A review, cited here for the framework rather than for primary data.

58. 58
    Committee of Neoplastic Supportive-Care, Chinese Anti-Cancer Association. CACA guidelines for the prevention and control of nausea and vomiting induced by anticancer therapy (2026 edition).Zhonghua Zhong Liu Za Zhi · 2026 · 48(9):1074-1093doi:10.3760/cma.j.cn112152-20260511-00230 · PMID 42633985

    Current multidisciplinary guideline using GRADE and Delphi methods, covering peripheral 5-HT3 and central neurokinin-1 pathways, stepwise regimens for drug therapy, radiotherapy and surgery, and management of refractory, breakthrough and anticipatory symptoms. In Chinese with English abstract; applies to adults.

59. 59
    Weibel S, Rücker G, Eberhart LH, Pace NL, Hartl HM, Jordan OL, Mayer D, Riemer M, Schaefer MS, Raj D, Backhaus I, Helf A, Schlesinger T, Kienbaum P, Kranke P. Drugs for preventing postoperative nausea and vomiting in adults after general anaesthesia: a network meta-analysis.Cochrane Database Syst Rev · 2020 · 10(10):CD012859doi:10.1002/14651858.CD012859.pub2 · PMID 33075160

    585 randomized trials and 97,516 participants, searched to April 2020. High-certainty evidence that ondansetron reduces vomiting within 24 hours, relative risk 0.55 (0.51-0.60), ranking thirteenth of 28 single drugs behind aprepitant (0.26), ramosetron (0.44), granisetron (0.45) and dexamethasone (0.51); combinations generally beat single drugs. Most included trials were small and of unclear risk of bias, and safety evidence was of low certainty.

60. 60
    Gan TJ, Belani KG, Bergese S, Chung F, Diemunsch P, Habib AS, Jin Z, Kovac AL, Meyer TA, Urman RD, Apfel CC, et al. Fourth Consensus Guidelines for the Management of Postoperative Nausea and Vomiting.Anesth Analg · 2020 · 131(2):411-448doi:10.1213/ANE.0000000000004833 · PMID 32467512

    International consensus guideline for adults and children from a panel convened under the American Society for Enhanced Recovery and the Society for Ambulatory Anesthesia, based on literature to September 2019 and endorsed by 23 societies. Covers risk stratification, prophylaxis including second-generation 5-HT3 antagonists and neurokinin-1 antagonists, rescue treatment and multimodal prophylaxis. Consensus built on systematic review, not new data.

61. 61
    Kassar O, Shaheen O, Selim A, Soliman M, Taman L, Elshahat A, Abouelmagd ME. Efficacy and safety of ondansetron in schizophrenia: A systematic review and meta-analysis of randomized controlled trials.Gen Hosp Psychiatry · 2025 · 96:37-46doi:10.1016/j.genhosppsych.2025.06.001 · PMID 40482395

    Eight randomized trials, 533 patients, searched to October 2024. Adjunctive ondansetron improved the negative subscale of the Positive and Negative Syndrome Scale at twelve weeks by 2.96 points (1.24-4.69) and general psychopathology by 2.71 points, with no effect on positive symptoms or depression and significantly more constipation. The authors urge caution given the limitations of the included trials.

62. 62
    Kishi T, Mukai T, Matsuda Y, Iwata N. Selective serotonin 3 receptor antagonist treatment for schizophrenia: meta-analysis and systematic review.Neuromolecular Med · 2014 · 16(1):61-9doi:10.1007/s12017-013-8251-0 · PMID 23896722

    Six trials, 311 patients, of ondansetron, granisetron or tropisetron added to an antipsychotic. Standardized mean difference on total Positive and Negative Syndrome Scale minus 1.03 (minus 1.70 to minus 0.36) with heterogeneity of 82 percent, a figure that means the trials disagreed substantially; no effect on positive symptoms; constipation more common, relative risk 2.05, number needed to harm 11. Small trials, high heterogeneity.

63. 63
    Olivas-Cano I, Rodriguez-Andreu JM, Blasco-Ibañez JM, Crespo C, Nácher J, Varea E. Fluoxetine increased adult neurogenesis is mediated by 5-HT3 receptor.Neurosci Lett · 2022 · 795:137027doi:10.1016/j.neulet.2022.137027 · PMID 36566831

    In adult rats, 5-HT3 receptors were found on stem cells, amplifying progenitors and immature neurons in the dentate gyrus, and co-treatment with ondansetron blocked the increase in proliferation and new neurons produced by chronic fluoxetine. Rat study; there is no human counterpart, and it does not show that ondansetron impairs antidepressant response in people.

64. 64
    Charbit B, Albaladejo P, Funck-Brentano C, Legrand M, Samain E, Marty J. Prolongation of QTc interval after postoperative nausea and vomiting treatment by droperidol or ondansetron.Anesthesiology · 2005 · 102(6):1094-100doi:10.1097/00000542-200506000-00006 · PMID 15915019

    85 patients treated for postoperative nausea: maximal QTc prolongation 17 plus or minus 9 milliseconds with droperidol 0.75 mg at two minutes and 20 plus or minus 13 milliseconds with ondansetron 4 mg at three minutes, both below baseline by 90 minutes. The QTc was already prolonged in 21 percent before any drug. Single-blind, prospective.

65. 65
    Charbit B, Alvarez JC, Dasque E, Abe E, Démolis JL, Funck-Brentano C. Droperidol and ondansetron-induced QT interval prolongation: a clinical drug interaction study.Anesthesiology · 2008 · 109(2):206-12doi:10.1097/ALN.0b013e31817fd8c8 · PMID 18648229

    Double-blind randomized crossover in 16 healthy volunteers: maximum placebo-corrected, baseline-subtracted QTcF prolongation 25 plus or minus 8 milliseconds for droperidol 1 mg, 17 plus or minus 10 for ondansetron 4 mg, 28 plus or minus 10 for both, with no pharmacokinetic interaction. Healthy volunteers, single doses.

66. 66
    Brenner SM, Boucher J. Fatal Cardiac Arrest in 2 Children: Possible Role of Ondansetron.Pediatr Emerg Care · 2016 · 32(11):779-784doi:10.1097/PEC.0000000000000317 · PMID 27814325

    Two deaths after ondansetron given for gastroenteritis: a previously healthy 10-year-old who developed wide-complex tachycardia, and an 86-day-old infant with unrecognized congenital cardiomyopathy. Case reports; the authors note that other proarrhythmic factors were present and that experience in children under two is very limited.

67. 67
    Gandhi A, Parhizgar A, Bhise V. QT-related adverse events with ondansetron and olanzapine: a real-world FAERS analysis with implications for oncology anti-emetic practice.Front Pharmacol · 2026 · 17:1748635doi:10.3389/fphar.2026.1748635 · PMID 41800097

    2608 ondansetron and 15,257 olanzapine primary-suspect reports from January 2015 to October 2025. Narrow-term QT events in 4.91 against 1.61 percent of reports; reporting odds ratio 27.24 (22.78-32.58) against 8.70 (7.65-9.88). The authors state explicitly that these results do not establish causality; spontaneous reporting cannot give incidence.

68. 68
    Parker SE, Van Bennekom C, Anderka M, Mitchell AA. Ondansetron for Treatment of Nausea and Vomiting of Pregnancy and the Risk of Specific Birth Defects.Obstet Gynecol · 2018 · 132(2):385-394doi:10.1097/AOG.0000000000002679 · PMID 29995744

    Two case-control studies, the National Birth Defects Prevention Study (1997-2011) and the Slone Birth Defects Study (1997-2014), using women with untreated first-trimester nausea as the reference. Use rose from under 1 percent before 2000 to 13 percent in 2013-2014. No increase for most of 51 defect groups; cleft palate adjusted odds ratio 1.6 (1.1-2.3) and renal agenesis-dysgenesis 1.8 (1.1-3.0), which the authors say may be chance. Interview-based exposure recall.

69. 69
    Abas MN, Tan PC, Azmi N, Omar SZ. Ondansetron compared with metoclopramide for hyperemesis gravidarum: a randomized controlled trial.Obstet Gynecol · 2014 · 123(6):1272-1279doi:10.1097/AOG.0000000000000242 · PMID 24807340

    160 women admitted with hyperemesis gravidarum, double-blind, intravenous ondansetron 4 mg against metoclopramide 10 mg eight-hourly for 24 hours. Similar well-being, nausea and vomiting, on a trial not designed as an equivalence study; less drowsiness (12.5 against 30 percent), dry mouth (10.0 against 23.8 percent) and persistent ketonuria (12.5 against 30 percent) with ondansetron. An active-comparator trial; neither arm was placebo.

70. 70
    Oliveira LG, Capp SM, You WB, Riffenburgh RH, Carstairs SD. Ondansetron compared with doxylamine and pyridoxine for treatment of nausea in pregnancy: a randomized controlled trial.Obstet Gynecol · 2014 · 124(4):735-742doi:10.1097/AOG.0000000000000479 · PMID 25198265

    36 women randomized, 30 completing follow-up. Median nausea fell 51 mm (interquartile range 37-64) on ondansetron against 20 mm (8-51) on doxylamine with pyridoxine over five days (p=0.019), with less vomiting (p=0.049) and no difference in sedation or constipation. Very small, and only five days long.

71. 71
    Pasternak B, Svanström H, Hviid A. Ondansetron in pregnancy and risk of adverse fetal outcomes.N Engl J Med · 2013 · 368(9):814-23doi:10.1056/NEJMoa1211035 · PMID 23445092

    Danish cohort of 608,385 pregnancies with 1:4 propensity-score matching: no significant increase in spontaneous abortion, stillbirth, major birth defect (2.9 against 2.9 percent, odds ratio 1.12, 0.69-1.82), preterm delivery, low birth weight or small size for gestational age. The exposed birth-defect group numbered 1233, too few for rare defects such as oral cleft.

72. 72
    Huybrechts KF, Hernandez-Diaz S, Straub L, Gray KJ, Zhu Y, Mogun H, Bateman BT. Intravenous Ondansetron in Pregnancy and Risk of Congenital Malformations.JAMA · 2020 · 323(4):372-374doi:10.1001/jama.2019.18587 · PMID 31730152

    A research letter from the same group using the same Medicaid data to examine intravenous rather than oral exposure in the first trimester. Short-format publication; its estimates appear only in the letter itself, which is why no figures from it are quoted on this page.

73. 73
    Zambelli-Weiner A, Via C, Yuen M, Weiner DJ, Kirby RS. First trimester ondansetron exposure and risk of structural birth defects.Reprod Toxicol · 2018 · 83:14-20doi:10.1016/j.reprotox.2018.10.010 · PMID 30385129

    864,083 commercially insured mother-infant pairs 2000-2014: cardiac defects odds ratio 1.52 (1.35-1.70) and orofacial clefts 1.32 (0.76-2.28, not significant), compared against women with no antiemetic exposure in pregnancy. That comparison group leaves confounding by indication unaddressed, since women needing an antiemetic differ systematically from women who need none. The first four authors work at TTi Health Research & Economics, a private contract research firm, with one academic co-author; the PubMed record carries no funding or conflict-of-interest statement, and the commercial setting of the work is something a reader should weigh. PubMed lists no retraction or expression of concern.

74. 74
    Lemon LS, Bodnar LM, Garrard W, Venkataramanan R, Platt RW, Marroquin OC, Caritis SN. Ondansetron use in the first trimester of pregnancy and the risk of neonatal ventricular septal defect.Int J Epidemiol · 2020 · 49(2):648-656doi:10.1093/ije/dyz255 · PMID 31860078

    33,677 deliveries at one Pittsburgh hospital, 3733 (11 percent) exposed in the first trimester at cumulative doses from 2.4 to 1008 mg. Ventricular septal defect risk ratios 1.7 (1.0-2.9) to 2.1 (1.1-4.0) across methods, corresponding to about one additional defect per 330 exposed pregnancies; highest against lowest cumulative dose adjusted risk ratio 3.2 (1.0-9.9). Single center; the dose-response interval only just excludes no effect.

75. 75
    Kaplan YC, Richardson JL, Keskin-Arslan E, Erol-Coskun H, Kennedy D. Use of ondansetron during pregnancy and the risk of major congenital malformations: A systematic review and meta-analysis.Reprod Toxicol · 2019 · 86:1-13doi:10.1016/j.reprotox.2019.03.001 · PMID 30849498

    No significant increase in major malformations, heart defects, orofacial clefts, genitourinary malformations or hypospadias in the primary analysis, with significant heterogeneity for isolated cleft palate and elevated point estimates in some secondary analyses. The authors call the absolute risks low and the result reassuring, while asking for continued surveillance.

76. 76
    Picot C, Berard A, Grenet G, Ripoche E, Cucherat M, Cottin J. Risk of malformation after ondansetron in pregnancy: An updated systematic review and meta-analysis.Birth Defects Res · 2020 · 112(13):996-1013doi:10.1002/bdr2.1705 · PMID 32420702

    Twelve studies searched to November 2019. Ventricular septal defects odds ratio 1.11 (1.00-1.23); oral clefts 1.22 (1.00-1.49); cleft palate 1.48 (1.19-1.84) after excluding the study responsible for heterogeneity. No significant association for major congenital malformations overall, total cardiac malformations, atrial septal defects or cleft lip. Pools studies with different comparison groups.

77. 77
    Dormuth CR, Winquist B, Fisher A, Wu F, Reynier P, Suissa S, Dahl M, Ma Z, Lu X, Zhang J, Raymond CB, Filion KB, Platt RW, Moriello C, Paterson JM. Comparison of Pregnancy Outcomes of Patients Treated With Ondansetron vs Alternative Antiemetic Medications in a Multinational, Population-Based Cohort.JAMA Netw Open · 2021 · 4(4):e215329doi:10.1001/jamanetworkopen.2021.5329 · PMID 33890993

    456,963 pregnancies across five Canadian provinces, United States claims and United Kingdom practice records, April 2002 to March 2016, with meta-analysis. Compared against other antiemetics rather than no treatment: fetal death adjusted hazard ratio 0.91 (0.67-1.23), spontaneous abortion 0.82 (0.64-1.04), stillbirth 0.97 (0.79-1.20), major congenital malformations odds ratio 1.06 (0.91-1.22). The active-comparator design removes confounding by indication but reports malformations only as a composite.

78. 78
    Cao X, Sun M, Yang Q, Wang Q, Hou L, Wang J, Wu Y, Ge L. Risk of abnormal pregnancy outcomes after using ondansetron during pregnancy: A systematic review and meta-analysis.Front Pharmacol · 2022 · 13:951072doi:10.3389/fphar.2022.951072 · PMID 36120333

    Twenty observational studies searched to December 2021. Preliminary analysis found cardiac defects 1.06 (1.01-1.10) and neural tube defects 1.12 (1.05-1.18); sensitivity analysis abolished both (1.15, 0.94-1.40 and 0.87, 0.46-1.66). No association with orofacial clefts (1.09, 0.95-1.25), any congenital malformation (1.03, 0.98-1.09), stillbirth or preterm birth; reduced miscarriage (0.53, 0.31-0.89). Illustrates how much the conclusions depend on which studies are included.

79. 79
    Andrade C. Major Congenital Malformation Risk After First Trimester Gestational Exposure to Oral or Intravenous Ondansetron.J Clin Psychiatry · 2020 · 81(3):20f13472doi:10.4088/JCP.20f13472 · PMID 32526103

    A clinical-reasoning review that accepts a signal exists for heart and orofacial defects while quantifying it in absolute terms: an increase of about 0.03 percent for orofacial clefts and 0.3 percent for ventricular septal defect, to be weighed against the risks of untreated nausea and vomiting, with decisions shared between clinician and patient. A review, not new data.

80. 80
    Morgan S, Quintrell E, Russell DJ, Rhaman A, Gebremedhin A, Wyrwoll CS, Kelty E. A Scoping Review of 5-HT3 Antagonist Antiemetic Medication Safety in the Management of Nausea and Vomiting of Pregnancy.Pharmacoepidemiol Drug Saf · 2026 · 35(9):e70449doi:10.1002/pds.70449 · PMID 42592880

    Four databases searched from inception to 20 March 2024 and updated 7 May 2025: 1942 records screened and 39 original studies included, 34 on ondansetron and 5 on granisetron. No elevated risk of congenital anomalies overall; cardiac and orofacial results conflicting, with minimal absolute changes in risk either way. A scoping review, so it maps the literature and produces no pooled estimate.

81. 81
    Danielsson B, Webster WS, Ritchie HE. Ondansetron and teratogenicity in rats: Evidence for a mechanism mediated via embryonic hERG blockade.Reprod Toxicol · 2018 · 81:237-245doi:10.1016/j.reprotox.2018.08.018 · PMID 30149139

    Gestational day 13 rat embryos in culture: ondansetron caused concentration-dependent bradycardia and arrhythmia, and cardiovascular malformations occurred in rats at exposures slightly above those of early human pregnancy. The authors propose embryonic hERG blockade as the mechanism and suggest risk could rise with cofactors such as low potassium. Rat embryo culture; plausibility, not human causation, and the human cardiac association it was offered to explain did not survive the 2018 Medicaid cohort.

82. 82
    Petersen JM, Kahrs JC, Adrien N, Wood ME, Olshan AF, Smith LH, Howley MM, Ailes EC, Romitti PA, Herring AH, Parker SE, Shaw GM, Politis MD. Bias analyses to investigate the impact of differential participation: Application to a birth defects case-control study.Paediatr Perinat Epidemiol · 2024 · 38(6):535-543doi:10.1111/ppe.13026 · PMID 38102868

    Deterministic quantitative bias analysis of three associations in the National Birth Defects Prevention Study. For ondansetron and cleft palate, exposed cases would need to have been 1.6 times more likely to participate than exposed controls to produce the observed estimate if the true odds ratio were null. It shows the association could be explained by participation bias; it does not show that it is.

83. 83
    Tolchinsky I, Beckinshtein M, Kohn E, Cohen R, De-Haan T, Ziv-Baran T, Stepensky D, Gueta I, Berkovitch M, Berlin M. Another prospective study on the safety of ondansetron for nausea and vomiting of pregnancy: addressing ongoing concerns.Front Pharmacol · 2026 · 17:1783632doi:10.3389/fphar.2026.1783632 · PMID 41883501

    Prospective cohort from an Israeli teratology information service, 260 women, 137 exposed and 123 controls, recruited 2014-2018. Major congenital malformations 5.1 against 4.1 percent (p=0.695), with no consistent pattern, and no difference in live birth rate, birth weight or neonatal complications. The exposed group had more severe vomiting at baseline; the authors state the sample cannot detect very rare outcomes.

84. 84
    Cho Y, Choi EY, Han JY, Kim H, Choe YJ, Shin JY. Antiemetic use during pregnancy and child neurodevelopment: population-based birth cohort study.BMC Med · 2025 · 23(1):667doi:10.1186/s12916-025-04497-x · PMID 41327197

    South Korean National Health Insurance mother-child cohort, 2009-2023, 630,904 children, of whom 5673 (0.9 percent) were exposed to ondansetron. Overlap-weighted Cox models found no substantial association between any of six antiemetics and seven neurodevelopmental disorders or screening-detected delay; the only signals were with metoclopramide. Claims data, so exposure is dispensing rather than ingestion.

85. 85
    Yang J, Hong M, Liang F, Han Y, Xie R, Krewski D, Mattison D, Corsi DJ, Walker M, El-Chaâr D, McNair D, Xu J, Wen SW. Ondansetron Use for Women With Hyperemesis Gravidarum and Risks of Spontaneous Abortion and Stillbirth: A Retrospective Cohort Study.Pharmacoepidemiol Drug Saf · 2026 · 35(2):e70335doi:10.1002/pds.70335 · PMID 41614660

    Records from over 500 United States facilities, 2001-2016, restricted to women with hyperemesis gravidarum: after propensity matching, 10,238 pairs for spontaneous abortion and 13,585 for stillbirth. Spontaneous abortion adjusted hazard ratio 0.82 (0.73-0.91) against no antiemetic and 0.85 (0.73-0.98) against other antiemetics; no association with stillbirth. Restricting to hyperemesis reduces, but cannot remove, confounding by severity.

86. 86
    Fejzo MS, MacGibbon KW, Mullin PM. Ondansetron in pregnancy and risk of adverse fetal outcomes in the United States.Reprod Toxicol · 2016 · 62:87-91doi:10.1016/j.reprotox.2016.04.027 · PMID 27151373

    Retrospective cohort of 1070 pregnancies exposed to ondansetron for hyperemesis gravidarum against 771 with a hyperemesis history but no exposure and 1555 with neither. Ventricular septal defects in 2 of 952 exposed live births and 4 of 1286 unexposed; cleft palate in 1 of 952 and 2 of 1286; fewer miscarriages and terminations and more live births among exposed women. Self-selected registry participants; the counts are too small to exclude a modest effect.

87. 87
    Chen T, Chen C, Zhou H, Zhang J. Signal mining of adverse reactions in the antiemetic drug ondansetron during pregnancy: A real-world analysis of the FDA adverse event reporting system (FAERS).Expert Opin Drug Saf · 2024 · published online 5 Aug 2024:1-9doi:10.1080/14740338.2024.2386684 · PMID 39099462

    1064 reports with ondansetron as primary suspect among 15,727 pregnancy reports, 2017 to early 2023. The strongest disproportionality signals were in congenital, familial and genetic disorders, ear and labyrinth disorders and cardiac disorders. The authors state that causal relationship and risk level cannot be inferred from these data; spontaneous reporting cannot give incidence and is shaped by publicity and litigation.

88. 88
    Ambrogi M, Racine JL, Vezina CM. Effects of Selective Serotonin Reuptake Inhibitors and Ondansetron on Urinary Tract Infections in Pregnancy.Reprod Sci · 2025 · 32(4):1155-1165doi:10.1007/s43032-025-01826-6 · PMID 40044992

    Retrospective analysis of 7023 pregnant women, of whom 1019 received ondansetron and 316 a selective serotonin reuptake inhibitor, found increased odds of urinary tract infection and preterm birth with ondansetron but not with the antidepressants; pregnant mice given ondansetron before Escherichia coli challenge had more severe bladder infection. A human association joined to an animal mechanism; it needs replication before it changes practice.

89. 89
    Moore C, Bourque MS, Halman A, Agundez JAG, Prows CA, Hikino K, Schwab M, Oxencis CJ, Chauhan D, Diekstra MHM, Long SE, Bell GC, Gaedigk A, Whirl-Carrillo M, Klein TE, Caudle KE, Conyers R. Clinical Pharmacogenetics Implementation Consortium (CPIC) Guideline for CYP2D6 Genotype and Use of 5-HT3 Receptor Antagonists: 2026 Update.Clin Pharmacol Ther · 2026 · 120(2):387-393doi:10.1002/cpt.70291 · PMID 41979467

    Update of the 2016 CPIC guideline. States that most 5-HT3 receptor antagonists are metabolized by CYP2D6 but that the extent of that involvement varies between drugs, that CYP2D6 variation influences metabolism and thereby efficacy particularly for ondansetron and tropisetron, and provides genotype-based therapeutic recommendations where variation is associated with reduced efficacy. A guideline, not new data.

90. 90
    Liu M, Shuey MM, Holley SE, Fountain DF, Jaworski J, Griffin C, Kim J, Stec M, Sunthankar SD, Ong HH, Gangireddy S, Wei WQ, Gaedigk A, Velez Edwards DR, Kawai VK. Impact of CYP2D6 Metabolizer Status on Ondansetron Efficacy in Early Pregnancy Induced Nausea and Vomiting: A Case Control Study.Clin Transl Sci · 2026 · 19(3):e70523doi:10.1111/cts.70523 · PMID 41820792

    264 pregnant patients genotyped for 11 CYP2D6 variants and copy number. Metabolizer status was not significantly associated with response; ultrarapid and normal metabolizers had an odds ratio of 1.53 (0.88-2.66) for non-response against poor and intermediate metabolizers. Retrospective; half the cohort took other antiemetics simultaneously.

91. 91
    Chang L, Tong K, Ameen V. Ischemic colitis and complications of constipation associated with the use of alosetron under a risk management plan: clinical characteristics, outcomes, and incidences.Am J Gastroenterol · 2010 · 105(4):866-75doi:10.1038/ajg.2010.25 · PMID 20197759

    Safety reports from November 2002 to June 2008: 29,072 patients, 203,939 prescriptions; ischemic colitis 0.95 and serious constipation complications 0.36 per 1000 patient-years, similar to rates before withdrawal. In trials since reintroduction, ischemic colitis 4 per 1000 on alosetron (n=998) against 2 per 1000 on placebo (n=411). No surgery, transfusion or death among cases. Commercial conflict of interest: two authors were affiliated with the sponsor.

92. 92
    Andrews MB, Adler DG. Diarrhea-predominant irritable bowel syndrome treatment options eluxadoline, rifaximin, and alosetron: analysis of the FDA adverse event reporting system (FAERS) database.Expert Opin Drug Saf · 2026 · 25(9):1785-1791 (online April 2025)doi:10.1080/14740338.2025.2490841 · PMID 40197054

    Reports from approval to June 2024. Alosetron accounted for 3832 adverse events: constipation 2007, colitis 235, ischemic colitis 140, obstruction 110, perforation 26. The percentages printed beside those counts, 23.1, 2.7, 1.6, 1.3 and 0.3 percent, are calculated on a larger denominator than the 3832 events. Counts of spontaneous reports, not incidence rates, and reporting is intensified by the drug's risk management program.

93. 93
    Bonnet U. [The cannabinoid hyperemesis syndrome - A narrative review].Nervenarzt · 2026 · 97(4):377-381 (online July 2025)doi:10.1007/s00115-025-01864-0 · PMID 40689972

    German-language narrative review with English abstract. States that the Rome IV criteria define cannabis-induced cyclic vomiting as cannabinoid hyperemesis syndrome and treat it as a special variant of cyclic vomiting syndrome. States that commonly used antiemetics often do not help, that hot showers and intramuscular haloperidol 5 mg give acute relief and capsaicin cream 0.075 to 0.1 percent works more slowly, that cure requires sustained cannabis abstinence, and that distinguishing the syndrome from cyclic vomiting, in which a third of patients also use cannabis, requires six to twelve months of abstinence. A selective review, not systematic.

94. 94
    Törnblom H, Carbone F, Hasler WL, Smout A, Suzuki H, Tack J, Talley NJ, Stanghellini V. Gastroduodenal Disorders.Gastroenterology · 2026 · 170(6):1240-1260doi:10.1053/j.gastro.2026.01.038 · PMID 41713705

    The Rome V chapter defining the gastroduodenal categories, including the nausea and vomiting disorders: chronic nausea vomiting syndrome, cyclic vomiting syndrome and cannabinoid hyperemesis syndrome as a separate category. Consensus criteria, not data.

95. 95
    Remes-Troche JM. Rome V: minimizing "functional", maximizing precision in gut-brain disorders.Rev Esp Enferm Dig · 2026 · 118(8):438-441doi:10.17235/reed.2026.12073/2026 · PMID 42439384

    Narrative account of what the fifth edition of the Rome criteria changed: the move away from the word functional toward disorders of gut-brain interaction, and clearer separation of chronic nausea and vomiting syndrome, cyclic vomiting syndrome and cannabinoid hyperemesis syndrome. A commentary, not primary data.

96. 96
    Ilgen MA, Price AM, Goldman P, Whittington BJ, Hicks BM. Prevalence and Correlates of Symptoms of Cannabinoid Hyperemesis Syndrome in the United States, 2025.Am J Public Health · 2026 · published online ahead of printdoi:10.2105/AJPH.2026.308753 · PMID 42821817

    Nationally representative survey of 7034 United States adults, May to September 2025: 2.7 percent of all respondents and 17.8 percent of daily cannabis users reported symptoms consistent with the syndrome, more often among younger, female, non-White and lower-income respondents and those with cannabis-related problems. Self-reported symptoms, not clinical diagnoses.

97. 97
    Plebon-Huff S, Goodman S, Abramovici H. Trends in hospitalizations for cannabinoid hyperemesis syndrome in Canada, 2016-17 to 2024-25.Front Public Health · 2026 · 14:1740300doi:10.3389/fpubh.2026.1740300 · PMID 41908777

    7533 hospital records from the Canadian Institute for Health Information. Rates rose from 1.5 per 100,000 population in 2016-17 to 3.8 per 100,000 in 2024-25, with most of the rise before 2020-21 and stabilization after; mean age 32.4 years. Administrative coding; rising recognition cannot be separated from rising incidence.

98. 98
    Hickey JL, Witsil JC, Mycyk MB. Haloperidol for treatment of cannabinoid hyperemesis syndrome.Am J Emerg Med · 2013 · 31(6):1003.e5-6doi:10.1016/j.ajem.2013.02.021 · PMID 23583118

    Single case of cannabinoid hyperemesis improving substantially with haloperidol after conventional antiemetics failed. A case report; it generated the hypothesis the HaVOC trial later tested.

99. 99
    Witsil JC, Mycyk MB. Haloperidol, a Novel Treatment for Cannabinoid Hyperemesis Syndrome.Am J Ther · 2017 · 24(1):e64-e67doi:10.1097/MJT.0000000000000157 · PMID 25393073

    Four patients whose cannabinoid hyperemesis failed standard emergency department therapy and improved significantly with haloperidol. The authors invoke interactions between dopamine and cannabinoid type 1 signaling in animals as a possible mechanism. Uncontrolled case series.

100. 100
     Lee C, Greene SL, Wong A. The utility of droperidol in the treatment of cannabinoid hyperemesis syndrome.Clin Toxicol (Phila) · 2019 · 57(9):773-777doi:10.1080/15563650.2018.1564324 · PMID 30729854

     Eleven-year (2006-2016) single-center review: 76 of 689 screened presentations met criteria, 37 treated with droperidol and 39 not. Median length of stay 6.7 against 13.9 hours (p=0.014); ondansetron and metoclopramide were used about twice as often in the untreated group; the commonest droperidol dose was 0.625 mg intravenously. Retrospective and unrandomized.

101. 101
     Kelly GS, Meeks G, McCoul B, Doshi VK, Moran TP. Droperidol is associated with reduced length of stay in the treatment of cannabinoid hyperemesis syndrome.Clin Toxicol (Phila) · 2025 · 63(8):550-555doi:10.1080/15563650.2025.2516128 · PMID 40511503

     211 encounters among 158 patients at two Atlanta academic emergency departments, March to August 2024. Droperidol used in 77 encounters at a median 1.25 mg: length of stay 409 against 641 minutes, adjusted mean ratio 0.76 (0.62-0.94, p=0.01), with less total medication and markedly less opioid use and two mild adverse reactions. Retrospective; the authors call for a randomized trial of dosing.

102. 102
     Chopra Q, Peyko V, Lee JA, Puhalla L, Gemmel DJ, Bolotin T. Droperidol Plus Diphenhydramine for Symptom Improvement in Suspected Cannabinoid Hyperemesis Syndrome: A Prospective Cohort Study.Open Access Emerg Med · 2024 · 16:267-273doi:10.2147/OAEM.S473627 · PMID 39619486

     Multicenter prospective interventional cohort, 47 patients given droperidol with diphenhydramine. Nausea and vomiting scores fell from 8.3 at baseline to 3.1 at 30 minutes and 1.4 at 120 minutes; abdominal pain from 7.8 to 3.6 to 1.7; 12.9 percent returned within seven days. Single-arm, with no comparison group, so regression to the mean cannot be excluded.

103. 103
     Pourmand A, Esmailian G, Mazer-Amirshahi M, Lee-Park O, Tran QK. Topical capsaicin for the treatment of cannabinoid hyperemesis syndrome, a systematic review and meta-analysis.Am J Emerg Med · 2021 · 43:35-40doi:10.1016/j.ajem.2021.01.004 · PMID 33493995

     Seven studies, 106 patients, searched to October 2020. Mean time to symptom resolution after application 325 minutes (234-787) and emergency department length of stay 379 minutes (10-747), from two studies reporting them. Wide intervals, small and mostly uncontrolled studies; the authors call for randomized trials.

104. 104
     Albert K, Sivilotti MLA, Gareri J, Day A, Ruberto AJ, Hookey LC. Hair cannabinoid concentrations in emergency patients with cannabis hyperemesis syndrome.CJEM · 2019 · 21(4):477-481doi:10.1017/cem.2018.479 · PMID 30806339

     16 cases, 16 recreational-user controls and 14 emergency department controls with positive urine tetrahydrocannabinol. Scalp hair cannabinoid concentrations were similar across groups (cases 220 pg/mg, recreational users 150, emergency controls 270); only the tetrahydrocannabinol to cannabinol ratio differed, 2.6-fold lower in cases (1.3-5.7). Suggests heavy use is necessary but not sufficient. Small, and hair measures months of exposure, not the attack.

105. 105
     Toomey D, Merchant-Borna K, Krotulski AJ, Baker S, Zhang V, Nacca N. Serum cannabinoid profiles of emergency department patients presenting with cannabinoid hyperemesis syndrome and acute cannabis intoxication.Clin Toxicol (Phila) · 2026 · 64(10):881-887doi:10.1080/15563650.2026.2703873 · PMID 42504769

     Retrospective cohort at one academic center: 35 patients with cannabinoid hyperemesis and 15 with acute intoxication, remnant blood analyzed by liquid chromatography and tandem mass spectrometry. No significant differences in tetrahydrocannabinol, its metabolites, delta-8-tetrahydrocannabinol or non-psychotropic phytocannabinoids, or their ratios. Argues against differential cannabinoid accumulation as the cause; host factors were uncontrolled.

106. 106
     Chiew AL, Buckley NA. The serotonin toxidrome: shortfalls of current diagnostic criteria for related syndromes.Clin Toxicol (Phila) · 2022 · 60(2):143-158doi:10.1080/15563650.2021.1993242 · PMID 34806513

     Narrative review with reanalysis of 55 published fatal cases: 12 (22 percent) judged unlikely to be serotonin toxicity; Sternbach and Radomski criteria met by 25 (45 percent) with 20 having insufficient data; only 13 documented as meeting Hunter criteria. The authors argue that diagnosing from symptom criteria without asking whether the drug can raise central serotonin has produced case reports and government warnings for drugs that cannot plausibly cause it, naming ondansetron and the antipsychotics.

107. 107
     Gollapudy S, Kumar V, Dhamee MS. A case of serotonin syndrome precipitated by fentanyl and ondansetron in a patient receiving paroxetine, duloxetine, and bupropion.J Clin Anesth · 2012 · 24(3):251-2doi:10.1016/j.jclinane.2011.04.017 · PMID 22537574

     A single perioperative case in a patient already taking three serotonergic antidepressants. The PubMed record carries no abstract. Representative of the case-report literature on which the warning rests, and of its central weakness: several drugs that do raise synaptic serotonin were present.

108. 108
     Guo MH, Monir RL, Wright A, Holland NP. Case of Serotonin Syndrome Initially Presenting as Diffuse Body Pain.Am J Case Rep · 2018 · 19:1227-1231doi:10.12659/AJCR.911204 · PMID 30318504

     A woman on peritoneal dialysis taking duloxetine and tramadol developed nausea, vomiting, hyperreflexia and inducible clonus; her ondansetron, given to treat the nausea, was listed among the serotonergic agents and withdrawn with the rest, and the authors suggest it exacerbated the syndrome. Illustrates the circularity in this literature: nausea is a feature of the syndrome, so an antiemetic will usually have been given before the diagnosis is made.

109. 109
     Frazier R, Li BUK, Venkatesan T. Diagnosis and Management of Cyclic Vomiting Syndrome: A Critical Review.Am J Gastroenterol · 2023 · 118(7):1157-1167doi:10.14309/ajg.0000000000002216 · PMID 36791365

     Review stating a United States prevalence of about 2 percent, close association with migraine, unknown pathophysiology with genetic, environmental, autonomic and neurohormonal contributions, common therapeutic cannabis use among patients, and the distinction from cannabinoid hyperemesis. Abortive therapy includes triptans, ondansetron and sedation; prophylaxis is amitriptyline, topiramate and aprepitant. A review.

110. 110
     Sharaf RN, Venkatesan T, Shah R, Levinthal DJ, Tarbell SE, Jaradeh SS, Hasler WL, Issenman RM, Adams KA, Sarosiek I, Stave CD, Li BUK, Sultan S. Management of cyclic vomiting syndrome in adults: Evidence review.Neurogastroenterol Motil · 2019 · 31 Suppl 2:e13605doi:10.1111/nmo.13605 · PMID 31241818

     The systematic evidence review behind the adult guideline. Grades the evidence for amitriptyline, topiramate, aprepitant, zonisamide, levetiracetam and mitochondrial supplements as very low quality, with moderate certainty only for triptans as abortive therapy, and describes the evidence supporting ondansetron as limited.

111. 111
     Karrento K, Rosen JM, Tarbell SE, Issenman RM, Gelfand AA, Gamboa H, Parikh S, Adams K, Wiercioch W, Li BUK. North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition 2025 guidelines for management of cyclic vomiting syndrome in children.J Pediatr Gastroenterol Nutr · 2025 · 80(6):1028-1061doi:10.1002/jpn3.70020 · PMID 40223700

     Sixteen GRADE-based recommendations. The only strong recommendation for aborting attacks is anti-migraine agents in children with personal or family migraine history; oral and intravenous 5-HT3 and neurokinin-1 antagonists are conditional, as are beta-blockers, 5-HT2A antagonists and tricyclics for prophylaxis, with anticonvulsants reserved for refractory cases. Uses indirect evidence from pediatric migraine where direct evidence is lacking.

112. 112
     Karrento K, Rosen JM, Gelfand AA, Parikh S, Tarbell SE, Issenman RM, Gamboa H, Adams K, Wiercioch W, Li BUK. North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition 2025 guidelines on the diagnosis of cyclic vomiting syndrome in children.J Pediatr Gastroenterol Nutr · 2025 · 81(5):1346-1359doi:10.1002/jpn3.70193 · PMID 40836301

     Companion diagnostic guideline: screening blood and urine tests and an upper gastrointestinal radiographic series for all children with suggestive symptoms, other screening tests low-yield, updated symptom-based criteria based on attack frequency, duration and associated symptoms, and a separate consensus approach to pediatric cannabinoid hyperemesis as a related but distinct entity. No diagnostic biomarker exists.

113. 113
     Sharma N, MacGibbon KW, Brecht-Doscher A, Cortessis VK, Fejzo MS. Prepregnancy metformin use associated with lower risk of severe nausea and vomiting of pregnancy and hyperemesis gravidarum.Am J Obstet Gynecol · 2025 · 233(6):649.e1-649.e14doi:10.1016/j.ajog.2025.06.055 · PMID 40588059

     5414 respondents to a structured questionnaire posted on the social media sites of a hyperemesis foundation. Daily metformin in the month before the first pregnancy: adjusted relative risk 0.29 (0.12-0.71) for hyperemesis gravidarum; selective serotonin reuptake inhibitors before the first pregnancy (2.41, 1.33-4.38) and cannabis before a second (adjusted odds ratio 3.48, 1.80-6.75) were associated with higher risk. Self-selected and recall-based; the authors call for trials.

114. 114
     Svenningsen JS, Ostenfeld A, Petersen JF, Svenningsen MS, Hillig T, Gillum MP, Grevengoed TJ, Løkkegaard ECL. Early pregnancy GDF15 trajectories and their association with normal gestation, pregnancy loss, and hyperemesis gravidarum.EBioMedicine · 2026 · 131:106428doi:10.1016/j.ebiom.2026.106428 · PMID 42561814

     Longitudinal serum GDF15 and Pregnancy-Unique Quantification of Emesis scores from two Danish early-pregnancy cohorts, one of them the VOMIT trial. GDF15 rises steeply to week 7 and plateaus by week 9; levels are about 19 percent higher with a female fetus; hyperemesis levels are higher early but their confidence intervals overlap those of normal pregnancy by week 8; early pregnancy loss shows lower levels and a slower rise. Observational; defines sampling windows rather than proving clinical usefulness.